{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9997469849034326,
  "eval_steps": 500,
  "global_step": 8892,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5e-05,
      "loss": 0.7047,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-05,
      "loss": 0.6771,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-05,
      "loss": 0.676,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-05,
      "loss": 0.6731,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 5e-05,
      "loss": 0.6757,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 5e-05,
      "loss": 0.6726,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 5e-05,
      "loss": 0.6734,
      "step": 70
    },
    {
      "epoch": 0.03,
      "learning_rate": 5e-05,
      "loss": 0.6718,
      "step": 80
    },
    {
      "epoch": 0.03,
      "learning_rate": 5e-05,
      "loss": 0.6678,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 5e-05,
      "loss": 0.6683,
      "step": 100
    },
    {
      "epoch": 0.04,
      "learning_rate": 5e-05,
      "loss": 0.6735,
      "step": 110
    },
    {
      "epoch": 0.04,
      "learning_rate": 5e-05,
      "loss": 0.6669,
      "step": 120
    },
    {
      "epoch": 0.04,
      "learning_rate": 5e-05,
      "loss": 0.6719,
      "step": 130
    },
    {
      "epoch": 0.05,
      "learning_rate": 5e-05,
      "loss": 0.6684,
      "step": 140
    },
    {
      "epoch": 0.05,
      "learning_rate": 5e-05,
      "loss": 0.6675,
      "step": 150
    },
    {
      "epoch": 0.05,
      "learning_rate": 5e-05,
      "loss": 0.6806,
      "step": 160
    },
    {
      "epoch": 0.06,
      "learning_rate": 5e-05,
      "loss": 0.6743,
      "step": 170
    },
    {
      "epoch": 0.06,
      "learning_rate": 5e-05,
      "loss": 0.6735,
      "step": 180
    },
    {
      "epoch": 0.06,
      "learning_rate": 5e-05,
      "loss": 0.6739,
      "step": 190
    },
    {
      "epoch": 0.07,
      "learning_rate": 5e-05,
      "loss": 0.6739,
      "step": 200
    },
    {
      "epoch": 0.07,
      "learning_rate": 5e-05,
      "loss": 0.6761,
      "step": 210
    },
    {
      "epoch": 0.07,
      "learning_rate": 5e-05,
      "loss": 0.6714,
      "step": 220
    },
    {
      "epoch": 0.08,
      "learning_rate": 5e-05,
      "loss": 0.6707,
      "step": 230
    },
    {
      "epoch": 0.08,
      "learning_rate": 5e-05,
      "loss": 0.6651,
      "step": 240
    },
    {
      "epoch": 0.08,
      "learning_rate": 5e-05,
      "loss": 0.6757,
      "step": 250
    },
    {
      "epoch": 0.09,
      "learning_rate": 5e-05,
      "loss": 0.6759,
      "step": 260
    },
    {
      "epoch": 0.09,
      "learning_rate": 5e-05,
      "loss": 0.6698,
      "step": 270
    },
    {
      "epoch": 0.09,
      "learning_rate": 5e-05,
      "loss": 0.6687,
      "step": 280
    },
    {
      "epoch": 0.1,
      "learning_rate": 5e-05,
      "loss": 0.6697,
      "step": 290
    },
    {
      "epoch": 0.1,
      "learning_rate": 5e-05,
      "loss": 0.6678,
      "step": 300
    },
    {
      "epoch": 0.1,
      "learning_rate": 5e-05,
      "loss": 0.6604,
      "step": 310
    },
    {
      "epoch": 0.11,
      "learning_rate": 5e-05,
      "loss": 0.6669,
      "step": 320
    },
    {
      "epoch": 0.11,
      "learning_rate": 5e-05,
      "loss": 0.6655,
      "step": 330
    },
    {
      "epoch": 0.11,
      "learning_rate": 5e-05,
      "loss": 0.6569,
      "step": 340
    },
    {
      "epoch": 0.12,
      "learning_rate": 5e-05,
      "loss": 0.6666,
      "step": 350
    },
    {
      "epoch": 0.12,
      "learning_rate": 5e-05,
      "loss": 0.6647,
      "step": 360
    },
    {
      "epoch": 0.12,
      "learning_rate": 5e-05,
      "loss": 0.6612,
      "step": 370
    },
    {
      "epoch": 0.13,
      "learning_rate": 5e-05,
      "loss": 0.6616,
      "step": 380
    },
    {
      "epoch": 0.13,
      "learning_rate": 5e-05,
      "loss": 0.6607,
      "step": 390
    },
    {
      "epoch": 0.13,
      "learning_rate": 5e-05,
      "loss": 0.6596,
      "step": 400
    },
    {
      "epoch": 0.14,
      "learning_rate": 5e-05,
      "loss": 0.6542,
      "step": 410
    },
    {
      "epoch": 0.14,
      "learning_rate": 5e-05,
      "loss": 0.6606,
      "step": 420
    },
    {
      "epoch": 0.15,
      "learning_rate": 5e-05,
      "loss": 0.654,
      "step": 430
    },
    {
      "epoch": 0.15,
      "learning_rate": 5e-05,
      "loss": 0.6631,
      "step": 440
    },
    {
      "epoch": 0.15,
      "learning_rate": 5e-05,
      "loss": 0.6558,
      "step": 450
    },
    {
      "epoch": 0.16,
      "learning_rate": 5e-05,
      "loss": 0.6498,
      "step": 460
    },
    {
      "epoch": 0.16,
      "learning_rate": 5e-05,
      "loss": 0.6728,
      "step": 470
    },
    {
      "epoch": 0.16,
      "learning_rate": 5e-05,
      "loss": 0.6649,
      "step": 480
    },
    {
      "epoch": 0.17,
      "learning_rate": 5e-05,
      "loss": 0.6569,
      "step": 490
    },
    {
      "epoch": 0.17,
      "learning_rate": 5e-05,
      "loss": 0.6582,
      "step": 500
    },
    {
      "epoch": 0.17,
      "learning_rate": 5e-05,
      "loss": 0.6484,
      "step": 510
    },
    {
      "epoch": 0.18,
      "learning_rate": 5e-05,
      "loss": 0.6476,
      "step": 520
    },
    {
      "epoch": 0.18,
      "learning_rate": 5e-05,
      "loss": 0.6505,
      "step": 530
    },
    {
      "epoch": 0.18,
      "learning_rate": 5e-05,
      "loss": 0.6523,
      "step": 540
    },
    {
      "epoch": 0.19,
      "learning_rate": 5e-05,
      "loss": 0.6521,
      "step": 550
    },
    {
      "epoch": 0.19,
      "learning_rate": 5e-05,
      "loss": 0.6467,
      "step": 560
    },
    {
      "epoch": 0.19,
      "learning_rate": 5e-05,
      "loss": 0.6628,
      "step": 570
    },
    {
      "epoch": 0.2,
      "learning_rate": 5e-05,
      "loss": 0.6523,
      "step": 580
    },
    {
      "epoch": 0.2,
      "learning_rate": 5e-05,
      "loss": 0.655,
      "step": 590
    },
    {
      "epoch": 0.2,
      "learning_rate": 5e-05,
      "loss": 0.6496,
      "step": 600
    },
    {
      "epoch": 0.21,
      "learning_rate": 5e-05,
      "loss": 0.6498,
      "step": 610
    },
    {
      "epoch": 0.21,
      "learning_rate": 5e-05,
      "loss": 0.6498,
      "step": 620
    },
    {
      "epoch": 0.21,
      "learning_rate": 5e-05,
      "loss": 0.6542,
      "step": 630
    },
    {
      "epoch": 0.22,
      "learning_rate": 5e-05,
      "loss": 0.6445,
      "step": 640
    },
    {
      "epoch": 0.22,
      "learning_rate": 5e-05,
      "loss": 0.6511,
      "step": 650
    },
    {
      "epoch": 0.22,
      "learning_rate": 5e-05,
      "loss": 0.6482,
      "step": 660
    },
    {
      "epoch": 0.23,
      "learning_rate": 5e-05,
      "loss": 0.6462,
      "step": 670
    },
    {
      "epoch": 0.23,
      "learning_rate": 5e-05,
      "loss": 0.6623,
      "step": 680
    },
    {
      "epoch": 0.23,
      "learning_rate": 5e-05,
      "loss": 0.6473,
      "step": 690
    },
    {
      "epoch": 0.24,
      "learning_rate": 5e-05,
      "loss": 0.651,
      "step": 700
    },
    {
      "epoch": 0.24,
      "learning_rate": 5e-05,
      "loss": 0.6453,
      "step": 710
    },
    {
      "epoch": 0.24,
      "learning_rate": 5e-05,
      "loss": 0.6508,
      "step": 720
    },
    {
      "epoch": 0.25,
      "learning_rate": 5e-05,
      "loss": 0.6541,
      "step": 730
    },
    {
      "epoch": 0.25,
      "learning_rate": 5e-05,
      "loss": 0.6526,
      "step": 740
    },
    {
      "epoch": 0.25,
      "learning_rate": 5e-05,
      "loss": 0.6391,
      "step": 750
    },
    {
      "epoch": 0.26,
      "learning_rate": 5e-05,
      "loss": 0.6467,
      "step": 760
    },
    {
      "epoch": 0.26,
      "learning_rate": 5e-05,
      "loss": 0.6505,
      "step": 770
    },
    {
      "epoch": 0.26,
      "learning_rate": 5e-05,
      "loss": 0.6405,
      "step": 780
    },
    {
      "epoch": 0.27,
      "learning_rate": 5e-05,
      "loss": 0.6533,
      "step": 790
    },
    {
      "epoch": 0.27,
      "learning_rate": 5e-05,
      "loss": 0.6432,
      "step": 800
    },
    {
      "epoch": 0.27,
      "learning_rate": 5e-05,
      "loss": 0.6448,
      "step": 810
    },
    {
      "epoch": 0.28,
      "learning_rate": 5e-05,
      "loss": 0.6361,
      "step": 820
    },
    {
      "epoch": 0.28,
      "learning_rate": 5e-05,
      "loss": 0.6416,
      "step": 830
    },
    {
      "epoch": 0.28,
      "learning_rate": 5e-05,
      "loss": 0.6381,
      "step": 840
    },
    {
      "epoch": 0.29,
      "learning_rate": 5e-05,
      "loss": 0.6461,
      "step": 850
    },
    {
      "epoch": 0.29,
      "learning_rate": 5e-05,
      "loss": 0.6439,
      "step": 860
    },
    {
      "epoch": 0.29,
      "learning_rate": 5e-05,
      "loss": 0.6433,
      "step": 870
    },
    {
      "epoch": 0.3,
      "learning_rate": 5e-05,
      "loss": 0.6468,
      "step": 880
    },
    {
      "epoch": 0.3,
      "learning_rate": 5e-05,
      "loss": 0.6507,
      "step": 890
    },
    {
      "epoch": 0.3,
      "learning_rate": 5e-05,
      "loss": 0.6481,
      "step": 900
    },
    {
      "epoch": 0.31,
      "learning_rate": 5e-05,
      "loss": 0.6442,
      "step": 910
    },
    {
      "epoch": 0.31,
      "learning_rate": 5e-05,
      "loss": 0.6384,
      "step": 920
    },
    {
      "epoch": 0.31,
      "learning_rate": 5e-05,
      "loss": 0.6418,
      "step": 930
    },
    {
      "epoch": 0.32,
      "learning_rate": 5e-05,
      "loss": 0.6505,
      "step": 940
    },
    {
      "epoch": 0.32,
      "learning_rate": 5e-05,
      "loss": 0.637,
      "step": 950
    },
    {
      "epoch": 0.32,
      "learning_rate": 5e-05,
      "loss": 0.6381,
      "step": 960
    },
    {
      "epoch": 0.33,
      "learning_rate": 5e-05,
      "loss": 0.6377,
      "step": 970
    },
    {
      "epoch": 0.33,
      "learning_rate": 5e-05,
      "loss": 0.6737,
      "step": 980
    },
    {
      "epoch": 0.33,
      "learning_rate": 5e-05,
      "loss": 0.64,
      "step": 990
    },
    {
      "epoch": 0.34,
      "learning_rate": 5e-05,
      "loss": 0.641,
      "step": 1000
    },
    {
      "epoch": 0.34,
      "learning_rate": 5e-05,
      "loss": 0.6335,
      "step": 1010
    },
    {
      "epoch": 0.34,
      "learning_rate": 5e-05,
      "loss": 0.6583,
      "step": 1020
    },
    {
      "epoch": 0.35,
      "learning_rate": 5e-05,
      "loss": 0.6388,
      "step": 1030
    },
    {
      "epoch": 0.35,
      "learning_rate": 5e-05,
      "loss": 0.6357,
      "step": 1040
    },
    {
      "epoch": 0.35,
      "learning_rate": 5e-05,
      "loss": 0.6326,
      "step": 1050
    },
    {
      "epoch": 0.36,
      "learning_rate": 5e-05,
      "loss": 0.6349,
      "step": 1060
    },
    {
      "epoch": 0.36,
      "learning_rate": 5e-05,
      "loss": 0.6468,
      "step": 1070
    },
    {
      "epoch": 0.36,
      "learning_rate": 5e-05,
      "loss": 0.6384,
      "step": 1080
    },
    {
      "epoch": 0.37,
      "learning_rate": 5e-05,
      "loss": 0.6427,
      "step": 1090
    },
    {
      "epoch": 0.37,
      "learning_rate": 5e-05,
      "loss": 0.6454,
      "step": 1100
    },
    {
      "epoch": 0.37,
      "learning_rate": 5e-05,
      "loss": 0.6379,
      "step": 1110
    },
    {
      "epoch": 0.38,
      "learning_rate": 5e-05,
      "loss": 0.6352,
      "step": 1120
    },
    {
      "epoch": 0.38,
      "learning_rate": 5e-05,
      "loss": 0.6299,
      "step": 1130
    },
    {
      "epoch": 0.38,
      "learning_rate": 5e-05,
      "loss": 0.6381,
      "step": 1140
    },
    {
      "epoch": 0.39,
      "learning_rate": 5e-05,
      "loss": 0.6322,
      "step": 1150
    },
    {
      "epoch": 0.39,
      "learning_rate": 5e-05,
      "loss": 0.653,
      "step": 1160
    },
    {
      "epoch": 0.39,
      "learning_rate": 5e-05,
      "loss": 0.6367,
      "step": 1170
    },
    {
      "epoch": 0.4,
      "learning_rate": 5e-05,
      "loss": 0.6477,
      "step": 1180
    },
    {
      "epoch": 0.4,
      "learning_rate": 5e-05,
      "loss": 0.6413,
      "step": 1190
    },
    {
      "epoch": 0.4,
      "learning_rate": 5e-05,
      "loss": 0.6374,
      "step": 1200
    },
    {
      "epoch": 0.41,
      "learning_rate": 5e-05,
      "loss": 0.6359,
      "step": 1210
    },
    {
      "epoch": 0.41,
      "learning_rate": 5e-05,
      "loss": 0.6325,
      "step": 1220
    },
    {
      "epoch": 0.41,
      "learning_rate": 5e-05,
      "loss": 0.6454,
      "step": 1230
    },
    {
      "epoch": 0.42,
      "learning_rate": 5e-05,
      "loss": 0.641,
      "step": 1240
    },
    {
      "epoch": 0.42,
      "learning_rate": 5e-05,
      "loss": 0.6332,
      "step": 1250
    },
    {
      "epoch": 0.43,
      "learning_rate": 5e-05,
      "loss": 0.6338,
      "step": 1260
    },
    {
      "epoch": 0.43,
      "learning_rate": 5e-05,
      "loss": 0.6351,
      "step": 1270
    },
    {
      "epoch": 0.43,
      "learning_rate": 5e-05,
      "loss": 0.6365,
      "step": 1280
    },
    {
      "epoch": 0.44,
      "learning_rate": 5e-05,
      "loss": 0.6376,
      "step": 1290
    },
    {
      "epoch": 0.44,
      "learning_rate": 5e-05,
      "loss": 0.6367,
      "step": 1300
    },
    {
      "epoch": 0.44,
      "learning_rate": 5e-05,
      "loss": 0.6386,
      "step": 1310
    },
    {
      "epoch": 0.45,
      "learning_rate": 5e-05,
      "loss": 0.6433,
      "step": 1320
    },
    {
      "epoch": 0.45,
      "learning_rate": 5e-05,
      "loss": 0.6377,
      "step": 1330
    },
    {
      "epoch": 0.45,
      "learning_rate": 5e-05,
      "loss": 0.6389,
      "step": 1340
    },
    {
      "epoch": 0.46,
      "learning_rate": 5e-05,
      "loss": 0.6298,
      "step": 1350
    },
    {
      "epoch": 0.46,
      "learning_rate": 5e-05,
      "loss": 0.6304,
      "step": 1360
    },
    {
      "epoch": 0.46,
      "learning_rate": 5e-05,
      "loss": 0.6361,
      "step": 1370
    },
    {
      "epoch": 0.47,
      "learning_rate": 5e-05,
      "loss": 0.6301,
      "step": 1380
    },
    {
      "epoch": 0.47,
      "learning_rate": 5e-05,
      "loss": 0.6303,
      "step": 1390
    },
    {
      "epoch": 0.47,
      "learning_rate": 5e-05,
      "loss": 0.6299,
      "step": 1400
    },
    {
      "epoch": 0.48,
      "learning_rate": 5e-05,
      "loss": 0.6376,
      "step": 1410
    },
    {
      "epoch": 0.48,
      "learning_rate": 5e-05,
      "loss": 0.6315,
      "step": 1420
    },
    {
      "epoch": 0.48,
      "learning_rate": 5e-05,
      "loss": 0.6447,
      "step": 1430
    },
    {
      "epoch": 0.49,
      "learning_rate": 5e-05,
      "loss": 0.6349,
      "step": 1440
    },
    {
      "epoch": 0.49,
      "learning_rate": 5e-05,
      "loss": 0.6343,
      "step": 1450
    },
    {
      "epoch": 0.49,
      "learning_rate": 5e-05,
      "loss": 0.6325,
      "step": 1460
    },
    {
      "epoch": 0.5,
      "learning_rate": 5e-05,
      "loss": 0.6343,
      "step": 1470
    },
    {
      "epoch": 0.5,
      "learning_rate": 5e-05,
      "loss": 0.6389,
      "step": 1480
    },
    {
      "epoch": 0.5,
      "learning_rate": 5e-05,
      "loss": 0.6359,
      "step": 1490
    },
    {
      "epoch": 0.51,
      "learning_rate": 5e-05,
      "loss": 0.6302,
      "step": 1500
    },
    {
      "epoch": 0.51,
      "learning_rate": 5e-05,
      "loss": 0.6329,
      "step": 1510
    },
    {
      "epoch": 0.51,
      "learning_rate": 5e-05,
      "loss": 0.6315,
      "step": 1520
    },
    {
      "epoch": 0.52,
      "learning_rate": 5e-05,
      "loss": 0.6237,
      "step": 1530
    },
    {
      "epoch": 0.52,
      "learning_rate": 5e-05,
      "loss": 0.6347,
      "step": 1540
    },
    {
      "epoch": 0.52,
      "learning_rate": 5e-05,
      "loss": 0.6293,
      "step": 1550
    },
    {
      "epoch": 0.53,
      "learning_rate": 5e-05,
      "loss": 0.6428,
      "step": 1560
    },
    {
      "epoch": 0.53,
      "learning_rate": 5e-05,
      "loss": 0.6466,
      "step": 1570
    },
    {
      "epoch": 0.53,
      "learning_rate": 5e-05,
      "loss": 0.6391,
      "step": 1580
    },
    {
      "epoch": 0.54,
      "learning_rate": 5e-05,
      "loss": 0.6371,
      "step": 1590
    },
    {
      "epoch": 0.54,
      "learning_rate": 5e-05,
      "loss": 0.6357,
      "step": 1600
    },
    {
      "epoch": 0.54,
      "learning_rate": 5e-05,
      "loss": 0.6323,
      "step": 1610
    },
    {
      "epoch": 0.55,
      "learning_rate": 5e-05,
      "loss": 0.6239,
      "step": 1620
    },
    {
      "epoch": 0.55,
      "learning_rate": 5e-05,
      "loss": 0.6316,
      "step": 1630
    },
    {
      "epoch": 0.55,
      "learning_rate": 5e-05,
      "loss": 0.638,
      "step": 1640
    },
    {
      "epoch": 0.56,
      "learning_rate": 5e-05,
      "loss": 0.6391,
      "step": 1650
    },
    {
      "epoch": 0.56,
      "learning_rate": 5e-05,
      "loss": 0.6494,
      "step": 1660
    },
    {
      "epoch": 0.56,
      "learning_rate": 5e-05,
      "loss": 0.6448,
      "step": 1670
    },
    {
      "epoch": 0.57,
      "learning_rate": 5e-05,
      "loss": 0.6454,
      "step": 1680
    },
    {
      "epoch": 0.57,
      "learning_rate": 5e-05,
      "loss": 0.6521,
      "step": 1690
    },
    {
      "epoch": 0.57,
      "learning_rate": 5e-05,
      "loss": 0.6653,
      "step": 1700
    },
    {
      "epoch": 0.58,
      "learning_rate": 5e-05,
      "loss": 0.6432,
      "step": 1710
    },
    {
      "epoch": 0.58,
      "learning_rate": 5e-05,
      "loss": 0.6284,
      "step": 1720
    },
    {
      "epoch": 0.58,
      "learning_rate": 5e-05,
      "loss": 0.6347,
      "step": 1730
    },
    {
      "epoch": 0.59,
      "learning_rate": 5e-05,
      "loss": 0.6373,
      "step": 1740
    },
    {
      "epoch": 0.59,
      "learning_rate": 5e-05,
      "loss": 0.635,
      "step": 1750
    },
    {
      "epoch": 0.59,
      "learning_rate": 5e-05,
      "loss": 0.6322,
      "step": 1760
    },
    {
      "epoch": 0.6,
      "learning_rate": 5e-05,
      "loss": 0.6297,
      "step": 1770
    },
    {
      "epoch": 0.6,
      "learning_rate": 5e-05,
      "loss": 0.6532,
      "step": 1780
    },
    {
      "epoch": 0.6,
      "learning_rate": 5e-05,
      "loss": 0.644,
      "step": 1790
    },
    {
      "epoch": 0.61,
      "learning_rate": 5e-05,
      "loss": 0.6428,
      "step": 1800
    },
    {
      "epoch": 0.61,
      "learning_rate": 5e-05,
      "loss": 0.6344,
      "step": 1810
    },
    {
      "epoch": 0.61,
      "learning_rate": 5e-05,
      "loss": 0.6268,
      "step": 1820
    },
    {
      "epoch": 0.62,
      "learning_rate": 5e-05,
      "loss": 0.6273,
      "step": 1830
    },
    {
      "epoch": 0.62,
      "learning_rate": 5e-05,
      "loss": 0.6372,
      "step": 1840
    },
    {
      "epoch": 0.62,
      "learning_rate": 5e-05,
      "loss": 0.6398,
      "step": 1850
    },
    {
      "epoch": 0.63,
      "learning_rate": 5e-05,
      "loss": 0.6491,
      "step": 1860
    },
    {
      "epoch": 0.63,
      "learning_rate": 5e-05,
      "loss": 0.6318,
      "step": 1870
    },
    {
      "epoch": 0.63,
      "learning_rate": 5e-05,
      "loss": 0.6339,
      "step": 1880
    },
    {
      "epoch": 0.64,
      "learning_rate": 5e-05,
      "loss": 0.6355,
      "step": 1890
    },
    {
      "epoch": 0.64,
      "learning_rate": 5e-05,
      "loss": 0.6381,
      "step": 1900
    },
    {
      "epoch": 0.64,
      "learning_rate": 5e-05,
      "loss": 0.6242,
      "step": 1910
    },
    {
      "epoch": 0.65,
      "learning_rate": 5e-05,
      "loss": 0.6314,
      "step": 1920
    },
    {
      "epoch": 0.65,
      "learning_rate": 5e-05,
      "loss": 0.6327,
      "step": 1930
    },
    {
      "epoch": 0.65,
      "learning_rate": 5e-05,
      "loss": 0.6354,
      "step": 1940
    },
    {
      "epoch": 0.66,
      "learning_rate": 5e-05,
      "loss": 0.627,
      "step": 1950
    },
    {
      "epoch": 0.66,
      "learning_rate": 5e-05,
      "loss": 0.6383,
      "step": 1960
    },
    {
      "epoch": 0.66,
      "learning_rate": 5e-05,
      "loss": 0.63,
      "step": 1970
    },
    {
      "epoch": 0.67,
      "learning_rate": 5e-05,
      "loss": 0.6313,
      "step": 1980
    },
    {
      "epoch": 0.67,
      "learning_rate": 5e-05,
      "loss": 0.6367,
      "step": 1990
    },
    {
      "epoch": 0.67,
      "learning_rate": 5e-05,
      "loss": 0.6393,
      "step": 2000
    },
    {
      "epoch": 0.68,
      "learning_rate": 5e-05,
      "loss": 0.644,
      "step": 2010
    },
    {
      "epoch": 0.68,
      "learning_rate": 5e-05,
      "loss": 0.644,
      "step": 2020
    },
    {
      "epoch": 0.68,
      "learning_rate": 5e-05,
      "loss": 0.6387,
      "step": 2030
    },
    {
      "epoch": 0.69,
      "learning_rate": 5e-05,
      "loss": 0.6388,
      "step": 2040
    },
    {
      "epoch": 0.69,
      "learning_rate": 5e-05,
      "loss": 0.6337,
      "step": 2050
    },
    {
      "epoch": 0.69,
      "learning_rate": 5e-05,
      "loss": 0.6284,
      "step": 2060
    },
    {
      "epoch": 0.7,
      "learning_rate": 5e-05,
      "loss": 0.6292,
      "step": 2070
    },
    {
      "epoch": 0.7,
      "learning_rate": 5e-05,
      "loss": 0.6267,
      "step": 2080
    },
    {
      "epoch": 0.71,
      "learning_rate": 5e-05,
      "loss": 0.6312,
      "step": 2090
    },
    {
      "epoch": 0.71,
      "learning_rate": 5e-05,
      "loss": 0.634,
      "step": 2100
    },
    {
      "epoch": 0.71,
      "learning_rate": 5e-05,
      "loss": 0.6267,
      "step": 2110
    },
    {
      "epoch": 0.72,
      "learning_rate": 5e-05,
      "loss": 0.6227,
      "step": 2120
    },
    {
      "epoch": 0.72,
      "learning_rate": 5e-05,
      "loss": 0.6319,
      "step": 2130
    },
    {
      "epoch": 0.72,
      "learning_rate": 5e-05,
      "loss": 0.627,
      "step": 2140
    },
    {
      "epoch": 0.73,
      "learning_rate": 5e-05,
      "loss": 0.627,
      "step": 2150
    },
    {
      "epoch": 0.73,
      "learning_rate": 5e-05,
      "loss": 0.6425,
      "step": 2160
    },
    {
      "epoch": 0.73,
      "learning_rate": 5e-05,
      "loss": 0.6301,
      "step": 2170
    },
    {
      "epoch": 0.74,
      "learning_rate": 5e-05,
      "loss": 0.6283,
      "step": 2180
    },
    {
      "epoch": 0.74,
      "learning_rate": 5e-05,
      "loss": 0.6311,
      "step": 2190
    },
    {
      "epoch": 0.74,
      "learning_rate": 5e-05,
      "loss": 0.6316,
      "step": 2200
    },
    {
      "epoch": 0.75,
      "learning_rate": 5e-05,
      "loss": 0.6277,
      "step": 2210
    },
    {
      "epoch": 0.75,
      "learning_rate": 5e-05,
      "loss": 0.6399,
      "step": 2220
    },
    {
      "epoch": 0.75,
      "learning_rate": 5e-05,
      "loss": 0.6324,
      "step": 2230
    },
    {
      "epoch": 0.76,
      "learning_rate": 5e-05,
      "loss": 0.6268,
      "step": 2240
    },
    {
      "epoch": 0.76,
      "learning_rate": 5e-05,
      "loss": 0.6306,
      "step": 2250
    },
    {
      "epoch": 0.76,
      "learning_rate": 5e-05,
      "loss": 0.6258,
      "step": 2260
    },
    {
      "epoch": 0.77,
      "learning_rate": 5e-05,
      "loss": 0.6203,
      "step": 2270
    },
    {
      "epoch": 0.77,
      "learning_rate": 5e-05,
      "loss": 0.6305,
      "step": 2280
    },
    {
      "epoch": 0.77,
      "learning_rate": 5e-05,
      "loss": 0.6385,
      "step": 2290
    },
    {
      "epoch": 0.78,
      "learning_rate": 5e-05,
      "loss": 0.6288,
      "step": 2300
    },
    {
      "epoch": 0.78,
      "learning_rate": 5e-05,
      "loss": 0.6283,
      "step": 2310
    },
    {
      "epoch": 0.78,
      "learning_rate": 5e-05,
      "loss": 0.6349,
      "step": 2320
    },
    {
      "epoch": 0.79,
      "learning_rate": 5e-05,
      "loss": 0.625,
      "step": 2330
    },
    {
      "epoch": 0.79,
      "learning_rate": 5e-05,
      "loss": 0.6223,
      "step": 2340
    },
    {
      "epoch": 0.79,
      "learning_rate": 5e-05,
      "loss": 0.6382,
      "step": 2350
    },
    {
      "epoch": 0.8,
      "learning_rate": 5e-05,
      "loss": 0.6345,
      "step": 2360
    },
    {
      "epoch": 0.8,
      "learning_rate": 5e-05,
      "loss": 0.6462,
      "step": 2370
    },
    {
      "epoch": 0.8,
      "learning_rate": 5e-05,
      "loss": 0.6367,
      "step": 2380
    },
    {
      "epoch": 0.81,
      "learning_rate": 5e-05,
      "loss": 0.6358,
      "step": 2390
    },
    {
      "epoch": 0.81,
      "learning_rate": 5e-05,
      "loss": 0.6357,
      "step": 2400
    },
    {
      "epoch": 0.81,
      "learning_rate": 5e-05,
      "loss": 0.6338,
      "step": 2410
    },
    {
      "epoch": 0.82,
      "learning_rate": 5e-05,
      "loss": 0.6255,
      "step": 2420
    },
    {
      "epoch": 0.82,
      "learning_rate": 5e-05,
      "loss": 0.6199,
      "step": 2430
    },
    {
      "epoch": 0.82,
      "learning_rate": 5e-05,
      "loss": 0.6252,
      "step": 2440
    },
    {
      "epoch": 0.83,
      "learning_rate": 5e-05,
      "loss": 0.6258,
      "step": 2450
    },
    {
      "epoch": 0.83,
      "learning_rate": 5e-05,
      "loss": 0.6235,
      "step": 2460
    },
    {
      "epoch": 0.83,
      "learning_rate": 5e-05,
      "loss": 0.6328,
      "step": 2470
    },
    {
      "epoch": 0.84,
      "learning_rate": 5e-05,
      "loss": 0.6359,
      "step": 2480
    },
    {
      "epoch": 0.84,
      "learning_rate": 5e-05,
      "loss": 0.6329,
      "step": 2490
    },
    {
      "epoch": 0.84,
      "learning_rate": 5e-05,
      "loss": 0.624,
      "step": 2500
    },
    {
      "epoch": 0.85,
      "learning_rate": 5e-05,
      "loss": 0.6463,
      "step": 2510
    },
    {
      "epoch": 0.85,
      "learning_rate": 5e-05,
      "loss": 0.6195,
      "step": 2520
    },
    {
      "epoch": 0.85,
      "learning_rate": 5e-05,
      "loss": 0.6446,
      "step": 2530
    },
    {
      "epoch": 0.86,
      "learning_rate": 5e-05,
      "loss": 0.6264,
      "step": 2540
    },
    {
      "epoch": 0.86,
      "learning_rate": 5e-05,
      "loss": 0.6234,
      "step": 2550
    },
    {
      "epoch": 0.86,
      "learning_rate": 5e-05,
      "loss": 0.6266,
      "step": 2560
    },
    {
      "epoch": 0.87,
      "learning_rate": 5e-05,
      "loss": 0.6209,
      "step": 2570
    },
    {
      "epoch": 0.87,
      "learning_rate": 5e-05,
      "loss": 0.6213,
      "step": 2580
    },
    {
      "epoch": 0.87,
      "learning_rate": 5e-05,
      "loss": 0.6226,
      "step": 2590
    },
    {
      "epoch": 0.88,
      "learning_rate": 5e-05,
      "loss": 0.627,
      "step": 2600
    },
    {
      "epoch": 0.88,
      "learning_rate": 5e-05,
      "loss": 0.6224,
      "step": 2610
    },
    {
      "epoch": 0.88,
      "learning_rate": 5e-05,
      "loss": 0.6212,
      "step": 2620
    },
    {
      "epoch": 0.89,
      "learning_rate": 5e-05,
      "loss": 0.6141,
      "step": 2630
    },
    {
      "epoch": 0.89,
      "learning_rate": 5e-05,
      "loss": 0.6249,
      "step": 2640
    },
    {
      "epoch": 0.89,
      "learning_rate": 5e-05,
      "loss": 0.6204,
      "step": 2650
    },
    {
      "epoch": 0.9,
      "learning_rate": 5e-05,
      "loss": 0.6259,
      "step": 2660
    },
    {
      "epoch": 0.9,
      "learning_rate": 5e-05,
      "loss": 0.6206,
      "step": 2670
    },
    {
      "epoch": 0.9,
      "learning_rate": 5e-05,
      "loss": 0.6241,
      "step": 2680
    },
    {
      "epoch": 0.91,
      "learning_rate": 5e-05,
      "loss": 0.6237,
      "step": 2690
    },
    {
      "epoch": 0.91,
      "learning_rate": 5e-05,
      "loss": 0.621,
      "step": 2700
    },
    {
      "epoch": 0.91,
      "learning_rate": 5e-05,
      "loss": 0.622,
      "step": 2710
    },
    {
      "epoch": 0.92,
      "learning_rate": 5e-05,
      "loss": 0.6217,
      "step": 2720
    },
    {
      "epoch": 0.92,
      "learning_rate": 5e-05,
      "loss": 0.6206,
      "step": 2730
    },
    {
      "epoch": 0.92,
      "learning_rate": 5e-05,
      "loss": 0.6366,
      "step": 2740
    },
    {
      "epoch": 0.93,
      "learning_rate": 5e-05,
      "loss": 0.6341,
      "step": 2750
    },
    {
      "epoch": 0.93,
      "learning_rate": 5e-05,
      "loss": 0.6414,
      "step": 2760
    },
    {
      "epoch": 0.93,
      "learning_rate": 5e-05,
      "loss": 0.6208,
      "step": 2770
    },
    {
      "epoch": 0.94,
      "learning_rate": 5e-05,
      "loss": 0.6218,
      "step": 2780
    },
    {
      "epoch": 0.94,
      "learning_rate": 5e-05,
      "loss": 0.6278,
      "step": 2790
    },
    {
      "epoch": 0.94,
      "learning_rate": 5e-05,
      "loss": 0.6295,
      "step": 2800
    },
    {
      "epoch": 0.95,
      "learning_rate": 5e-05,
      "loss": 0.6194,
      "step": 2810
    },
    {
      "epoch": 0.95,
      "learning_rate": 5e-05,
      "loss": 0.6278,
      "step": 2820
    },
    {
      "epoch": 0.95,
      "learning_rate": 5e-05,
      "loss": 0.6248,
      "step": 2830
    },
    {
      "epoch": 0.96,
      "learning_rate": 5e-05,
      "loss": 0.6213,
      "step": 2840
    },
    {
      "epoch": 0.96,
      "learning_rate": 5e-05,
      "loss": 0.6219,
      "step": 2850
    },
    {
      "epoch": 0.96,
      "learning_rate": 5e-05,
      "loss": 0.6362,
      "step": 2860
    },
    {
      "epoch": 0.97,
      "learning_rate": 5e-05,
      "loss": 0.6377,
      "step": 2870
    },
    {
      "epoch": 0.97,
      "learning_rate": 5e-05,
      "loss": 0.6333,
      "step": 2880
    },
    {
      "epoch": 0.97,
      "learning_rate": 5e-05,
      "loss": 0.6526,
      "step": 2890
    },
    {
      "epoch": 0.98,
      "learning_rate": 5e-05,
      "loss": 0.6221,
      "step": 2900
    },
    {
      "epoch": 0.98,
      "learning_rate": 5e-05,
      "loss": 0.6274,
      "step": 2910
    },
    {
      "epoch": 0.99,
      "learning_rate": 5e-05,
      "loss": 0.6226,
      "step": 2920
    },
    {
      "epoch": 0.99,
      "learning_rate": 5e-05,
      "loss": 0.6217,
      "step": 2930
    },
    {
      "epoch": 0.99,
      "learning_rate": 5e-05,
      "loss": 0.6222,
      "step": 2940
    },
    {
      "epoch": 1.0,
      "learning_rate": 5e-05,
      "loss": 0.6246,
      "step": 2950
    },
    {
      "epoch": 1.0,
      "learning_rate": 5e-05,
      "loss": 0.6311,
      "step": 2960
    },
    {
      "epoch": 1.0,
      "learning_rate": 5e-05,
      "loss": 0.6182,
      "step": 2970
    },
    {
      "epoch": 1.01,
      "learning_rate": 5e-05,
      "loss": 0.6112,
      "step": 2980
    },
    {
      "epoch": 1.01,
      "learning_rate": 5e-05,
      "loss": 0.6145,
      "step": 2990
    },
    {
      "epoch": 1.01,
      "learning_rate": 5e-05,
      "loss": 0.6233,
      "step": 3000
    },
    {
      "epoch": 1.02,
      "learning_rate": 5e-05,
      "loss": 0.6314,
      "step": 3010
    },
    {
      "epoch": 1.02,
      "learning_rate": 5e-05,
      "loss": 0.6157,
      "step": 3020
    },
    {
      "epoch": 1.02,
      "learning_rate": 5e-05,
      "loss": 0.6247,
      "step": 3030
    },
    {
      "epoch": 1.03,
      "learning_rate": 5e-05,
      "loss": 0.6207,
      "step": 3040
    },
    {
      "epoch": 1.03,
      "learning_rate": 5e-05,
      "loss": 0.6357,
      "step": 3050
    },
    {
      "epoch": 1.03,
      "learning_rate": 5e-05,
      "loss": 0.6295,
      "step": 3060
    },
    {
      "epoch": 1.04,
      "learning_rate": 5e-05,
      "loss": 0.6096,
      "step": 3070
    },
    {
      "epoch": 1.04,
      "learning_rate": 5e-05,
      "loss": 0.6205,
      "step": 3080
    },
    {
      "epoch": 1.04,
      "learning_rate": 5e-05,
      "loss": 0.6187,
      "step": 3090
    },
    {
      "epoch": 1.05,
      "learning_rate": 5e-05,
      "loss": 0.6242,
      "step": 3100
    },
    {
      "epoch": 1.05,
      "learning_rate": 5e-05,
      "loss": 0.6204,
      "step": 3110
    },
    {
      "epoch": 1.05,
      "learning_rate": 5e-05,
      "loss": 0.6233,
      "step": 3120
    },
    {
      "epoch": 1.06,
      "learning_rate": 5e-05,
      "loss": 0.6235,
      "step": 3130
    },
    {
      "epoch": 1.06,
      "learning_rate": 5e-05,
      "loss": 0.6207,
      "step": 3140
    },
    {
      "epoch": 1.06,
      "learning_rate": 5e-05,
      "loss": 0.6137,
      "step": 3150
    },
    {
      "epoch": 1.07,
      "learning_rate": 5e-05,
      "loss": 0.6151,
      "step": 3160
    },
    {
      "epoch": 1.07,
      "learning_rate": 5e-05,
      "loss": 0.6189,
      "step": 3170
    },
    {
      "epoch": 1.07,
      "learning_rate": 5e-05,
      "loss": 0.6212,
      "step": 3180
    },
    {
      "epoch": 1.08,
      "learning_rate": 5e-05,
      "loss": 0.6173,
      "step": 3190
    },
    {
      "epoch": 1.08,
      "learning_rate": 5e-05,
      "loss": 0.6224,
      "step": 3200
    },
    {
      "epoch": 1.08,
      "learning_rate": 5e-05,
      "loss": 0.612,
      "step": 3210
    },
    {
      "epoch": 1.09,
      "learning_rate": 5e-05,
      "loss": 0.6159,
      "step": 3220
    },
    {
      "epoch": 1.09,
      "learning_rate": 5e-05,
      "loss": 0.6335,
      "step": 3230
    },
    {
      "epoch": 1.09,
      "learning_rate": 5e-05,
      "loss": 0.6117,
      "step": 3240
    },
    {
      "epoch": 1.1,
      "learning_rate": 5e-05,
      "loss": 0.6171,
      "step": 3250
    },
    {
      "epoch": 1.1,
      "learning_rate": 5e-05,
      "loss": 0.6152,
      "step": 3260
    },
    {
      "epoch": 1.1,
      "learning_rate": 5e-05,
      "loss": 0.6112,
      "step": 3270
    },
    {
      "epoch": 1.11,
      "learning_rate": 5e-05,
      "loss": 0.6151,
      "step": 3280
    },
    {
      "epoch": 1.11,
      "learning_rate": 5e-05,
      "loss": 0.6153,
      "step": 3290
    },
    {
      "epoch": 1.11,
      "learning_rate": 5e-05,
      "loss": 0.6238,
      "step": 3300
    },
    {
      "epoch": 1.12,
      "learning_rate": 5e-05,
      "loss": 0.6266,
      "step": 3310
    },
    {
      "epoch": 1.12,
      "learning_rate": 5e-05,
      "loss": 0.6185,
      "step": 3320
    },
    {
      "epoch": 1.12,
      "learning_rate": 5e-05,
      "loss": 0.6071,
      "step": 3330
    },
    {
      "epoch": 1.13,
      "learning_rate": 5e-05,
      "loss": 0.6188,
      "step": 3340
    },
    {
      "epoch": 1.13,
      "learning_rate": 5e-05,
      "loss": 0.6204,
      "step": 3350
    },
    {
      "epoch": 1.13,
      "learning_rate": 5e-05,
      "loss": 0.6185,
      "step": 3360
    },
    {
      "epoch": 1.14,
      "learning_rate": 5e-05,
      "loss": 0.6229,
      "step": 3370
    },
    {
      "epoch": 1.14,
      "learning_rate": 5e-05,
      "loss": 0.6241,
      "step": 3380
    },
    {
      "epoch": 1.14,
      "learning_rate": 5e-05,
      "loss": 0.618,
      "step": 3390
    },
    {
      "epoch": 1.15,
      "learning_rate": 5e-05,
      "loss": 0.6137,
      "step": 3400
    },
    {
      "epoch": 1.15,
      "learning_rate": 5e-05,
      "loss": 0.6197,
      "step": 3410
    },
    {
      "epoch": 1.15,
      "learning_rate": 5e-05,
      "loss": 0.625,
      "step": 3420
    },
    {
      "epoch": 1.16,
      "learning_rate": 5e-05,
      "loss": 0.6217,
      "step": 3430
    },
    {
      "epoch": 1.16,
      "learning_rate": 5e-05,
      "loss": 0.6234,
      "step": 3440
    },
    {
      "epoch": 1.16,
      "learning_rate": 5e-05,
      "loss": 0.6185,
      "step": 3450
    },
    {
      "epoch": 1.17,
      "learning_rate": 5e-05,
      "loss": 0.6107,
      "step": 3460
    },
    {
      "epoch": 1.17,
      "learning_rate": 5e-05,
      "loss": 0.6189,
      "step": 3470
    },
    {
      "epoch": 1.17,
      "learning_rate": 5e-05,
      "loss": 0.6173,
      "step": 3480
    },
    {
      "epoch": 1.18,
      "learning_rate": 5e-05,
      "loss": 0.6143,
      "step": 3490
    },
    {
      "epoch": 1.18,
      "learning_rate": 5e-05,
      "loss": 0.6243,
      "step": 3500
    },
    {
      "epoch": 1.18,
      "learning_rate": 5e-05,
      "loss": 0.6156,
      "step": 3510
    },
    {
      "epoch": 1.19,
      "learning_rate": 5e-05,
      "loss": 0.6129,
      "step": 3520
    },
    {
      "epoch": 1.19,
      "learning_rate": 5e-05,
      "loss": 0.6131,
      "step": 3530
    },
    {
      "epoch": 1.19,
      "learning_rate": 5e-05,
      "loss": 0.621,
      "step": 3540
    },
    {
      "epoch": 1.2,
      "learning_rate": 5e-05,
      "loss": 0.6202,
      "step": 3550
    },
    {
      "epoch": 1.2,
      "learning_rate": 5e-05,
      "loss": 0.6158,
      "step": 3560
    },
    {
      "epoch": 1.2,
      "learning_rate": 5e-05,
      "loss": 0.6089,
      "step": 3570
    },
    {
      "epoch": 1.21,
      "learning_rate": 5e-05,
      "loss": 0.6229,
      "step": 3580
    },
    {
      "epoch": 1.21,
      "learning_rate": 5e-05,
      "loss": 0.6081,
      "step": 3590
    },
    {
      "epoch": 1.21,
      "learning_rate": 5e-05,
      "loss": 0.6082,
      "step": 3600
    },
    {
      "epoch": 1.22,
      "learning_rate": 5e-05,
      "loss": 0.6251,
      "step": 3610
    },
    {
      "epoch": 1.22,
      "learning_rate": 5e-05,
      "loss": 0.639,
      "step": 3620
    },
    {
      "epoch": 1.22,
      "learning_rate": 5e-05,
      "loss": 0.6265,
      "step": 3630
    },
    {
      "epoch": 1.23,
      "learning_rate": 5e-05,
      "loss": 0.6103,
      "step": 3640
    },
    {
      "epoch": 1.23,
      "learning_rate": 5e-05,
      "loss": 0.6247,
      "step": 3650
    },
    {
      "epoch": 1.23,
      "learning_rate": 5e-05,
      "loss": 0.6201,
      "step": 3660
    },
    {
      "epoch": 1.24,
      "learning_rate": 5e-05,
      "loss": 0.6212,
      "step": 3670
    },
    {
      "epoch": 1.24,
      "learning_rate": 5e-05,
      "loss": 0.6197,
      "step": 3680
    },
    {
      "epoch": 1.24,
      "learning_rate": 5e-05,
      "loss": 0.6094,
      "step": 3690
    },
    {
      "epoch": 1.25,
      "learning_rate": 5e-05,
      "loss": 0.6228,
      "step": 3700
    },
    {
      "epoch": 1.25,
      "learning_rate": 5e-05,
      "loss": 0.6119,
      "step": 3710
    },
    {
      "epoch": 1.25,
      "learning_rate": 5e-05,
      "loss": 0.6138,
      "step": 3720
    },
    {
      "epoch": 1.26,
      "learning_rate": 5e-05,
      "loss": 0.6274,
      "step": 3730
    },
    {
      "epoch": 1.26,
      "learning_rate": 5e-05,
      "loss": 0.627,
      "step": 3740
    },
    {
      "epoch": 1.27,
      "learning_rate": 5e-05,
      "loss": 0.6088,
      "step": 3750
    },
    {
      "epoch": 1.27,
      "learning_rate": 5e-05,
      "loss": 0.6144,
      "step": 3760
    },
    {
      "epoch": 1.27,
      "learning_rate": 5e-05,
      "loss": 0.6176,
      "step": 3770
    },
    {
      "epoch": 1.28,
      "learning_rate": 5e-05,
      "loss": 0.6195,
      "step": 3780
    },
    {
      "epoch": 1.28,
      "learning_rate": 5e-05,
      "loss": 0.6148,
      "step": 3790
    },
    {
      "epoch": 1.28,
      "learning_rate": 5e-05,
      "loss": 0.6119,
      "step": 3800
    },
    {
      "epoch": 1.29,
      "learning_rate": 5e-05,
      "loss": 0.6155,
      "step": 3810
    },
    {
      "epoch": 1.29,
      "learning_rate": 5e-05,
      "loss": 0.6111,
      "step": 3820
    },
    {
      "epoch": 1.29,
      "learning_rate": 5e-05,
      "loss": 0.6153,
      "step": 3830
    },
    {
      "epoch": 1.3,
      "learning_rate": 5e-05,
      "loss": 0.6164,
      "step": 3840
    },
    {
      "epoch": 1.3,
      "learning_rate": 5e-05,
      "loss": 0.6109,
      "step": 3850
    },
    {
      "epoch": 1.3,
      "learning_rate": 5e-05,
      "loss": 0.6216,
      "step": 3860
    },
    {
      "epoch": 1.31,
      "learning_rate": 5e-05,
      "loss": 0.6187,
      "step": 3870
    },
    {
      "epoch": 1.31,
      "learning_rate": 5e-05,
      "loss": 0.6049,
      "step": 3880
    },
    {
      "epoch": 1.31,
      "learning_rate": 5e-05,
      "loss": 0.6148,
      "step": 3890
    },
    {
      "epoch": 1.32,
      "learning_rate": 5e-05,
      "loss": 0.6193,
      "step": 3900
    },
    {
      "epoch": 1.32,
      "learning_rate": 5e-05,
      "loss": 0.6157,
      "step": 3910
    },
    {
      "epoch": 1.32,
      "learning_rate": 5e-05,
      "loss": 0.6135,
      "step": 3920
    },
    {
      "epoch": 1.33,
      "learning_rate": 5e-05,
      "loss": 0.6104,
      "step": 3930
    },
    {
      "epoch": 1.33,
      "learning_rate": 5e-05,
      "loss": 0.6178,
      "step": 3940
    },
    {
      "epoch": 1.33,
      "learning_rate": 5e-05,
      "loss": 0.6037,
      "step": 3950
    },
    {
      "epoch": 1.34,
      "learning_rate": 5e-05,
      "loss": 0.6222,
      "step": 3960
    },
    {
      "epoch": 1.34,
      "learning_rate": 5e-05,
      "loss": 0.6119,
      "step": 3970
    },
    {
      "epoch": 1.34,
      "learning_rate": 5e-05,
      "loss": 0.6071,
      "step": 3980
    },
    {
      "epoch": 1.35,
      "learning_rate": 5e-05,
      "loss": 0.6017,
      "step": 3990
    },
    {
      "epoch": 1.35,
      "learning_rate": 5e-05,
      "loss": 0.607,
      "step": 4000
    },
    {
      "epoch": 1.35,
      "learning_rate": 5e-05,
      "loss": 0.6129,
      "step": 4010
    },
    {
      "epoch": 1.36,
      "learning_rate": 5e-05,
      "loss": 0.6061,
      "step": 4020
    },
    {
      "epoch": 1.36,
      "learning_rate": 5e-05,
      "loss": 0.6121,
      "step": 4030
    },
    {
      "epoch": 1.36,
      "learning_rate": 5e-05,
      "loss": 0.6121,
      "step": 4040
    },
    {
      "epoch": 1.37,
      "learning_rate": 5e-05,
      "loss": 0.6084,
      "step": 4050
    },
    {
      "epoch": 1.37,
      "learning_rate": 5e-05,
      "loss": 0.6162,
      "step": 4060
    },
    {
      "epoch": 1.37,
      "learning_rate": 5e-05,
      "loss": 0.6253,
      "step": 4070
    },
    {
      "epoch": 1.38,
      "learning_rate": 5e-05,
      "loss": 0.6176,
      "step": 4080
    },
    {
      "epoch": 1.38,
      "learning_rate": 5e-05,
      "loss": 0.6043,
      "step": 4090
    },
    {
      "epoch": 1.38,
      "learning_rate": 5e-05,
      "loss": 0.6072,
      "step": 4100
    },
    {
      "epoch": 1.39,
      "learning_rate": 5e-05,
      "loss": 0.6086,
      "step": 4110
    },
    {
      "epoch": 1.39,
      "learning_rate": 5e-05,
      "loss": 0.6114,
      "step": 4120
    },
    {
      "epoch": 1.39,
      "learning_rate": 5e-05,
      "loss": 0.6048,
      "step": 4130
    },
    {
      "epoch": 1.4,
      "learning_rate": 5e-05,
      "loss": 0.6114,
      "step": 4140
    },
    {
      "epoch": 1.4,
      "learning_rate": 5e-05,
      "loss": 0.6214,
      "step": 4150
    },
    {
      "epoch": 1.4,
      "learning_rate": 5e-05,
      "loss": 0.6138,
      "step": 4160
    },
    {
      "epoch": 1.41,
      "learning_rate": 5e-05,
      "loss": 0.6158,
      "step": 4170
    },
    {
      "epoch": 1.41,
      "learning_rate": 5e-05,
      "loss": 0.6121,
      "step": 4180
    },
    {
      "epoch": 1.41,
      "learning_rate": 5e-05,
      "loss": 0.6064,
      "step": 4190
    },
    {
      "epoch": 1.42,
      "learning_rate": 5e-05,
      "loss": 0.6073,
      "step": 4200
    },
    {
      "epoch": 1.42,
      "learning_rate": 5e-05,
      "loss": 0.6123,
      "step": 4210
    },
    {
      "epoch": 1.42,
      "learning_rate": 5e-05,
      "loss": 0.6159,
      "step": 4220
    },
    {
      "epoch": 1.43,
      "learning_rate": 5e-05,
      "loss": 0.6219,
      "step": 4230
    },
    {
      "epoch": 1.43,
      "learning_rate": 5e-05,
      "loss": 0.6187,
      "step": 4240
    },
    {
      "epoch": 1.43,
      "learning_rate": 5e-05,
      "loss": 0.6071,
      "step": 4250
    },
    {
      "epoch": 1.44,
      "learning_rate": 5e-05,
      "loss": 0.6095,
      "step": 4260
    },
    {
      "epoch": 1.44,
      "learning_rate": 5e-05,
      "loss": 0.6188,
      "step": 4270
    },
    {
      "epoch": 1.44,
      "learning_rate": 5e-05,
      "loss": 0.6208,
      "step": 4280
    },
    {
      "epoch": 1.45,
      "learning_rate": 5e-05,
      "loss": 0.623,
      "step": 4290
    },
    {
      "epoch": 1.45,
      "learning_rate": 5e-05,
      "loss": 0.6115,
      "step": 4300
    },
    {
      "epoch": 1.45,
      "learning_rate": 5e-05,
      "loss": 0.6085,
      "step": 4310
    },
    {
      "epoch": 1.46,
      "learning_rate": 5e-05,
      "loss": 0.6012,
      "step": 4320
    },
    {
      "epoch": 1.46,
      "learning_rate": 5e-05,
      "loss": 0.6042,
      "step": 4330
    },
    {
      "epoch": 1.46,
      "learning_rate": 5e-05,
      "loss": 0.607,
      "step": 4340
    },
    {
      "epoch": 1.47,
      "learning_rate": 5e-05,
      "loss": 0.6148,
      "step": 4350
    },
    {
      "epoch": 1.47,
      "learning_rate": 5e-05,
      "loss": 0.6104,
      "step": 4360
    },
    {
      "epoch": 1.47,
      "learning_rate": 5e-05,
      "loss": 0.6091,
      "step": 4370
    },
    {
      "epoch": 1.48,
      "learning_rate": 5e-05,
      "loss": 0.6116,
      "step": 4380
    },
    {
      "epoch": 1.48,
      "learning_rate": 5e-05,
      "loss": 0.6108,
      "step": 4390
    },
    {
      "epoch": 1.48,
      "learning_rate": 5e-05,
      "loss": 0.6175,
      "step": 4400
    },
    {
      "epoch": 1.49,
      "learning_rate": 5e-05,
      "loss": 0.6114,
      "step": 4410
    },
    {
      "epoch": 1.49,
      "learning_rate": 5e-05,
      "loss": 0.6151,
      "step": 4420
    },
    {
      "epoch": 1.49,
      "learning_rate": 5e-05,
      "loss": 0.6086,
      "step": 4430
    },
    {
      "epoch": 1.5,
      "learning_rate": 5e-05,
      "loss": 0.6084,
      "step": 4440
    },
    {
      "epoch": 1.5,
      "learning_rate": 5e-05,
      "loss": 0.6051,
      "step": 4450
    },
    {
      "epoch": 1.5,
      "learning_rate": 5e-05,
      "loss": 0.6152,
      "step": 4460
    },
    {
      "epoch": 1.51,
      "learning_rate": 5e-05,
      "loss": 0.6043,
      "step": 4470
    },
    {
      "epoch": 1.51,
      "learning_rate": 5e-05,
      "loss": 0.617,
      "step": 4480
    },
    {
      "epoch": 1.51,
      "learning_rate": 5e-05,
      "loss": 0.6151,
      "step": 4490
    },
    {
      "epoch": 1.52,
      "learning_rate": 5e-05,
      "loss": 0.6146,
      "step": 4500
    },
    {
      "epoch": 1.52,
      "learning_rate": 5e-05,
      "loss": 0.6107,
      "step": 4510
    },
    {
      "epoch": 1.52,
      "learning_rate": 5e-05,
      "loss": 0.6107,
      "step": 4520
    },
    {
      "epoch": 1.53,
      "learning_rate": 5e-05,
      "loss": 0.6219,
      "step": 4530
    },
    {
      "epoch": 1.53,
      "learning_rate": 5e-05,
      "loss": 0.613,
      "step": 4540
    },
    {
      "epoch": 1.53,
      "learning_rate": 5e-05,
      "loss": 0.604,
      "step": 4550
    },
    {
      "epoch": 1.54,
      "learning_rate": 5e-05,
      "loss": 0.5997,
      "step": 4560
    },
    {
      "epoch": 1.54,
      "learning_rate": 5e-05,
      "loss": 0.6056,
      "step": 4570
    },
    {
      "epoch": 1.55,
      "learning_rate": 5e-05,
      "loss": 0.6211,
      "step": 4580
    },
    {
      "epoch": 1.55,
      "learning_rate": 5e-05,
      "loss": 0.6064,
      "step": 4590
    },
    {
      "epoch": 1.55,
      "learning_rate": 5e-05,
      "loss": 0.6124,
      "step": 4600
    },
    {
      "epoch": 1.56,
      "learning_rate": 5e-05,
      "loss": 0.6097,
      "step": 4610
    },
    {
      "epoch": 1.56,
      "learning_rate": 5e-05,
      "loss": 0.6024,
      "step": 4620
    },
    {
      "epoch": 1.56,
      "learning_rate": 5e-05,
      "loss": 0.6177,
      "step": 4630
    },
    {
      "epoch": 1.57,
      "learning_rate": 5e-05,
      "loss": 0.6126,
      "step": 4640
    },
    {
      "epoch": 1.57,
      "learning_rate": 5e-05,
      "loss": 0.603,
      "step": 4650
    },
    {
      "epoch": 1.57,
      "learning_rate": 5e-05,
      "loss": 0.606,
      "step": 4660
    },
    {
      "epoch": 1.58,
      "learning_rate": 5e-05,
      "loss": 0.6219,
      "step": 4670
    },
    {
      "epoch": 1.58,
      "learning_rate": 5e-05,
      "loss": 0.6075,
      "step": 4680
    },
    {
      "epoch": 1.58,
      "learning_rate": 5e-05,
      "loss": 0.6016,
      "step": 4690
    },
    {
      "epoch": 1.59,
      "learning_rate": 5e-05,
      "loss": 0.6083,
      "step": 4700
    },
    {
      "epoch": 1.59,
      "learning_rate": 5e-05,
      "loss": 0.6103,
      "step": 4710
    },
    {
      "epoch": 1.59,
      "learning_rate": 5e-05,
      "loss": 0.6128,
      "step": 4720
    },
    {
      "epoch": 1.6,
      "learning_rate": 5e-05,
      "loss": 0.6101,
      "step": 4730
    },
    {
      "epoch": 1.6,
      "learning_rate": 5e-05,
      "loss": 0.5993,
      "step": 4740
    },
    {
      "epoch": 1.6,
      "learning_rate": 5e-05,
      "loss": 0.6075,
      "step": 4750
    },
    {
      "epoch": 1.61,
      "learning_rate": 5e-05,
      "loss": 0.6173,
      "step": 4760
    },
    {
      "epoch": 1.61,
      "learning_rate": 5e-05,
      "loss": 0.6072,
      "step": 4770
    },
    {
      "epoch": 1.61,
      "learning_rate": 5e-05,
      "loss": 0.628,
      "step": 4780
    },
    {
      "epoch": 1.62,
      "learning_rate": 5e-05,
      "loss": 0.604,
      "step": 4790
    },
    {
      "epoch": 1.62,
      "learning_rate": 5e-05,
      "loss": 0.6024,
      "step": 4800
    },
    {
      "epoch": 1.62,
      "learning_rate": 5e-05,
      "loss": 0.6063,
      "step": 4810
    },
    {
      "epoch": 1.63,
      "learning_rate": 5e-05,
      "loss": 0.5967,
      "step": 4820
    },
    {
      "epoch": 1.63,
      "learning_rate": 5e-05,
      "loss": 0.6096,
      "step": 4830
    },
    {
      "epoch": 1.63,
      "learning_rate": 5e-05,
      "loss": 0.5992,
      "step": 4840
    },
    {
      "epoch": 1.64,
      "learning_rate": 5e-05,
      "loss": 0.6027,
      "step": 4850
    },
    {
      "epoch": 1.64,
      "learning_rate": 5e-05,
      "loss": 0.6051,
      "step": 4860
    },
    {
      "epoch": 1.64,
      "learning_rate": 5e-05,
      "loss": 0.6025,
      "step": 4870
    },
    {
      "epoch": 1.65,
      "learning_rate": 5e-05,
      "loss": 0.5954,
      "step": 4880
    },
    {
      "epoch": 1.65,
      "learning_rate": 5e-05,
      "loss": 0.6022,
      "step": 4890
    },
    {
      "epoch": 1.65,
      "learning_rate": 5e-05,
      "loss": 0.6103,
      "step": 4900
    },
    {
      "epoch": 1.66,
      "learning_rate": 5e-05,
      "loss": 0.5979,
      "step": 4910
    },
    {
      "epoch": 1.66,
      "learning_rate": 5e-05,
      "loss": 0.6065,
      "step": 4920
    },
    {
      "epoch": 1.66,
      "learning_rate": 5e-05,
      "loss": 0.6061,
      "step": 4930
    },
    {
      "epoch": 1.67,
      "learning_rate": 5e-05,
      "loss": 0.6208,
      "step": 4940
    },
    {
      "epoch": 1.67,
      "learning_rate": 5e-05,
      "loss": 0.6096,
      "step": 4950
    },
    {
      "epoch": 1.67,
      "learning_rate": 5e-05,
      "loss": 0.606,
      "step": 4960
    },
    {
      "epoch": 1.68,
      "learning_rate": 5e-05,
      "loss": 0.6032,
      "step": 4970
    },
    {
      "epoch": 1.68,
      "learning_rate": 5e-05,
      "loss": 0.6084,
      "step": 4980
    },
    {
      "epoch": 1.68,
      "learning_rate": 5e-05,
      "loss": 0.61,
      "step": 4990
    },
    {
      "epoch": 1.69,
      "learning_rate": 5e-05,
      "loss": 0.605,
      "step": 5000
    },
    {
      "epoch": 1.69,
      "learning_rate": 5e-05,
      "loss": 0.5961,
      "step": 5010
    },
    {
      "epoch": 1.69,
      "learning_rate": 5e-05,
      "loss": 0.5957,
      "step": 5020
    },
    {
      "epoch": 1.7,
      "learning_rate": 5e-05,
      "loss": 0.6073,
      "step": 5030
    },
    {
      "epoch": 1.7,
      "learning_rate": 5e-05,
      "loss": 0.6069,
      "step": 5040
    },
    {
      "epoch": 1.7,
      "learning_rate": 5e-05,
      "loss": 0.5924,
      "step": 5050
    },
    {
      "epoch": 1.71,
      "learning_rate": 5e-05,
      "loss": 0.6059,
      "step": 5060
    },
    {
      "epoch": 1.71,
      "learning_rate": 5e-05,
      "loss": 0.5999,
      "step": 5070
    },
    {
      "epoch": 1.71,
      "learning_rate": 5e-05,
      "loss": 0.5981,
      "step": 5080
    },
    {
      "epoch": 1.72,
      "learning_rate": 5e-05,
      "loss": 0.5953,
      "step": 5090
    },
    {
      "epoch": 1.72,
      "learning_rate": 5e-05,
      "loss": 0.5974,
      "step": 5100
    },
    {
      "epoch": 1.72,
      "learning_rate": 5e-05,
      "loss": 0.5941,
      "step": 5110
    },
    {
      "epoch": 1.73,
      "learning_rate": 5e-05,
      "loss": 0.6051,
      "step": 5120
    },
    {
      "epoch": 1.73,
      "learning_rate": 5e-05,
      "loss": 0.5919,
      "step": 5130
    },
    {
      "epoch": 1.73,
      "learning_rate": 5e-05,
      "loss": 0.618,
      "step": 5140
    },
    {
      "epoch": 1.74,
      "learning_rate": 5e-05,
      "loss": 0.5972,
      "step": 5150
    },
    {
      "epoch": 1.74,
      "learning_rate": 5e-05,
      "loss": 0.606,
      "step": 5160
    },
    {
      "epoch": 1.74,
      "learning_rate": 5e-05,
      "loss": 0.6066,
      "step": 5170
    },
    {
      "epoch": 1.75,
      "learning_rate": 5e-05,
      "loss": 0.6047,
      "step": 5180
    },
    {
      "epoch": 1.75,
      "learning_rate": 5e-05,
      "loss": 0.6096,
      "step": 5190
    },
    {
      "epoch": 1.75,
      "learning_rate": 5e-05,
      "loss": 0.6077,
      "step": 5200
    },
    {
      "epoch": 1.76,
      "learning_rate": 5e-05,
      "loss": 0.5961,
      "step": 5210
    },
    {
      "epoch": 1.76,
      "learning_rate": 5e-05,
      "loss": 0.593,
      "step": 5220
    },
    {
      "epoch": 1.76,
      "learning_rate": 5e-05,
      "loss": 0.6012,
      "step": 5230
    },
    {
      "epoch": 1.77,
      "learning_rate": 5e-05,
      "loss": 0.6027,
      "step": 5240
    },
    {
      "epoch": 1.77,
      "learning_rate": 5e-05,
      "loss": 0.5976,
      "step": 5250
    },
    {
      "epoch": 1.77,
      "learning_rate": 5e-05,
      "loss": 0.6027,
      "step": 5260
    },
    {
      "epoch": 1.78,
      "learning_rate": 5e-05,
      "loss": 0.5945,
      "step": 5270
    },
    {
      "epoch": 1.78,
      "learning_rate": 5e-05,
      "loss": 0.6092,
      "step": 5280
    },
    {
      "epoch": 1.78,
      "learning_rate": 5e-05,
      "loss": 0.6062,
      "step": 5290
    },
    {
      "epoch": 1.79,
      "learning_rate": 5e-05,
      "loss": 0.6021,
      "step": 5300
    },
    {
      "epoch": 1.79,
      "learning_rate": 5e-05,
      "loss": 0.5988,
      "step": 5310
    },
    {
      "epoch": 1.79,
      "learning_rate": 5e-05,
      "loss": 0.6035,
      "step": 5320
    },
    {
      "epoch": 1.8,
      "learning_rate": 5e-05,
      "loss": 0.6163,
      "step": 5330
    },
    {
      "epoch": 1.8,
      "learning_rate": 5e-05,
      "loss": 0.6015,
      "step": 5340
    },
    {
      "epoch": 1.8,
      "learning_rate": 5e-05,
      "loss": 0.5948,
      "step": 5350
    },
    {
      "epoch": 1.81,
      "learning_rate": 5e-05,
      "loss": 0.5959,
      "step": 5360
    },
    {
      "epoch": 1.81,
      "learning_rate": 5e-05,
      "loss": 0.5871,
      "step": 5370
    },
    {
      "epoch": 1.81,
      "learning_rate": 5e-05,
      "loss": 0.5958,
      "step": 5380
    },
    {
      "epoch": 1.82,
      "learning_rate": 5e-05,
      "loss": 0.5939,
      "step": 5390
    },
    {
      "epoch": 1.82,
      "learning_rate": 5e-05,
      "loss": 0.5985,
      "step": 5400
    },
    {
      "epoch": 1.83,
      "learning_rate": 5e-05,
      "loss": 0.6004,
      "step": 5410
    },
    {
      "epoch": 1.83,
      "learning_rate": 5e-05,
      "loss": 0.5908,
      "step": 5420
    },
    {
      "epoch": 1.83,
      "learning_rate": 5e-05,
      "loss": 0.6019,
      "step": 5430
    },
    {
      "epoch": 1.84,
      "learning_rate": 5e-05,
      "loss": 0.6048,
      "step": 5440
    },
    {
      "epoch": 1.84,
      "learning_rate": 5e-05,
      "loss": 0.602,
      "step": 5450
    },
    {
      "epoch": 1.84,
      "learning_rate": 5e-05,
      "loss": 0.6098,
      "step": 5460
    },
    {
      "epoch": 1.85,
      "learning_rate": 5e-05,
      "loss": 0.6045,
      "step": 5470
    },
    {
      "epoch": 1.85,
      "learning_rate": 5e-05,
      "loss": 0.5921,
      "step": 5480
    },
    {
      "epoch": 1.85,
      "learning_rate": 5e-05,
      "loss": 0.5962,
      "step": 5490
    },
    {
      "epoch": 1.86,
      "learning_rate": 5e-05,
      "loss": 0.5969,
      "step": 5500
    },
    {
      "epoch": 1.86,
      "learning_rate": 5e-05,
      "loss": 0.5975,
      "step": 5510
    },
    {
      "epoch": 1.86,
      "learning_rate": 5e-05,
      "loss": 0.5998,
      "step": 5520
    },
    {
      "epoch": 1.87,
      "learning_rate": 5e-05,
      "loss": 0.6049,
      "step": 5530
    },
    {
      "epoch": 1.87,
      "learning_rate": 5e-05,
      "loss": 0.5984,
      "step": 5540
    },
    {
      "epoch": 1.87,
      "learning_rate": 5e-05,
      "loss": 0.599,
      "step": 5550
    },
    {
      "epoch": 1.88,
      "learning_rate": 5e-05,
      "loss": 0.6075,
      "step": 5560
    },
    {
      "epoch": 1.88,
      "learning_rate": 5e-05,
      "loss": 0.6019,
      "step": 5570
    },
    {
      "epoch": 1.88,
      "learning_rate": 5e-05,
      "loss": 0.601,
      "step": 5580
    },
    {
      "epoch": 1.89,
      "learning_rate": 5e-05,
      "loss": 0.592,
      "step": 5590
    },
    {
      "epoch": 1.89,
      "learning_rate": 5e-05,
      "loss": 0.5925,
      "step": 5600
    },
    {
      "epoch": 1.89,
      "learning_rate": 5e-05,
      "loss": 0.6064,
      "step": 5610
    },
    {
      "epoch": 1.9,
      "learning_rate": 5e-05,
      "loss": 0.6089,
      "step": 5620
    },
    {
      "epoch": 1.9,
      "learning_rate": 5e-05,
      "loss": 0.5952,
      "step": 5630
    },
    {
      "epoch": 1.9,
      "learning_rate": 5e-05,
      "loss": 0.5954,
      "step": 5640
    },
    {
      "epoch": 1.91,
      "learning_rate": 5e-05,
      "loss": 0.5987,
      "step": 5650
    },
    {
      "epoch": 1.91,
      "learning_rate": 5e-05,
      "loss": 0.6125,
      "step": 5660
    },
    {
      "epoch": 1.91,
      "learning_rate": 5e-05,
      "loss": 0.5889,
      "step": 5670
    },
    {
      "epoch": 1.92,
      "learning_rate": 5e-05,
      "loss": 0.5944,
      "step": 5680
    },
    {
      "epoch": 1.92,
      "learning_rate": 5e-05,
      "loss": 0.5984,
      "step": 5690
    },
    {
      "epoch": 1.92,
      "learning_rate": 5e-05,
      "loss": 0.5973,
      "step": 5700
    },
    {
      "epoch": 1.93,
      "learning_rate": 5e-05,
      "loss": 0.6004,
      "step": 5710
    },
    {
      "epoch": 1.93,
      "learning_rate": 5e-05,
      "loss": 0.5989,
      "step": 5720
    },
    {
      "epoch": 1.93,
      "learning_rate": 5e-05,
      "loss": 0.5983,
      "step": 5730
    },
    {
      "epoch": 1.94,
      "learning_rate": 5e-05,
      "loss": 0.6079,
      "step": 5740
    },
    {
      "epoch": 1.94,
      "learning_rate": 5e-05,
      "loss": 0.6028,
      "step": 5750
    },
    {
      "epoch": 1.94,
      "learning_rate": 5e-05,
      "loss": 0.5953,
      "step": 5760
    },
    {
      "epoch": 1.95,
      "learning_rate": 5e-05,
      "loss": 0.6108,
      "step": 5770
    },
    {
      "epoch": 1.95,
      "learning_rate": 5e-05,
      "loss": 0.6158,
      "step": 5780
    },
    {
      "epoch": 1.95,
      "learning_rate": 5e-05,
      "loss": 0.5943,
      "step": 5790
    },
    {
      "epoch": 1.96,
      "learning_rate": 5e-05,
      "loss": 0.5845,
      "step": 5800
    },
    {
      "epoch": 1.96,
      "learning_rate": 5e-05,
      "loss": 0.5952,
      "step": 5810
    },
    {
      "epoch": 1.96,
      "learning_rate": 5e-05,
      "loss": 0.6038,
      "step": 5820
    },
    {
      "epoch": 1.97,
      "learning_rate": 5e-05,
      "loss": 0.5987,
      "step": 5830
    },
    {
      "epoch": 1.97,
      "learning_rate": 5e-05,
      "loss": 0.5895,
      "step": 5840
    },
    {
      "epoch": 1.97,
      "learning_rate": 5e-05,
      "loss": 0.6007,
      "step": 5850
    },
    {
      "epoch": 1.98,
      "learning_rate": 5e-05,
      "loss": 0.5979,
      "step": 5860
    },
    {
      "epoch": 1.98,
      "learning_rate": 5e-05,
      "loss": 0.6054,
      "step": 5870
    },
    {
      "epoch": 1.98,
      "learning_rate": 5e-05,
      "loss": 0.5896,
      "step": 5880
    },
    {
      "epoch": 1.99,
      "learning_rate": 5e-05,
      "loss": 0.598,
      "step": 5890
    },
    {
      "epoch": 1.99,
      "learning_rate": 5e-05,
      "loss": 0.5963,
      "step": 5900
    },
    {
      "epoch": 1.99,
      "learning_rate": 5e-05,
      "loss": 0.5984,
      "step": 5910
    },
    {
      "epoch": 2.0,
      "learning_rate": 5e-05,
      "loss": 0.5944,
      "step": 5920
    },
    {
      "epoch": 2.0,
      "learning_rate": 5e-05,
      "loss": 0.606,
      "step": 5930
    },
    {
      "epoch": 2.0,
      "learning_rate": 5e-05,
      "loss": 0.6001,
      "step": 5940
    },
    {
      "epoch": 2.01,
      "learning_rate": 5e-05,
      "loss": 0.5975,
      "step": 5950
    },
    {
      "epoch": 2.01,
      "learning_rate": 5e-05,
      "loss": 0.5864,
      "step": 5960
    },
    {
      "epoch": 2.01,
      "learning_rate": 5e-05,
      "loss": 0.5899,
      "step": 5970
    },
    {
      "epoch": 2.02,
      "learning_rate": 5e-05,
      "loss": 0.604,
      "step": 5980
    },
    {
      "epoch": 2.02,
      "learning_rate": 5e-05,
      "loss": 0.5957,
      "step": 5990
    },
    {
      "epoch": 2.02,
      "learning_rate": 5e-05,
      "loss": 0.5888,
      "step": 6000
    },
    {
      "epoch": 2.03,
      "learning_rate": 5e-05,
      "loss": 0.5994,
      "step": 6010
    },
    {
      "epoch": 2.03,
      "learning_rate": 5e-05,
      "loss": 0.5914,
      "step": 6020
    },
    {
      "epoch": 2.03,
      "learning_rate": 5e-05,
      "loss": 0.5981,
      "step": 6030
    },
    {
      "epoch": 2.04,
      "learning_rate": 5e-05,
      "loss": 0.5807,
      "step": 6040
    },
    {
      "epoch": 2.04,
      "learning_rate": 5e-05,
      "loss": 0.594,
      "step": 6050
    },
    {
      "epoch": 2.04,
      "learning_rate": 5e-05,
      "loss": 0.5983,
      "step": 6060
    },
    {
      "epoch": 2.05,
      "learning_rate": 5e-05,
      "loss": 0.5895,
      "step": 6070
    },
    {
      "epoch": 2.05,
      "learning_rate": 5e-05,
      "loss": 0.5896,
      "step": 6080
    },
    {
      "epoch": 2.05,
      "learning_rate": 5e-05,
      "loss": 0.5949,
      "step": 6090
    },
    {
      "epoch": 2.06,
      "learning_rate": 5e-05,
      "loss": 0.579,
      "step": 6100
    },
    {
      "epoch": 2.06,
      "learning_rate": 5e-05,
      "loss": 0.591,
      "step": 6110
    },
    {
      "epoch": 2.06,
      "learning_rate": 5e-05,
      "loss": 0.5881,
      "step": 6120
    },
    {
      "epoch": 2.07,
      "learning_rate": 5e-05,
      "loss": 0.6055,
      "step": 6130
    },
    {
      "epoch": 2.07,
      "learning_rate": 5e-05,
      "loss": 0.6037,
      "step": 6140
    },
    {
      "epoch": 2.07,
      "learning_rate": 5e-05,
      "loss": 0.5881,
      "step": 6150
    },
    {
      "epoch": 2.08,
      "learning_rate": 5e-05,
      "loss": 0.5937,
      "step": 6160
    },
    {
      "epoch": 2.08,
      "learning_rate": 5e-05,
      "loss": 0.5962,
      "step": 6170
    },
    {
      "epoch": 2.08,
      "learning_rate": 5e-05,
      "loss": 0.5838,
      "step": 6180
    },
    {
      "epoch": 2.09,
      "learning_rate": 5e-05,
      "loss": 0.6035,
      "step": 6190
    },
    {
      "epoch": 2.09,
      "learning_rate": 5e-05,
      "loss": 0.5993,
      "step": 6200
    },
    {
      "epoch": 2.09,
      "learning_rate": 5e-05,
      "loss": 0.6014,
      "step": 6210
    },
    {
      "epoch": 2.1,
      "learning_rate": 5e-05,
      "loss": 0.5968,
      "step": 6220
    },
    {
      "epoch": 2.1,
      "learning_rate": 5e-05,
      "loss": 0.5967,
      "step": 6230
    },
    {
      "epoch": 2.11,
      "learning_rate": 5e-05,
      "loss": 0.6003,
      "step": 6240
    },
    {
      "epoch": 2.11,
      "learning_rate": 5e-05,
      "loss": 0.5939,
      "step": 6250
    },
    {
      "epoch": 2.11,
      "learning_rate": 5e-05,
      "loss": 0.5855,
      "step": 6260
    },
    {
      "epoch": 2.12,
      "learning_rate": 5e-05,
      "loss": 0.5876,
      "step": 6270
    },
    {
      "epoch": 2.12,
      "learning_rate": 5e-05,
      "loss": 0.5898,
      "step": 6280
    },
    {
      "epoch": 2.12,
      "learning_rate": 5e-05,
      "loss": 0.5845,
      "step": 6290
    },
    {
      "epoch": 2.13,
      "learning_rate": 5e-05,
      "loss": 0.5929,
      "step": 6300
    },
    {
      "epoch": 2.13,
      "learning_rate": 5e-05,
      "loss": 0.5898,
      "step": 6310
    },
    {
      "epoch": 2.13,
      "learning_rate": 5e-05,
      "loss": 0.5938,
      "step": 6320
    },
    {
      "epoch": 2.14,
      "learning_rate": 5e-05,
      "loss": 0.5935,
      "step": 6330
    },
    {
      "epoch": 2.14,
      "learning_rate": 5e-05,
      "loss": 0.5904,
      "step": 6340
    },
    {
      "epoch": 2.14,
      "learning_rate": 5e-05,
      "loss": 0.6015,
      "step": 6350
    },
    {
      "epoch": 2.15,
      "learning_rate": 5e-05,
      "loss": 0.59,
      "step": 6360
    },
    {
      "epoch": 2.15,
      "learning_rate": 5e-05,
      "loss": 0.5894,
      "step": 6370
    },
    {
      "epoch": 2.15,
      "learning_rate": 5e-05,
      "loss": 0.6066,
      "step": 6380
    },
    {
      "epoch": 2.16,
      "learning_rate": 5e-05,
      "loss": 0.5921,
      "step": 6390
    },
    {
      "epoch": 2.16,
      "learning_rate": 5e-05,
      "loss": 0.5967,
      "step": 6400
    },
    {
      "epoch": 2.16,
      "learning_rate": 5e-05,
      "loss": 0.5919,
      "step": 6410
    },
    {
      "epoch": 2.17,
      "learning_rate": 5e-05,
      "loss": 0.5898,
      "step": 6420
    },
    {
      "epoch": 2.17,
      "learning_rate": 5e-05,
      "loss": 0.5948,
      "step": 6430
    },
    {
      "epoch": 2.17,
      "learning_rate": 5e-05,
      "loss": 0.5858,
      "step": 6440
    },
    {
      "epoch": 2.18,
      "learning_rate": 5e-05,
      "loss": 0.5927,
      "step": 6450
    },
    {
      "epoch": 2.18,
      "learning_rate": 5e-05,
      "loss": 0.5877,
      "step": 6460
    },
    {
      "epoch": 2.18,
      "learning_rate": 5e-05,
      "loss": 0.5874,
      "step": 6470
    },
    {
      "epoch": 2.19,
      "learning_rate": 5e-05,
      "loss": 0.5934,
      "step": 6480
    },
    {
      "epoch": 2.19,
      "learning_rate": 5e-05,
      "loss": 0.6038,
      "step": 6490
    },
    {
      "epoch": 2.19,
      "learning_rate": 5e-05,
      "loss": 0.5994,
      "step": 6500
    },
    {
      "epoch": 2.2,
      "learning_rate": 5e-05,
      "loss": 0.5986,
      "step": 6510
    },
    {
      "epoch": 2.2,
      "learning_rate": 5e-05,
      "loss": 0.5863,
      "step": 6520
    },
    {
      "epoch": 2.2,
      "learning_rate": 5e-05,
      "loss": 0.5844,
      "step": 6530
    },
    {
      "epoch": 2.21,
      "learning_rate": 5e-05,
      "loss": 0.5891,
      "step": 6540
    },
    {
      "epoch": 2.21,
      "learning_rate": 5e-05,
      "loss": 0.5848,
      "step": 6550
    },
    {
      "epoch": 2.21,
      "learning_rate": 5e-05,
      "loss": 0.6097,
      "step": 6560
    },
    {
      "epoch": 2.22,
      "learning_rate": 5e-05,
      "loss": 0.6017,
      "step": 6570
    },
    {
      "epoch": 2.22,
      "learning_rate": 5e-05,
      "loss": 0.5913,
      "step": 6580
    },
    {
      "epoch": 2.22,
      "learning_rate": 5e-05,
      "loss": 0.5933,
      "step": 6590
    },
    {
      "epoch": 2.23,
      "learning_rate": 5e-05,
      "loss": 0.5906,
      "step": 6600
    },
    {
      "epoch": 2.23,
      "learning_rate": 5e-05,
      "loss": 0.5811,
      "step": 6610
    },
    {
      "epoch": 2.23,
      "learning_rate": 5e-05,
      "loss": 0.5866,
      "step": 6620
    },
    {
      "epoch": 2.24,
      "learning_rate": 5e-05,
      "loss": 0.5979,
      "step": 6630
    },
    {
      "epoch": 2.24,
      "learning_rate": 5e-05,
      "loss": 0.5871,
      "step": 6640
    },
    {
      "epoch": 2.24,
      "learning_rate": 5e-05,
      "loss": 0.5849,
      "step": 6650
    },
    {
      "epoch": 2.25,
      "learning_rate": 5e-05,
      "loss": 0.5923,
      "step": 6660
    },
    {
      "epoch": 2.25,
      "learning_rate": 5e-05,
      "loss": 0.59,
      "step": 6670
    },
    {
      "epoch": 2.25,
      "learning_rate": 5e-05,
      "loss": 0.5842,
      "step": 6680
    },
    {
      "epoch": 2.26,
      "learning_rate": 5e-05,
      "loss": 0.5902,
      "step": 6690
    },
    {
      "epoch": 2.26,
      "learning_rate": 5e-05,
      "loss": 0.5971,
      "step": 6700
    },
    {
      "epoch": 2.26,
      "learning_rate": 5e-05,
      "loss": 0.5869,
      "step": 6710
    },
    {
      "epoch": 2.27,
      "learning_rate": 5e-05,
      "loss": 0.5885,
      "step": 6720
    },
    {
      "epoch": 2.27,
      "learning_rate": 5e-05,
      "loss": 0.5924,
      "step": 6730
    },
    {
      "epoch": 2.27,
      "learning_rate": 5e-05,
      "loss": 0.5837,
      "step": 6740
    },
    {
      "epoch": 2.28,
      "learning_rate": 5e-05,
      "loss": 0.588,
      "step": 6750
    },
    {
      "epoch": 2.28,
      "learning_rate": 5e-05,
      "loss": 0.5903,
      "step": 6760
    },
    {
      "epoch": 2.28,
      "learning_rate": 5e-05,
      "loss": 0.5896,
      "step": 6770
    },
    {
      "epoch": 2.29,
      "learning_rate": 5e-05,
      "loss": 0.5885,
      "step": 6780
    },
    {
      "epoch": 2.29,
      "learning_rate": 5e-05,
      "loss": 0.588,
      "step": 6790
    },
    {
      "epoch": 2.29,
      "learning_rate": 5e-05,
      "loss": 0.5856,
      "step": 6800
    },
    {
      "epoch": 2.3,
      "learning_rate": 5e-05,
      "loss": 0.5869,
      "step": 6810
    },
    {
      "epoch": 2.3,
      "learning_rate": 5e-05,
      "loss": 0.5811,
      "step": 6820
    },
    {
      "epoch": 2.3,
      "learning_rate": 5e-05,
      "loss": 0.5904,
      "step": 6830
    },
    {
      "epoch": 2.31,
      "learning_rate": 5e-05,
      "loss": 0.5817,
      "step": 6840
    },
    {
      "epoch": 2.31,
      "learning_rate": 5e-05,
      "loss": 0.5856,
      "step": 6850
    },
    {
      "epoch": 2.31,
      "learning_rate": 5e-05,
      "loss": 0.5847,
      "step": 6860
    },
    {
      "epoch": 2.32,
      "learning_rate": 5e-05,
      "loss": 0.5869,
      "step": 6870
    },
    {
      "epoch": 2.32,
      "learning_rate": 5e-05,
      "loss": 0.5886,
      "step": 6880
    },
    {
      "epoch": 2.32,
      "learning_rate": 5e-05,
      "loss": 0.5828,
      "step": 6890
    },
    {
      "epoch": 2.33,
      "learning_rate": 5e-05,
      "loss": 0.5906,
      "step": 6900
    },
    {
      "epoch": 2.33,
      "learning_rate": 5e-05,
      "loss": 0.5843,
      "step": 6910
    },
    {
      "epoch": 2.33,
      "learning_rate": 5e-05,
      "loss": 0.5907,
      "step": 6920
    },
    {
      "epoch": 2.34,
      "learning_rate": 5e-05,
      "loss": 0.5802,
      "step": 6930
    },
    {
      "epoch": 2.34,
      "learning_rate": 5e-05,
      "loss": 0.5883,
      "step": 6940
    },
    {
      "epoch": 2.34,
      "learning_rate": 5e-05,
      "loss": 0.5842,
      "step": 6950
    },
    {
      "epoch": 2.35,
      "learning_rate": 5e-05,
      "loss": 0.5908,
      "step": 6960
    },
    {
      "epoch": 2.35,
      "learning_rate": 5e-05,
      "loss": 0.593,
      "step": 6970
    },
    {
      "epoch": 2.35,
      "learning_rate": 5e-05,
      "loss": 0.5913,
      "step": 6980
    },
    {
      "epoch": 2.36,
      "learning_rate": 5e-05,
      "loss": 0.5846,
      "step": 6990
    },
    {
      "epoch": 2.36,
      "learning_rate": 5e-05,
      "loss": 0.5924,
      "step": 7000
    },
    {
      "epoch": 2.36,
      "learning_rate": 5e-05,
      "loss": 0.5959,
      "step": 7010
    },
    {
      "epoch": 2.37,
      "learning_rate": 5e-05,
      "loss": 0.5869,
      "step": 7020
    },
    {
      "epoch": 2.37,
      "learning_rate": 5e-05,
      "loss": 0.581,
      "step": 7030
    },
    {
      "epoch": 2.37,
      "learning_rate": 5e-05,
      "loss": 0.5893,
      "step": 7040
    },
    {
      "epoch": 2.38,
      "learning_rate": 5e-05,
      "loss": 0.5792,
      "step": 7050
    },
    {
      "epoch": 2.38,
      "learning_rate": 5e-05,
      "loss": 0.6,
      "step": 7060
    },
    {
      "epoch": 2.39,
      "learning_rate": 5e-05,
      "loss": 0.5856,
      "step": 7070
    },
    {
      "epoch": 2.39,
      "learning_rate": 5e-05,
      "loss": 0.579,
      "step": 7080
    },
    {
      "epoch": 2.39,
      "learning_rate": 5e-05,
      "loss": 0.5834,
      "step": 7090
    },
    {
      "epoch": 2.4,
      "learning_rate": 5e-05,
      "loss": 0.5823,
      "step": 7100
    },
    {
      "epoch": 2.4,
      "learning_rate": 5e-05,
      "loss": 0.5852,
      "step": 7110
    },
    {
      "epoch": 2.4,
      "learning_rate": 5e-05,
      "loss": 0.5856,
      "step": 7120
    },
    {
      "epoch": 2.41,
      "learning_rate": 5e-05,
      "loss": 0.5831,
      "step": 7130
    },
    {
      "epoch": 2.41,
      "learning_rate": 5e-05,
      "loss": 0.581,
      "step": 7140
    },
    {
      "epoch": 2.41,
      "learning_rate": 5e-05,
      "loss": 0.5791,
      "step": 7150
    },
    {
      "epoch": 2.42,
      "learning_rate": 5e-05,
      "loss": 0.5779,
      "step": 7160
    },
    {
      "epoch": 2.42,
      "learning_rate": 5e-05,
      "loss": 0.5843,
      "step": 7170
    },
    {
      "epoch": 2.42,
      "learning_rate": 5e-05,
      "loss": 0.5849,
      "step": 7180
    },
    {
      "epoch": 2.43,
      "learning_rate": 5e-05,
      "loss": 0.588,
      "step": 7190
    },
    {
      "epoch": 2.43,
      "learning_rate": 5e-05,
      "loss": 0.5856,
      "step": 7200
    },
    {
      "epoch": 2.43,
      "learning_rate": 5e-05,
      "loss": 0.5785,
      "step": 7210
    },
    {
      "epoch": 2.44,
      "learning_rate": 5e-05,
      "loss": 0.59,
      "step": 7220
    },
    {
      "epoch": 2.44,
      "learning_rate": 5e-05,
      "loss": 0.5897,
      "step": 7230
    },
    {
      "epoch": 2.44,
      "learning_rate": 5e-05,
      "loss": 0.5827,
      "step": 7240
    },
    {
      "epoch": 2.45,
      "learning_rate": 5e-05,
      "loss": 0.5874,
      "step": 7250
    },
    {
      "epoch": 2.45,
      "learning_rate": 5e-05,
      "loss": 0.5853,
      "step": 7260
    },
    {
      "epoch": 2.45,
      "learning_rate": 5e-05,
      "loss": 0.5849,
      "step": 7270
    },
    {
      "epoch": 2.46,
      "learning_rate": 5e-05,
      "loss": 0.5926,
      "step": 7280
    },
    {
      "epoch": 2.46,
      "learning_rate": 5e-05,
      "loss": 0.586,
      "step": 7290
    },
    {
      "epoch": 2.46,
      "learning_rate": 5e-05,
      "loss": 0.5897,
      "step": 7300
    },
    {
      "epoch": 2.47,
      "learning_rate": 5e-05,
      "loss": 0.5869,
      "step": 7310
    },
    {
      "epoch": 2.47,
      "learning_rate": 5e-05,
      "loss": 0.5805,
      "step": 7320
    },
    {
      "epoch": 2.47,
      "learning_rate": 5e-05,
      "loss": 0.5942,
      "step": 7330
    },
    {
      "epoch": 2.48,
      "learning_rate": 5e-05,
      "loss": 0.5872,
      "step": 7340
    },
    {
      "epoch": 2.48,
      "learning_rate": 5e-05,
      "loss": 0.5785,
      "step": 7350
    },
    {
      "epoch": 2.48,
      "learning_rate": 5e-05,
      "loss": 0.5852,
      "step": 7360
    },
    {
      "epoch": 2.49,
      "learning_rate": 5e-05,
      "loss": 0.5964,
      "step": 7370
    },
    {
      "epoch": 2.49,
      "learning_rate": 5e-05,
      "loss": 0.584,
      "step": 7380
    },
    {
      "epoch": 2.49,
      "learning_rate": 5e-05,
      "loss": 0.5775,
      "step": 7390
    },
    {
      "epoch": 2.5,
      "learning_rate": 5e-05,
      "loss": 0.5909,
      "step": 7400
    },
    {
      "epoch": 2.5,
      "learning_rate": 5e-05,
      "loss": 0.5876,
      "step": 7410
    },
    {
      "epoch": 2.5,
      "learning_rate": 5e-05,
      "loss": 0.585,
      "step": 7420
    },
    {
      "epoch": 2.51,
      "learning_rate": 5e-05,
      "loss": 0.585,
      "step": 7430
    },
    {
      "epoch": 2.51,
      "learning_rate": 5e-05,
      "loss": 0.5876,
      "step": 7440
    },
    {
      "epoch": 2.51,
      "learning_rate": 5e-05,
      "loss": 0.5876,
      "step": 7450
    },
    {
      "epoch": 2.52,
      "learning_rate": 5e-05,
      "loss": 0.5846,
      "step": 7460
    },
    {
      "epoch": 2.52,
      "learning_rate": 5e-05,
      "loss": 0.5909,
      "step": 7470
    },
    {
      "epoch": 2.52,
      "learning_rate": 5e-05,
      "loss": 0.5862,
      "step": 7480
    },
    {
      "epoch": 2.53,
      "learning_rate": 5e-05,
      "loss": 0.575,
      "step": 7490
    },
    {
      "epoch": 2.53,
      "learning_rate": 5e-05,
      "loss": 0.5819,
      "step": 7500
    },
    {
      "epoch": 2.53,
      "learning_rate": 5e-05,
      "loss": 0.5867,
      "step": 7510
    },
    {
      "epoch": 2.54,
      "learning_rate": 5e-05,
      "loss": 0.5963,
      "step": 7520
    },
    {
      "epoch": 2.54,
      "learning_rate": 5e-05,
      "loss": 0.5869,
      "step": 7530
    },
    {
      "epoch": 2.54,
      "learning_rate": 5e-05,
      "loss": 0.5914,
      "step": 7540
    },
    {
      "epoch": 2.55,
      "learning_rate": 5e-05,
      "loss": 0.5847,
      "step": 7550
    },
    {
      "epoch": 2.55,
      "learning_rate": 5e-05,
      "loss": 0.5905,
      "step": 7560
    },
    {
      "epoch": 2.55,
      "learning_rate": 5e-05,
      "loss": 0.5744,
      "step": 7570
    },
    {
      "epoch": 2.56,
      "learning_rate": 5e-05,
      "loss": 0.5771,
      "step": 7580
    },
    {
      "epoch": 2.56,
      "learning_rate": 5e-05,
      "loss": 0.5827,
      "step": 7590
    },
    {
      "epoch": 2.56,
      "learning_rate": 5e-05,
      "loss": 0.5844,
      "step": 7600
    },
    {
      "epoch": 2.57,
      "learning_rate": 5e-05,
      "loss": 0.5807,
      "step": 7610
    },
    {
      "epoch": 2.57,
      "learning_rate": 5e-05,
      "loss": 0.5872,
      "step": 7620
    },
    {
      "epoch": 2.57,
      "learning_rate": 5e-05,
      "loss": 0.5827,
      "step": 7630
    },
    {
      "epoch": 2.58,
      "learning_rate": 5e-05,
      "loss": 0.5827,
      "step": 7640
    },
    {
      "epoch": 2.58,
      "learning_rate": 5e-05,
      "loss": 0.5894,
      "step": 7650
    },
    {
      "epoch": 2.58,
      "learning_rate": 5e-05,
      "loss": 0.5827,
      "step": 7660
    },
    {
      "epoch": 2.59,
      "learning_rate": 5e-05,
      "loss": 0.5815,
      "step": 7670
    },
    {
      "epoch": 2.59,
      "learning_rate": 5e-05,
      "loss": 0.5828,
      "step": 7680
    },
    {
      "epoch": 2.59,
      "learning_rate": 5e-05,
      "loss": 0.5835,
      "step": 7690
    },
    {
      "epoch": 2.6,
      "learning_rate": 5e-05,
      "loss": 0.5798,
      "step": 7700
    },
    {
      "epoch": 2.6,
      "learning_rate": 5e-05,
      "loss": 0.5886,
      "step": 7710
    },
    {
      "epoch": 2.6,
      "learning_rate": 5e-05,
      "loss": 0.5966,
      "step": 7720
    },
    {
      "epoch": 2.61,
      "learning_rate": 5e-05,
      "loss": 0.5878,
      "step": 7730
    },
    {
      "epoch": 2.61,
      "learning_rate": 5e-05,
      "loss": 0.5855,
      "step": 7740
    },
    {
      "epoch": 2.61,
      "learning_rate": 5e-05,
      "loss": 0.5816,
      "step": 7750
    },
    {
      "epoch": 2.62,
      "learning_rate": 5e-05,
      "loss": 0.5856,
      "step": 7760
    },
    {
      "epoch": 2.62,
      "learning_rate": 5e-05,
      "loss": 0.5893,
      "step": 7770
    },
    {
      "epoch": 2.62,
      "learning_rate": 5e-05,
      "loss": 0.5839,
      "step": 7780
    },
    {
      "epoch": 2.63,
      "learning_rate": 5e-05,
      "loss": 0.5791,
      "step": 7790
    },
    {
      "epoch": 2.63,
      "learning_rate": 5e-05,
      "loss": 0.5974,
      "step": 7800
    },
    {
      "epoch": 2.63,
      "learning_rate": 5e-05,
      "loss": 0.5769,
      "step": 7810
    },
    {
      "epoch": 2.64,
      "learning_rate": 5e-05,
      "loss": 0.5816,
      "step": 7820
    },
    {
      "epoch": 2.64,
      "learning_rate": 5e-05,
      "loss": 0.5831,
      "step": 7830
    },
    {
      "epoch": 2.64,
      "learning_rate": 5e-05,
      "loss": 0.5885,
      "step": 7840
    },
    {
      "epoch": 2.65,
      "learning_rate": 5e-05,
      "loss": 0.5866,
      "step": 7850
    },
    {
      "epoch": 2.65,
      "learning_rate": 5e-05,
      "loss": 0.5805,
      "step": 7860
    },
    {
      "epoch": 2.65,
      "learning_rate": 5e-05,
      "loss": 0.5888,
      "step": 7870
    },
    {
      "epoch": 2.66,
      "learning_rate": 5e-05,
      "loss": 0.5713,
      "step": 7880
    },
    {
      "epoch": 2.66,
      "learning_rate": 5e-05,
      "loss": 0.5797,
      "step": 7890
    },
    {
      "epoch": 2.67,
      "learning_rate": 5e-05,
      "loss": 0.5757,
      "step": 7900
    },
    {
      "epoch": 2.67,
      "learning_rate": 5e-05,
      "loss": 0.5954,
      "step": 7910
    },
    {
      "epoch": 2.67,
      "learning_rate": 5e-05,
      "loss": 0.5921,
      "step": 7920
    },
    {
      "epoch": 2.68,
      "learning_rate": 5e-05,
      "loss": 0.5825,
      "step": 7930
    },
    {
      "epoch": 2.68,
      "learning_rate": 5e-05,
      "loss": 0.5749,
      "step": 7940
    },
    {
      "epoch": 2.68,
      "learning_rate": 5e-05,
      "loss": 0.5858,
      "step": 7950
    },
    {
      "epoch": 2.69,
      "learning_rate": 5e-05,
      "loss": 0.5773,
      "step": 7960
    },
    {
      "epoch": 2.69,
      "learning_rate": 5e-05,
      "loss": 0.5913,
      "step": 7970
    },
    {
      "epoch": 2.69,
      "learning_rate": 5e-05,
      "loss": 0.5758,
      "step": 7980
    },
    {
      "epoch": 2.7,
      "learning_rate": 5e-05,
      "loss": 0.5898,
      "step": 7990
    },
    {
      "epoch": 2.7,
      "learning_rate": 5e-05,
      "loss": 0.5885,
      "step": 8000
    },
    {
      "epoch": 2.7,
      "learning_rate": 5e-05,
      "loss": 0.5803,
      "step": 8010
    },
    {
      "epoch": 2.71,
      "learning_rate": 5e-05,
      "loss": 0.576,
      "step": 8020
    },
    {
      "epoch": 2.71,
      "learning_rate": 5e-05,
      "loss": 0.5821,
      "step": 8030
    },
    {
      "epoch": 2.71,
      "learning_rate": 5e-05,
      "loss": 0.5784,
      "step": 8040
    },
    {
      "epoch": 2.72,
      "learning_rate": 5e-05,
      "loss": 0.5787,
      "step": 8050
    },
    {
      "epoch": 2.72,
      "learning_rate": 5e-05,
      "loss": 0.5862,
      "step": 8060
    },
    {
      "epoch": 2.72,
      "learning_rate": 5e-05,
      "loss": 0.5778,
      "step": 8070
    },
    {
      "epoch": 2.73,
      "learning_rate": 5e-05,
      "loss": 0.5894,
      "step": 8080
    },
    {
      "epoch": 2.73,
      "learning_rate": 5e-05,
      "loss": 0.5868,
      "step": 8090
    },
    {
      "epoch": 2.73,
      "learning_rate": 5e-05,
      "loss": 0.578,
      "step": 8100
    },
    {
      "epoch": 2.74,
      "learning_rate": 5e-05,
      "loss": 0.5721,
      "step": 8110
    },
    {
      "epoch": 2.74,
      "learning_rate": 5e-05,
      "loss": 0.5771,
      "step": 8120
    },
    {
      "epoch": 2.74,
      "learning_rate": 5e-05,
      "loss": 0.5795,
      "step": 8130
    },
    {
      "epoch": 2.75,
      "learning_rate": 5e-05,
      "loss": 0.5847,
      "step": 8140
    },
    {
      "epoch": 2.75,
      "learning_rate": 5e-05,
      "loss": 0.5801,
      "step": 8150
    },
    {
      "epoch": 2.75,
      "learning_rate": 5e-05,
      "loss": 0.5759,
      "step": 8160
    },
    {
      "epoch": 2.76,
      "learning_rate": 5e-05,
      "loss": 0.581,
      "step": 8170
    },
    {
      "epoch": 2.76,
      "learning_rate": 5e-05,
      "loss": 0.5797,
      "step": 8180
    },
    {
      "epoch": 2.76,
      "learning_rate": 5e-05,
      "loss": 0.5798,
      "step": 8190
    },
    {
      "epoch": 2.77,
      "learning_rate": 5e-05,
      "loss": 0.5811,
      "step": 8200
    },
    {
      "epoch": 2.77,
      "learning_rate": 5e-05,
      "loss": 0.5724,
      "step": 8210
    },
    {
      "epoch": 2.77,
      "learning_rate": 5e-05,
      "loss": 0.577,
      "step": 8220
    },
    {
      "epoch": 2.78,
      "learning_rate": 5e-05,
      "loss": 0.5823,
      "step": 8230
    },
    {
      "epoch": 2.78,
      "learning_rate": 5e-05,
      "loss": 0.5812,
      "step": 8240
    },
    {
      "epoch": 2.78,
      "learning_rate": 5e-05,
      "loss": 0.5763,
      "step": 8250
    },
    {
      "epoch": 2.79,
      "learning_rate": 5e-05,
      "loss": 0.5789,
      "step": 8260
    },
    {
      "epoch": 2.79,
      "learning_rate": 5e-05,
      "loss": 0.578,
      "step": 8270
    },
    {
      "epoch": 2.79,
      "learning_rate": 5e-05,
      "loss": 0.5826,
      "step": 8280
    },
    {
      "epoch": 2.8,
      "learning_rate": 5e-05,
      "loss": 0.587,
      "step": 8290
    },
    {
      "epoch": 2.8,
      "learning_rate": 5e-05,
      "loss": 0.5859,
      "step": 8300
    },
    {
      "epoch": 2.8,
      "learning_rate": 5e-05,
      "loss": 0.5904,
      "step": 8310
    },
    {
      "epoch": 2.81,
      "learning_rate": 5e-05,
      "loss": 0.5706,
      "step": 8320
    },
    {
      "epoch": 2.81,
      "learning_rate": 5e-05,
      "loss": 0.5772,
      "step": 8330
    },
    {
      "epoch": 2.81,
      "learning_rate": 5e-05,
      "loss": 0.5859,
      "step": 8340
    },
    {
      "epoch": 2.82,
      "learning_rate": 5e-05,
      "loss": 0.5824,
      "step": 8350
    },
    {
      "epoch": 2.82,
      "learning_rate": 5e-05,
      "loss": 0.5798,
      "step": 8360
    },
    {
      "epoch": 2.82,
      "learning_rate": 5e-05,
      "loss": 0.5788,
      "step": 8370
    },
    {
      "epoch": 2.83,
      "learning_rate": 5e-05,
      "loss": 0.5742,
      "step": 8380
    },
    {
      "epoch": 2.83,
      "learning_rate": 5e-05,
      "loss": 0.5829,
      "step": 8390
    },
    {
      "epoch": 2.83,
      "learning_rate": 5e-05,
      "loss": 0.5823,
      "step": 8400
    },
    {
      "epoch": 2.84,
      "learning_rate": 5e-05,
      "loss": 0.5741,
      "step": 8410
    },
    {
      "epoch": 2.84,
      "learning_rate": 5e-05,
      "loss": 0.5789,
      "step": 8420
    },
    {
      "epoch": 2.84,
      "learning_rate": 5e-05,
      "loss": 0.5873,
      "step": 8430
    },
    {
      "epoch": 2.85,
      "learning_rate": 5e-05,
      "loss": 0.5781,
      "step": 8440
    },
    {
      "epoch": 2.85,
      "learning_rate": 5e-05,
      "loss": 0.5789,
      "step": 8450
    },
    {
      "epoch": 2.85,
      "learning_rate": 5e-05,
      "loss": 0.5738,
      "step": 8460
    },
    {
      "epoch": 2.86,
      "learning_rate": 5e-05,
      "loss": 0.5822,
      "step": 8470
    },
    {
      "epoch": 2.86,
      "learning_rate": 5e-05,
      "loss": 0.593,
      "step": 8480
    },
    {
      "epoch": 2.86,
      "learning_rate": 5e-05,
      "loss": 0.58,
      "step": 8490
    },
    {
      "epoch": 2.87,
      "learning_rate": 5e-05,
      "loss": 0.5768,
      "step": 8500
    },
    {
      "epoch": 2.87,
      "learning_rate": 5e-05,
      "loss": 0.5818,
      "step": 8510
    },
    {
      "epoch": 2.87,
      "learning_rate": 5e-05,
      "loss": 0.5802,
      "step": 8520
    },
    {
      "epoch": 2.88,
      "learning_rate": 5e-05,
      "loss": 0.5753,
      "step": 8530
    },
    {
      "epoch": 2.88,
      "learning_rate": 5e-05,
      "loss": 0.5794,
      "step": 8540
    },
    {
      "epoch": 2.88,
      "learning_rate": 5e-05,
      "loss": 0.574,
      "step": 8550
    },
    {
      "epoch": 2.89,
      "learning_rate": 5e-05,
      "loss": 0.5712,
      "step": 8560
    },
    {
      "epoch": 2.89,
      "learning_rate": 5e-05,
      "loss": 0.5756,
      "step": 8570
    },
    {
      "epoch": 2.89,
      "learning_rate": 5e-05,
      "loss": 0.5687,
      "step": 8580
    },
    {
      "epoch": 2.9,
      "learning_rate": 5e-05,
      "loss": 0.5673,
      "step": 8590
    },
    {
      "epoch": 2.9,
      "learning_rate": 5e-05,
      "loss": 0.5771,
      "step": 8600
    },
    {
      "epoch": 2.9,
      "learning_rate": 5e-05,
      "loss": 0.5781,
      "step": 8610
    },
    {
      "epoch": 2.91,
      "learning_rate": 5e-05,
      "loss": 0.573,
      "step": 8620
    },
    {
      "epoch": 2.91,
      "learning_rate": 5e-05,
      "loss": 0.5821,
      "step": 8630
    },
    {
      "epoch": 2.91,
      "learning_rate": 5e-05,
      "loss": 0.5797,
      "step": 8640
    },
    {
      "epoch": 2.92,
      "learning_rate": 5e-05,
      "loss": 0.5812,
      "step": 8650
    },
    {
      "epoch": 2.92,
      "learning_rate": 5e-05,
      "loss": 0.5777,
      "step": 8660
    },
    {
      "epoch": 2.92,
      "learning_rate": 5e-05,
      "loss": 0.5755,
      "step": 8670
    },
    {
      "epoch": 2.93,
      "learning_rate": 5e-05,
      "loss": 0.5988,
      "step": 8680
    },
    {
      "epoch": 2.93,
      "learning_rate": 5e-05,
      "loss": 0.5892,
      "step": 8690
    },
    {
      "epoch": 2.93,
      "learning_rate": 5e-05,
      "loss": 0.576,
      "step": 8700
    },
    {
      "epoch": 2.94,
      "learning_rate": 5e-05,
      "loss": 0.5888,
      "step": 8710
    },
    {
      "epoch": 2.94,
      "learning_rate": 5e-05,
      "loss": 0.5845,
      "step": 8720
    },
    {
      "epoch": 2.95,
      "learning_rate": 5e-05,
      "loss": 0.5799,
      "step": 8730
    },
    {
      "epoch": 2.95,
      "learning_rate": 5e-05,
      "loss": 0.5784,
      "step": 8740
    },
    {
      "epoch": 2.95,
      "learning_rate": 5e-05,
      "loss": 0.5774,
      "step": 8750
    },
    {
      "epoch": 2.96,
      "learning_rate": 5e-05,
      "loss": 0.5776,
      "step": 8760
    },
    {
      "epoch": 2.96,
      "learning_rate": 5e-05,
      "loss": 0.5793,
      "step": 8770
    },
    {
      "epoch": 2.96,
      "learning_rate": 5e-05,
      "loss": 0.5814,
      "step": 8780
    },
    {
      "epoch": 2.97,
      "learning_rate": 5e-05,
      "loss": 0.5645,
      "step": 8790
    },
    {
      "epoch": 2.97,
      "learning_rate": 5e-05,
      "loss": 0.5851,
      "step": 8800
    },
    {
      "epoch": 2.97,
      "learning_rate": 5e-05,
      "loss": 0.5745,
      "step": 8810
    },
    {
      "epoch": 2.98,
      "learning_rate": 5e-05,
      "loss": 0.5776,
      "step": 8820
    },
    {
      "epoch": 2.98,
      "learning_rate": 5e-05,
      "loss": 0.5755,
      "step": 8830
    },
    {
      "epoch": 2.98,
      "learning_rate": 5e-05,
      "loss": 0.581,
      "step": 8840
    },
    {
      "epoch": 2.99,
      "learning_rate": 5e-05,
      "loss": 0.577,
      "step": 8850
    },
    {
      "epoch": 2.99,
      "learning_rate": 5e-05,
      "loss": 0.5703,
      "step": 8860
    },
    {
      "epoch": 2.99,
      "learning_rate": 5e-05,
      "loss": 0.5934,
      "step": 8870
    },
    {
      "epoch": 3.0,
      "learning_rate": 5e-05,
      "loss": 0.5754,
      "step": 8880
    },
    {
      "epoch": 3.0,
      "learning_rate": 5e-05,
      "loss": 0.5745,
      "step": 8890
    },
    {
      "epoch": 3.0,
      "step": 8892,
      "total_flos": 0.0,
      "train_loss": 0.6122167715069629,
      "train_runtime": 62943.9838,
      "train_samples_per_second": 36.166,
      "train_steps_per_second": 0.141
    }
  ],
  "logging_steps": 10,
  "max_steps": 8892,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}