{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 8969,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 14.375,
      "learning_rate": 0.001,
      "loss": 10.66,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.15625,
      "learning_rate": 0.001,
      "loss": 9.0861,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.89453125,
      "learning_rate": 0.001,
      "loss": 7.5929,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.80078125,
      "learning_rate": 0.001,
      "loss": 7.2385,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.890625,
      "learning_rate": 0.001,
      "loss": 7.1247,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.55078125,
      "learning_rate": 0.001,
      "loss": 6.9193,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001,
      "loss": 6.8173,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.68359375,
      "learning_rate": 0.001,
      "loss": 6.7162,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001,
      "loss": 6.6121,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7265625,
      "learning_rate": 0.001,
      "loss": 6.4806,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.63671875,
      "learning_rate": 0.001,
      "loss": 6.4596,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.875,
      "learning_rate": 0.001,
      "loss": 6.3081,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8125,
      "learning_rate": 0.001,
      "loss": 6.2485,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.765625,
      "learning_rate": 0.001,
      "loss": 6.1595,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.80859375,
      "learning_rate": 0.001,
      "loss": 6.113,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.89453125,
      "learning_rate": 0.001,
      "loss": 6.0143,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.203125,
      "learning_rate": 0.001,
      "loss": 5.9454,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.78515625,
      "learning_rate": 0.001,
      "loss": 5.8563,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.89453125,
      "learning_rate": 0.001,
      "loss": 5.8343,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6015625,
      "learning_rate": 0.001,
      "loss": 5.8936,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.765625,
      "learning_rate": 0.001,
      "loss": 5.6957,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 5.6288,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001,
      "loss": 5.601,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001,
      "loss": 5.6385,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5625,
      "learning_rate": 0.001,
      "loss": 5.5606,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.625,
      "learning_rate": 0.001,
      "loss": 5.53,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001,
      "loss": 5.512,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.91015625,
      "learning_rate": 0.001,
      "loss": 5.3661,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5546875,
      "learning_rate": 0.001,
      "loss": 5.4498,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.54296875,
      "learning_rate": 0.001,
      "loss": 5.3858,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.59375,
      "learning_rate": 0.001,
      "loss": 5.3242,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9140625,
      "learning_rate": 0.001,
      "loss": 5.3414,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.15625,
      "learning_rate": 0.001,
      "loss": 5.3306,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.79296875,
      "learning_rate": 0.001,
      "loss": 5.2121,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.578125,
      "learning_rate": 0.001,
      "loss": 5.1801,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001,
      "loss": 5.1879,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.76953125,
      "learning_rate": 0.001,
      "loss": 5.0849,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.75,
      "learning_rate": 0.001,
      "loss": 5.1201,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.77734375,
      "learning_rate": 0.001,
      "loss": 5.0651,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7109375,
      "learning_rate": 0.001,
      "loss": 5.0007,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001,
      "loss": 5.0496,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5390625,
      "learning_rate": 0.001,
      "loss": 4.9546,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.73828125,
      "learning_rate": 0.001,
      "loss": 4.9174,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.609375,
      "learning_rate": 0.001,
      "loss": 4.8822,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6171875,
      "learning_rate": 0.001,
      "loss": 4.9179,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001,
      "loss": 4.8214,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.625,
      "learning_rate": 0.001,
      "loss": 4.8761,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.55078125,
      "learning_rate": 0.001,
      "loss": 4.9586,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001,
      "loss": 4.8608,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0703125,
      "learning_rate": 0.001,
      "loss": 4.8165,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001,
      "loss": 4.8616,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.671875,
      "learning_rate": 0.001,
      "loss": 4.7908,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.52734375,
      "learning_rate": 0.001,
      "loss": 4.7085,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6484375,
      "learning_rate": 0.001,
      "loss": 4.6968,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.67578125,
      "learning_rate": 0.001,
      "loss": 4.648,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001,
      "loss": 4.7703,
      "step": 275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.71484375,
      "learning_rate": 0.001,
      "loss": 4.6552,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.51953125,
      "learning_rate": 0.001,
      "loss": 4.6484,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.546875,
      "learning_rate": 0.001,
      "loss": 4.6077,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.69140625,
      "learning_rate": 0.001,
      "loss": 4.5301,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 4.5485,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.73046875,
      "learning_rate": 0.001,
      "loss": 4.6582,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.546875,
      "learning_rate": 0.001,
      "loss": 4.6396,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5625,
      "learning_rate": 0.001,
      "loss": 4.4431,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5859375,
      "learning_rate": 0.001,
      "loss": 4.4276,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.58203125,
      "learning_rate": 0.001,
      "loss": 4.5243,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001,
      "loss": 4.4766,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4765625,
      "learning_rate": 0.001,
      "loss": 4.5035,
      "step": 335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 4.4449,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5625,
      "learning_rate": 0.001,
      "loss": 4.3884,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.58203125,
      "learning_rate": 0.001,
      "loss": 4.4832,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.546875,
      "learning_rate": 0.001,
      "loss": 4.3619,
      "step": 355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001,
      "loss": 4.4318,
      "step": 360
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.59375,
      "learning_rate": 0.001,
      "loss": 4.3091,
      "step": 365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.609375,
      "learning_rate": 0.001,
      "loss": 4.3793,
      "step": 370
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001,
      "loss": 4.3673,
      "step": 375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7734375,
      "learning_rate": 0.001,
      "loss": 4.2584,
      "step": 380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5078125,
      "learning_rate": 0.001,
      "loss": 4.432,
      "step": 385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.53515625,
      "learning_rate": 0.001,
      "loss": 4.2351,
      "step": 390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 4.3768,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 4.2398,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.484375,
      "learning_rate": 0.001,
      "loss": 4.2591,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.51953125,
      "learning_rate": 0.001,
      "loss": 4.2083,
      "step": 410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.52734375,
      "learning_rate": 0.001,
      "loss": 4.2354,
      "step": 415
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.640625,
      "learning_rate": 0.001,
      "loss": 4.2133,
      "step": 420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.58203125,
      "learning_rate": 0.001,
      "loss": 4.2055,
      "step": 425
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.56640625,
      "learning_rate": 0.001,
      "loss": 4.1682,
      "step": 430
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.53515625,
      "learning_rate": 0.001,
      "loss": 4.2379,
      "step": 435
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.578125,
      "learning_rate": 0.001,
      "loss": 4.1974,
      "step": 440
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49609375,
      "learning_rate": 0.001,
      "loss": 4.2177,
      "step": 445
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.59375,
      "learning_rate": 0.001,
      "loss": 4.1886,
      "step": 450
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.58203125,
      "learning_rate": 0.001,
      "loss": 4.1948,
      "step": 455
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5390625,
      "learning_rate": 0.001,
      "loss": 4.2911,
      "step": 460
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.515625,
      "learning_rate": 0.001,
      "loss": 4.2212,
      "step": 465
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49609375,
      "learning_rate": 0.001,
      "loss": 4.1132,
      "step": 470
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6796875,
      "learning_rate": 0.001,
      "loss": 4.1323,
      "step": 475
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5546875,
      "learning_rate": 0.001,
      "loss": 4.0926,
      "step": 480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.53125,
      "learning_rate": 0.001,
      "loss": 4.0398,
      "step": 485
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.484375,
      "learning_rate": 0.001,
      "loss": 4.1016,
      "step": 490
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.578125,
      "learning_rate": 0.001,
      "loss": 3.9908,
      "step": 495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 4.1144,
      "step": 500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.59375,
      "learning_rate": 0.001,
      "loss": 4.0445,
      "step": 505
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.52734375,
      "learning_rate": 0.001,
      "loss": 4.0426,
      "step": 510
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5234375,
      "learning_rate": 0.001,
      "loss": 4.0339,
      "step": 515
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001,
      "loss": 4.0517,
      "step": 520
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5625,
      "learning_rate": 0.001,
      "loss": 3.9779,
      "step": 525
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.482421875,
      "learning_rate": 0.001,
      "loss": 3.9787,
      "step": 530
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.494140625,
      "learning_rate": 0.001,
      "loss": 3.9433,
      "step": 535
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.58984375,
      "learning_rate": 0.001,
      "loss": 3.9909,
      "step": 540
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.498046875,
      "learning_rate": 0.001,
      "loss": 3.9582,
      "step": 545
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.453125,
      "learning_rate": 0.001,
      "loss": 3.996,
      "step": 550
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.546875,
      "learning_rate": 0.001,
      "loss": 4.0016,
      "step": 555
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.64453125,
      "learning_rate": 0.001,
      "loss": 3.9915,
      "step": 560
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.640625,
      "learning_rate": 0.001,
      "loss": 3.8457,
      "step": 565
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 3.8556,
      "step": 570
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.578125,
      "learning_rate": 0.001,
      "loss": 3.9233,
      "step": 575
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.62109375,
      "learning_rate": 0.001,
      "loss": 3.9413,
      "step": 580
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.474609375,
      "learning_rate": 0.001,
      "loss": 3.968,
      "step": 585
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.486328125,
      "learning_rate": 0.001,
      "loss": 3.971,
      "step": 590
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4765625,
      "learning_rate": 0.001,
      "loss": 4.0015,
      "step": 595
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001,
      "loss": 3.8761,
      "step": 600
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48828125,
      "learning_rate": 0.001,
      "loss": 3.8149,
      "step": 605
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 3.8305,
      "step": 610
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.498046875,
      "learning_rate": 0.001,
      "loss": 3.7626,
      "step": 615
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46875,
      "learning_rate": 0.001,
      "loss": 3.8127,
      "step": 620
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.52734375,
      "learning_rate": 0.001,
      "loss": 3.8408,
      "step": 625
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4765625,
      "learning_rate": 0.001,
      "loss": 3.9199,
      "step": 630
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.60546875,
      "learning_rate": 0.001,
      "loss": 3.8861,
      "step": 635
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.8037,
      "step": 640
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 3.8155,
      "step": 645
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5859375,
      "learning_rate": 0.001,
      "loss": 3.8044,
      "step": 650
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.7604,
      "step": 655
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5390625,
      "learning_rate": 0.001,
      "loss": 3.7374,
      "step": 660
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.54296875,
      "learning_rate": 0.001,
      "loss": 3.7128,
      "step": 665
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 3.6821,
      "step": 670
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.466796875,
      "learning_rate": 0.001,
      "loss": 3.689,
      "step": 675
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.53125,
      "learning_rate": 0.001,
      "loss": 3.7706,
      "step": 680
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.53125,
      "learning_rate": 0.001,
      "loss": 3.7512,
      "step": 685
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.53515625,
      "learning_rate": 0.001,
      "loss": 3.687,
      "step": 690
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.57421875,
      "learning_rate": 0.001,
      "loss": 3.6497,
      "step": 695
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.55859375,
      "learning_rate": 0.001,
      "loss": 3.7325,
      "step": 700
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.59375,
      "learning_rate": 0.001,
      "loss": 3.6482,
      "step": 705
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48046875,
      "learning_rate": 0.001,
      "loss": 3.7139,
      "step": 710
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.58984375,
      "learning_rate": 0.001,
      "loss": 3.6265,
      "step": 715
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.478515625,
      "learning_rate": 0.001,
      "loss": 3.6359,
      "step": 720
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4609375,
      "learning_rate": 0.001,
      "loss": 3.689,
      "step": 725
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.664,
      "step": 730
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5234375,
      "learning_rate": 0.001,
      "loss": 3.5956,
      "step": 735
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.458984375,
      "learning_rate": 0.001,
      "loss": 3.6903,
      "step": 740
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47265625,
      "learning_rate": 0.001,
      "loss": 3.7775,
      "step": 745
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 3.604,
      "step": 750
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5390625,
      "learning_rate": 0.001,
      "loss": 3.5746,
      "step": 755
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 3.5907,
      "step": 760
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 3.6306,
      "step": 765
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.498046875,
      "learning_rate": 0.001,
      "loss": 3.7143,
      "step": 770
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.478515625,
      "learning_rate": 0.001,
      "loss": 3.6161,
      "step": 775
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 3.6404,
      "step": 780
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 3.6052,
      "step": 785
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.55078125,
      "learning_rate": 0.001,
      "loss": 3.5776,
      "step": 790
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.484375,
      "learning_rate": 0.001,
      "loss": 3.56,
      "step": 795
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.484375,
      "learning_rate": 0.001,
      "loss": 3.5469,
      "step": 800
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5625,
      "learning_rate": 0.001,
      "loss": 3.637,
      "step": 805
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 3.582,
      "step": 810
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.515625,
      "learning_rate": 0.001,
      "loss": 3.6892,
      "step": 815
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.494140625,
      "learning_rate": 0.001,
      "loss": 3.5234,
      "step": 820
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43359375,
      "learning_rate": 0.001,
      "loss": 3.5884,
      "step": 825
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.515625,
      "learning_rate": 0.001,
      "loss": 3.541,
      "step": 830
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.474609375,
      "learning_rate": 0.001,
      "loss": 3.4975,
      "step": 835
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 3.5908,
      "step": 840
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4765625,
      "learning_rate": 0.001,
      "loss": 3.5859,
      "step": 845
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 3.4903,
      "step": 850
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49609375,
      "learning_rate": 0.001,
      "loss": 3.4214,
      "step": 855
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 3.4849,
      "step": 860
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.4287,
      "step": 865
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 3.3918,
      "step": 870
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.4713,
      "step": 875
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.51171875,
      "learning_rate": 0.001,
      "loss": 3.4109,
      "step": 880
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5625,
      "learning_rate": 0.001,
      "loss": 3.4379,
      "step": 885
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.4546,
      "step": 890
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.482421875,
      "learning_rate": 0.001,
      "loss": 3.3923,
      "step": 895
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47265625,
      "learning_rate": 0.001,
      "loss": 3.4302,
      "step": 900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.494140625,
      "learning_rate": 0.001,
      "loss": 3.4475,
      "step": 905
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.51953125,
      "learning_rate": 0.001,
      "loss": 3.3023,
      "step": 910
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 3.3485,
      "step": 915
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4375,
      "learning_rate": 0.001,
      "loss": 3.3677,
      "step": 920
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.52734375,
      "learning_rate": 0.001,
      "loss": 3.5076,
      "step": 925
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.53515625,
      "learning_rate": 0.001,
      "loss": 3.421,
      "step": 930
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.474609375,
      "learning_rate": 0.001,
      "loss": 3.4751,
      "step": 935
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5234375,
      "learning_rate": 0.001,
      "loss": 3.2864,
      "step": 940
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 3.3964,
      "step": 945
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.51171875,
      "learning_rate": 0.001,
      "loss": 3.367,
      "step": 950
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45703125,
      "learning_rate": 0.001,
      "loss": 3.3819,
      "step": 955
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 3.2415,
      "step": 960
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48828125,
      "learning_rate": 0.001,
      "loss": 3.2919,
      "step": 965
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47265625,
      "learning_rate": 0.001,
      "loss": 3.4433,
      "step": 970
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.470703125,
      "learning_rate": 0.001,
      "loss": 3.3476,
      "step": 975
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.2323,
      "step": 980
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5390625,
      "learning_rate": 0.001,
      "loss": 3.3249,
      "step": 985
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.51171875,
      "learning_rate": 0.001,
      "loss": 3.215,
      "step": 990
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 3.2893,
      "step": 995
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.48828125,
      "learning_rate": 0.001,
      "loss": 3.3386,
      "step": 1000
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.2875,
      "step": 1005
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5234375,
      "learning_rate": 0.001,
      "loss": 3.3014,
      "step": 1010
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.458984375,
      "learning_rate": 0.001,
      "loss": 3.176,
      "step": 1015
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46484375,
      "learning_rate": 0.001,
      "loss": 3.2795,
      "step": 1020
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.494140625,
      "learning_rate": 0.001,
      "loss": 3.2852,
      "step": 1025
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5703125,
      "learning_rate": 0.001,
      "loss": 3.278,
      "step": 1030
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.54296875,
      "learning_rate": 0.001,
      "loss": 3.2953,
      "step": 1035
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.482421875,
      "learning_rate": 0.001,
      "loss": 3.217,
      "step": 1040
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.48046875,
      "learning_rate": 0.001,
      "loss": 3.1406,
      "step": 1045
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 3.2346,
      "step": 1050
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.1984,
      "step": 1055
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49609375,
      "learning_rate": 0.001,
      "loss": 3.2343,
      "step": 1060
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 3.1503,
      "step": 1065
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.427734375,
      "learning_rate": 0.001,
      "loss": 3.224,
      "step": 1070
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43359375,
      "learning_rate": 0.001,
      "loss": 3.1678,
      "step": 1075
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.1891,
      "step": 1080
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.515625,
      "learning_rate": 0.001,
      "loss": 3.197,
      "step": 1085
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 3.1608,
      "step": 1090
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46484375,
      "learning_rate": 0.001,
      "loss": 3.2191,
      "step": 1095
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 3.1926,
      "step": 1100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.474609375,
      "learning_rate": 0.001,
      "loss": 3.2302,
      "step": 1105
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.466796875,
      "learning_rate": 0.001,
      "loss": 3.1557,
      "step": 1110
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 3.1696,
      "step": 1115
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.2021,
      "step": 1120
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4609375,
      "learning_rate": 0.001,
      "loss": 3.1396,
      "step": 1125
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 3.126,
      "step": 1130
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44921875,
      "learning_rate": 0.001,
      "loss": 3.1153,
      "step": 1135
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46484375,
      "learning_rate": 0.001,
      "loss": 3.2244,
      "step": 1140
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44140625,
      "learning_rate": 0.001,
      "loss": 3.1862,
      "step": 1145
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.458984375,
      "learning_rate": 0.001,
      "loss": 3.1397,
      "step": 1150
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44921875,
      "learning_rate": 0.001,
      "loss": 3.1137,
      "step": 1155
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 3.0842,
      "step": 1160
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4765625,
      "learning_rate": 0.001,
      "loss": 3.1646,
      "step": 1165
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 3.1085,
      "step": 1170
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.431640625,
      "learning_rate": 0.001,
      "loss": 3.1618,
      "step": 1175
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 3.1494,
      "step": 1180
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 3.1187,
      "step": 1185
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.1668,
      "step": 1190
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.47265625,
      "learning_rate": 0.001,
      "loss": 3.0082,
      "step": 1195
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5234375,
      "learning_rate": 0.001,
      "loss": 3.0856,
      "step": 1200
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43359375,
      "learning_rate": 0.001,
      "loss": 3.0244,
      "step": 1205
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.51953125,
      "learning_rate": 0.001,
      "loss": 3.1119,
      "step": 1210
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.423828125,
      "learning_rate": 0.001,
      "loss": 3.1893,
      "step": 1215
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46484375,
      "learning_rate": 0.001,
      "loss": 3.0317,
      "step": 1220
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.498046875,
      "learning_rate": 0.001,
      "loss": 3.0754,
      "step": 1225
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.0712,
      "step": 1230
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.443359375,
      "learning_rate": 0.001,
      "loss": 3.0459,
      "step": 1235
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.1248,
      "step": 1240
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.515625,
      "learning_rate": 0.001,
      "loss": 3.0001,
      "step": 1245
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 3.0746,
      "step": 1250
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 3.0217,
      "step": 1255
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.65234375,
      "learning_rate": 0.001,
      "loss": 3.0922,
      "step": 1260
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.478515625,
      "learning_rate": 0.001,
      "loss": 3.0769,
      "step": 1265
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 3.0214,
      "step": 1270
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 3.0603,
      "step": 1275
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 3.0081,
      "step": 1280
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 3.1073,
      "step": 1285
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 3.0161,
      "step": 1290
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 3.0399,
      "step": 1295
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.427734375,
      "learning_rate": 0.001,
      "loss": 3.0176,
      "step": 1300
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.466796875,
      "learning_rate": 0.001,
      "loss": 3.0438,
      "step": 1305
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45703125,
      "learning_rate": 0.001,
      "loss": 2.9605,
      "step": 1310
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.490234375,
      "learning_rate": 0.001,
      "loss": 3.1799,
      "step": 1315
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 3.0367,
      "step": 1320
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.48046875,
      "learning_rate": 0.001,
      "loss": 3.0772,
      "step": 1325
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.453125,
      "learning_rate": 0.001,
      "loss": 2.9758,
      "step": 1330
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 2.9411,
      "step": 1335
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 3.0461,
      "step": 1340
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.466796875,
      "learning_rate": 0.001,
      "loss": 3.0639,
      "step": 1345
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.51171875,
      "learning_rate": 0.001,
      "loss": 3.0514,
      "step": 1350
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.396484375,
      "learning_rate": 0.001,
      "loss": 2.9288,
      "step": 1355
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44921875,
      "learning_rate": 0.001,
      "loss": 2.9909,
      "step": 1360
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 3.0404,
      "step": 1365
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4296875,
      "learning_rate": 0.001,
      "loss": 2.9636,
      "step": 1370
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44921875,
      "learning_rate": 0.001,
      "loss": 2.9691,
      "step": 1375
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 3.0421,
      "step": 1380
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.453125,
      "learning_rate": 0.001,
      "loss": 3.0106,
      "step": 1385
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.9206,
      "step": 1390
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.412109375,
      "learning_rate": 0.001,
      "loss": 2.9851,
      "step": 1395
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.427734375,
      "learning_rate": 0.001,
      "loss": 2.8973,
      "step": 1400
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 2.9926,
      "step": 1405
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 2.9713,
      "step": 1410
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5,
      "learning_rate": 0.001,
      "loss": 2.9641,
      "step": 1415
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.431640625,
      "learning_rate": 0.001,
      "loss": 2.9165,
      "step": 1420
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 2.9043,
      "step": 1425
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 3.0078,
      "step": 1430
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 2.9917,
      "step": 1435
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.51953125,
      "learning_rate": 0.001,
      "loss": 2.9517,
      "step": 1440
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.431640625,
      "learning_rate": 0.001,
      "loss": 2.9723,
      "step": 1445
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.8912,
      "step": 1450
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.997,
      "step": 1455
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 2.8812,
      "step": 1460
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.396484375,
      "learning_rate": 0.001,
      "loss": 2.9449,
      "step": 1465
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.8727,
      "step": 1470
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4296875,
      "learning_rate": 0.001,
      "loss": 2.9621,
      "step": 1475
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.8718,
      "step": 1480
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.412109375,
      "learning_rate": 0.001,
      "loss": 2.924,
      "step": 1485
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.431640625,
      "learning_rate": 0.001,
      "loss": 2.975,
      "step": 1490
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.8352,
      "step": 1495
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4375,
      "learning_rate": 0.001,
      "loss": 2.8661,
      "step": 1500
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.8946,
      "step": 1505
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 2.8299,
      "step": 1510
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.8354,
      "step": 1515
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 2.8615,
      "step": 1520
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.858,
      "step": 1525
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43359375,
      "learning_rate": 0.001,
      "loss": 2.9946,
      "step": 1530
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.470703125,
      "learning_rate": 0.001,
      "loss": 2.9449,
      "step": 1535
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.447265625,
      "learning_rate": 0.001,
      "loss": 2.8862,
      "step": 1540
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.404296875,
      "learning_rate": 0.001,
      "loss": 2.8575,
      "step": 1545
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.8596,
      "step": 1550
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 2.9418,
      "step": 1555
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.8957,
      "step": 1560
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4296875,
      "learning_rate": 0.001,
      "loss": 2.8291,
      "step": 1565
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.455078125,
      "learning_rate": 0.001,
      "loss": 2.8985,
      "step": 1570
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.9283,
      "step": 1575
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 2.9447,
      "step": 1580
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 2.9154,
      "step": 1585
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.470703125,
      "learning_rate": 0.001,
      "loss": 2.9456,
      "step": 1590
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.8641,
      "step": 1595
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.996,
      "step": 1600
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.431640625,
      "learning_rate": 0.001,
      "loss": 2.9102,
      "step": 1605
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 2.8548,
      "step": 1610
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.423828125,
      "learning_rate": 0.001,
      "loss": 2.8116,
      "step": 1615
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.9482,
      "step": 1620
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 2.8992,
      "step": 1625
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.8682,
      "step": 1630
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.412109375,
      "learning_rate": 0.001,
      "loss": 2.8866,
      "step": 1635
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 2.8674,
      "step": 1640
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 2.8261,
      "step": 1645
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 3.0015,
      "step": 1650
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 2.859,
      "step": 1655
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.451171875,
      "learning_rate": 0.001,
      "loss": 2.8403,
      "step": 1660
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.8866,
      "step": 1665
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.8863,
      "step": 1670
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 2.8138,
      "step": 1675
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.9055,
      "step": 1680
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.494140625,
      "learning_rate": 0.001,
      "loss": 2.8598,
      "step": 1685
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.8609,
      "step": 1690
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.7287,
      "step": 1695
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43359375,
      "learning_rate": 0.001,
      "loss": 2.8634,
      "step": 1700
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 2.8631,
      "step": 1705
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43359375,
      "learning_rate": 0.001,
      "loss": 2.8012,
      "step": 1710
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 2.8285,
      "step": 1715
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.8212,
      "step": 1720
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.8448,
      "step": 1725
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.453125,
      "learning_rate": 0.001,
      "loss": 2.8296,
      "step": 1730
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.412109375,
      "learning_rate": 0.001,
      "loss": 2.8172,
      "step": 1735
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 2.8287,
      "step": 1740
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.396484375,
      "learning_rate": 0.001,
      "loss": 2.8966,
      "step": 1745
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 2.7409,
      "step": 1750
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 2.8212,
      "step": 1755
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.7865,
      "step": 1760
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.423828125,
      "learning_rate": 0.001,
      "loss": 2.8383,
      "step": 1765
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 2.8843,
      "step": 1770
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.7954,
      "step": 1775
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.404296875,
      "learning_rate": 0.001,
      "loss": 2.7362,
      "step": 1780
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.7698,
      "step": 1785
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4921875,
      "learning_rate": 0.001,
      "loss": 2.9273,
      "step": 1790
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.774,
      "step": 1795
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.842,
      "step": 1800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.443359375,
      "learning_rate": 0.001,
      "loss": 2.7766,
      "step": 1805
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 2.7165,
      "step": 1810
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.427734375,
      "learning_rate": 0.001,
      "loss": 2.8396,
      "step": 1815
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.7133,
      "step": 1820
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.7195,
      "step": 1825
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.47265625,
      "learning_rate": 0.001,
      "loss": 2.7769,
      "step": 1830
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 2.8127,
      "step": 1835
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 2.7074,
      "step": 1840
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.380859375,
      "learning_rate": 0.001,
      "loss": 2.8092,
      "step": 1845
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4375,
      "learning_rate": 0.001,
      "loss": 2.8355,
      "step": 1850
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.412109375,
      "learning_rate": 0.001,
      "loss": 2.7602,
      "step": 1855
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.7212,
      "step": 1860
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.8304,
      "step": 1865
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.8164,
      "step": 1870
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 2.8169,
      "step": 1875
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.7487,
      "step": 1880
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.7734,
      "step": 1885
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 2.7757,
      "step": 1890
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.8023,
      "step": 1895
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.7523,
      "step": 1900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.818,
      "step": 1905
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.7938,
      "step": 1910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.7914,
      "step": 1915
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.431640625,
      "learning_rate": 0.001,
      "loss": 2.7482,
      "step": 1920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.474609375,
      "learning_rate": 0.001,
      "loss": 2.7364,
      "step": 1925
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.7383,
      "step": 1930
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 2.7172,
      "step": 1935
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.6953,
      "step": 1940
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.6664,
      "step": 1945
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.6742,
      "step": 1950
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44140625,
      "learning_rate": 0.001,
      "loss": 2.7493,
      "step": 1955
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.7087,
      "step": 1960
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.7275,
      "step": 1965
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4296875,
      "learning_rate": 0.001,
      "loss": 2.7522,
      "step": 1970
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.7288,
      "step": 1975
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.7241,
      "step": 1980
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.7156,
      "step": 1985
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 2.7864,
      "step": 1990
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46875,
      "learning_rate": 0.001,
      "loss": 2.7317,
      "step": 1995
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 2.6977,
      "step": 2000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.392578125,
      "learning_rate": 0.001,
      "loss": 2.7876,
      "step": 2005
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.380859375,
      "learning_rate": 0.001,
      "loss": 2.8518,
      "step": 2010
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.6466,
      "step": 2015
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.8142,
      "step": 2020
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.6646,
      "step": 2025
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 2.7533,
      "step": 2030
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.7672,
      "step": 2035
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.6538,
      "step": 2040
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.7903,
      "step": 2045
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.6225,
      "step": 2050
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.6761,
      "step": 2055
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.7514,
      "step": 2060
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.7695,
      "step": 2065
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.6763,
      "step": 2070
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.7686,
      "step": 2075
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.6337,
      "step": 2080
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 2.7801,
      "step": 2085
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.7185,
      "step": 2090
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.7413,
      "step": 2095
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6984,
      "step": 2100
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.7253,
      "step": 2105
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.7485,
      "step": 2110
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.6853,
      "step": 2115
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.6931,
      "step": 2120
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.6173,
      "step": 2125
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 2.624,
      "step": 2130
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.6629,
      "step": 2135
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.7097,
      "step": 2140
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.7286,
      "step": 2145
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.404296875,
      "learning_rate": 0.001,
      "loss": 2.7264,
      "step": 2150
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.47265625,
      "learning_rate": 0.001,
      "loss": 2.6504,
      "step": 2155
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.392578125,
      "learning_rate": 0.001,
      "loss": 2.7058,
      "step": 2160
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.7127,
      "step": 2165
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6349,
      "step": 2170
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.375,
      "learning_rate": 0.001,
      "loss": 2.7502,
      "step": 2175
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.7021,
      "step": 2180
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.470703125,
      "learning_rate": 0.001,
      "loss": 2.6564,
      "step": 2185
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.6715,
      "step": 2190
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.421875,
      "learning_rate": 0.001,
      "loss": 2.7572,
      "step": 2195
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.6954,
      "step": 2200
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.7125,
      "step": 2205
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.6772,
      "step": 2210
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.8311,
      "step": 2215
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.6858,
      "step": 2220
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44140625,
      "learning_rate": 0.001,
      "loss": 2.6552,
      "step": 2225
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.380859375,
      "learning_rate": 0.001,
      "loss": 2.7589,
      "step": 2230
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.6547,
      "step": 2235
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.6727,
      "step": 2240
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.7218,
      "step": 2245
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.439453125,
      "learning_rate": 0.001,
      "loss": 2.6825,
      "step": 2250
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.7252,
      "step": 2255
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.7026,
      "step": 2260
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.7449,
      "step": 2265
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.5876,
      "step": 2270
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.6584,
      "step": 2275
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.6238,
      "step": 2280
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.7556,
      "step": 2285
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.6804,
      "step": 2290
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.6864,
      "step": 2295
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.5995,
      "step": 2300
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.6537,
      "step": 2305
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 2.6645,
      "step": 2310
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.7136,
      "step": 2315
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.6908,
      "step": 2320
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.7102,
      "step": 2325
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.50390625,
      "learning_rate": 0.001,
      "loss": 2.7089,
      "step": 2330
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.6295,
      "step": 2335
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.6105,
      "step": 2340
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41796875,
      "learning_rate": 0.001,
      "loss": 2.6332,
      "step": 2345
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.6549,
      "step": 2350
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.7258,
      "step": 2355
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.6932,
      "step": 2360
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.5853,
      "step": 2365
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6075,
      "step": 2370
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.7695,
      "step": 2375
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.675,
      "step": 2380
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.6556,
      "step": 2385
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.6259,
      "step": 2390
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.627,
      "step": 2395
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 0.001,
      "loss": 2.6976,
      "step": 2400
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.6732,
      "step": 2405
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.6387,
      "step": 2410
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.613,
      "step": 2415
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.6863,
      "step": 2420
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.6637,
      "step": 2425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.5994,
      "step": 2430
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.6521,
      "step": 2435
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.423828125,
      "learning_rate": 0.001,
      "loss": 2.6928,
      "step": 2440
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.6497,
      "step": 2445
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.412109375,
      "learning_rate": 0.001,
      "loss": 2.6095,
      "step": 2450
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.637,
      "step": 2455
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.5598,
      "step": 2460
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.419921875,
      "learning_rate": 0.001,
      "loss": 2.5683,
      "step": 2465
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.7179,
      "step": 2470
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.5267,
      "step": 2475
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.643,
      "step": 2480
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.5632,
      "step": 2485
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.611,
      "step": 2490
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.572,
      "step": 2495
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.6059,
      "step": 2500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.5368,
      "step": 2505
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.447265625,
      "learning_rate": 0.001,
      "loss": 2.6391,
      "step": 2510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.6533,
      "step": 2515
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 2.6177,
      "step": 2520
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.5711,
      "step": 2525
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.392578125,
      "learning_rate": 0.001,
      "loss": 2.6115,
      "step": 2530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.674,
      "step": 2535
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.6826,
      "step": 2540
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.5226,
      "step": 2545
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.6192,
      "step": 2550
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.6914,
      "step": 2555
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.6802,
      "step": 2560
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.5906,
      "step": 2565
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.5617,
      "step": 2570
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.5559,
      "step": 2575
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.5432,
      "step": 2580
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.6188,
      "step": 2585
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.375,
      "learning_rate": 0.001,
      "loss": 2.594,
      "step": 2590
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.6327,
      "step": 2595
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.396484375,
      "learning_rate": 0.001,
      "loss": 2.6117,
      "step": 2600
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.6003,
      "step": 2605
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.5405,
      "step": 2610
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.5711,
      "step": 2615
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.6618,
      "step": 2620
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.63,
      "step": 2625
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.404296875,
      "learning_rate": 0.001,
      "loss": 2.5909,
      "step": 2630
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.5804,
      "step": 2635
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.392578125,
      "learning_rate": 0.001,
      "loss": 2.6332,
      "step": 2640
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6762,
      "step": 2645
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.5429,
      "step": 2650
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.6548,
      "step": 2655
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.5452,
      "step": 2660
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.6768,
      "step": 2665
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.6133,
      "step": 2670
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.6124,
      "step": 2675
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5989,
      "step": 2680
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.5802,
      "step": 2685
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.5502,
      "step": 2690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.416015625,
      "learning_rate": 0.001,
      "loss": 2.5251,
      "step": 2695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.6167,
      "step": 2700
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5672,
      "step": 2705
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.5343,
      "step": 2710
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5582,
      "step": 2715
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5223,
      "step": 2720
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.5679,
      "step": 2725
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5854,
      "step": 2730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.375,
      "learning_rate": 0.001,
      "loss": 2.5077,
      "step": 2735
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5738,
      "step": 2740
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5829,
      "step": 2745
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.5914,
      "step": 2750
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.5735,
      "step": 2755
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.400390625,
      "learning_rate": 0.001,
      "loss": 2.5015,
      "step": 2760
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6339,
      "step": 2765
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.396484375,
      "learning_rate": 0.001,
      "loss": 2.6208,
      "step": 2770
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.5872,
      "step": 2775
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.5484,
      "step": 2780
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.6061,
      "step": 2785
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.7287,
      "step": 2790
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4569,
      "step": 2795
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.5641,
      "step": 2800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5877,
      "step": 2805
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40234375,
      "learning_rate": 0.001,
      "loss": 2.5578,
      "step": 2810
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.5764,
      "step": 2815
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.5233,
      "step": 2820
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.5845,
      "step": 2825
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.6143,
      "step": 2830
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.5839,
      "step": 2835
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5518,
      "step": 2840
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5763,
      "step": 2845
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.5789,
      "step": 2850
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.6161,
      "step": 2855
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.4979,
      "step": 2860
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.462890625,
      "learning_rate": 0.001,
      "loss": 2.5239,
      "step": 2865
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.5278,
      "step": 2870
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.5839,
      "step": 2875
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5451,
      "step": 2880
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.5832,
      "step": 2885
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.6994,
      "step": 2890
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5226,
      "step": 2895
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.568,
      "step": 2900
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.5504,
      "step": 2905
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3984375,
      "learning_rate": 0.001,
      "loss": 2.5429,
      "step": 2910
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.5131,
      "step": 2915
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.5571,
      "step": 2920
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 2.63,
      "step": 2925
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.5114,
      "step": 2930
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.587,
      "step": 2935
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6734,
      "step": 2940
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.6092,
      "step": 2945
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.4977,
      "step": 2950
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.5451,
      "step": 2955
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.6318,
      "step": 2960
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.489,
      "step": 2965
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.4592,
      "step": 2970
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.392578125,
      "learning_rate": 0.001,
      "loss": 2.5569,
      "step": 2975
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.380859375,
      "learning_rate": 0.001,
      "loss": 2.4976,
      "step": 2980
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.5439,
      "step": 2985
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5567,
      "step": 2990
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.5471,
      "step": 2995
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.5934,
      "step": 3000
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.527,
      "step": 3005
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5382,
      "step": 3010
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.5225,
      "step": 3015
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.571,
      "step": 3020
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5245,
      "step": 3025
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5326,
      "step": 3030
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.5773,
      "step": 3035
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.5171,
      "step": 3040
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.5531,
      "step": 3045
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.5284,
      "step": 3050
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.5586,
      "step": 3055
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.5146,
      "step": 3060
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.5981,
      "step": 3065
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.5523,
      "step": 3070
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5508,
      "step": 3075
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.5361,
      "step": 3080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.396484375,
      "learning_rate": 0.001,
      "loss": 2.6033,
      "step": 3085
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.6071,
      "step": 3090
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.5747,
      "step": 3095
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.4875,
      "step": 3100
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5199,
      "step": 3105
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.5829,
      "step": 3110
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4627,
      "step": 3115
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.375,
      "learning_rate": 0.001,
      "loss": 2.5592,
      "step": 3120
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.6057,
      "step": 3125
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.380859375,
      "learning_rate": 0.001,
      "loss": 2.538,
      "step": 3130
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.5633,
      "step": 3135
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.5373,
      "step": 3140
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.4773,
      "step": 3145
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41015625,
      "learning_rate": 0.001,
      "loss": 2.5214,
      "step": 3150
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.498,
      "step": 3155
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4634,
      "step": 3160
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4578,
      "step": 3165
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.5839,
      "step": 3170
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40625,
      "learning_rate": 0.001,
      "loss": 2.5751,
      "step": 3175
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4664,
      "step": 3180
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4624,
      "step": 3185
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42578125,
      "learning_rate": 0.001,
      "loss": 2.5735,
      "step": 3190
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5191,
      "step": 3195
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.408203125,
      "learning_rate": 0.001,
      "loss": 2.4841,
      "step": 3200
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.5304,
      "step": 3205
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4461,
      "step": 3210
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.5183,
      "step": 3215
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.481,
      "step": 3220
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.6264,
      "step": 3225
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.544,
      "step": 3230
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5118,
      "step": 3235
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.5537,
      "step": 3240
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5551,
      "step": 3245
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4869,
      "step": 3250
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5788,
      "step": 3255
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.5389,
      "step": 3260
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5014,
      "step": 3265
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.5224,
      "step": 3270
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4611,
      "step": 3275
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.5188,
      "step": 3280
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5296,
      "step": 3285
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5364,
      "step": 3290
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.5207,
      "step": 3295
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.5104,
      "step": 3300
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.513,
      "step": 3305
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.523,
      "step": 3310
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.4677,
      "step": 3315
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.528,
      "step": 3320
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.5352,
      "step": 3325
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4736,
      "step": 3330
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.5234,
      "step": 3335
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.4768,
      "step": 3340
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.5709,
      "step": 3345
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.4996,
      "step": 3350
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.4553,
      "step": 3355
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.5061,
      "step": 3360
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.4686,
      "step": 3365
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.5281,
      "step": 3370
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.5412,
      "step": 3375
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.4708,
      "step": 3380
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.4491,
      "step": 3385
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4432,
      "step": 3390
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.5601,
      "step": 3395
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4834,
      "step": 3400
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.5377,
      "step": 3405
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4399,
      "step": 3410
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4336,
      "step": 3415
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.5348,
      "step": 3420
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4751,
      "step": 3425
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39453125,
      "learning_rate": 0.001,
      "loss": 2.4556,
      "step": 3430
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.426,
      "step": 3435
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4736,
      "step": 3440
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.4077,
      "step": 3445
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.5495,
      "step": 3450
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.380859375,
      "learning_rate": 0.001,
      "loss": 2.4385,
      "step": 3455
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.5411,
      "step": 3460
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.4391,
      "step": 3465
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.5512,
      "step": 3470
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.5375,
      "step": 3475
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.4875,
      "step": 3480
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.5244,
      "step": 3485
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.5694,
      "step": 3490
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5128,
      "step": 3495
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4621,
      "step": 3500
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.3882,
      "step": 3505
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4457,
      "step": 3510
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.6129,
      "step": 3515
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.4673,
      "step": 3520
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4322,
      "step": 3525
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4767,
      "step": 3530
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.5015,
      "step": 3535
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4496,
      "step": 3540
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.4416,
      "step": 3545
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.4924,
      "step": 3550
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.444,
      "step": 3555
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.5554,
      "step": 3560
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.5222,
      "step": 3565
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4904,
      "step": 3570
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.51,
      "step": 3575
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.491,
      "step": 3580
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.5569,
      "step": 3585
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4702,
      "step": 3590
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.4147,
      "step": 3595
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.5733,
      "step": 3600
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.4938,
      "step": 3605
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4076,
      "step": 3610
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.4868,
      "step": 3615
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.4626,
      "step": 3620
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.5284,
      "step": 3625
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4889,
      "step": 3630
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.4571,
      "step": 3635
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.5282,
      "step": 3640
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.5391,
      "step": 3645
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.5301,
      "step": 3650
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.4556,
      "step": 3655
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4623,
      "step": 3660
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.5289,
      "step": 3665
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.4913,
      "step": 3670
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4986,
      "step": 3675
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.5109,
      "step": 3680
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.5675,
      "step": 3685
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.404296875,
      "learning_rate": 0.001,
      "loss": 2.5055,
      "step": 3690
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4677,
      "step": 3695
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5347,
      "step": 3700
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4140625,
      "learning_rate": 0.001,
      "loss": 2.4168,
      "step": 3705
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4156,
      "step": 3710
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.5117,
      "step": 3715
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.5008,
      "step": 3720
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.5145,
      "step": 3725
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4302,
      "step": 3730
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.5553,
      "step": 3735
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4049,
      "step": 3740
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4319,
      "step": 3745
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.5425,
      "step": 3750
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.5042,
      "step": 3755
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.4445,
      "step": 3760
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4256,
      "step": 3765
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4933,
      "step": 3770
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.4679,
      "step": 3775
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4482,
      "step": 3780
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4219,
      "step": 3785
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.5463,
      "step": 3790
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4565,
      "step": 3795
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4338,
      "step": 3800
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4545,
      "step": 3805
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4431,
      "step": 3810
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.5105,
      "step": 3815
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4994,
      "step": 3820
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.4036,
      "step": 3825
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.3943,
      "step": 3830
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4016,
      "step": 3835
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.3733,
      "step": 3840
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.4746,
      "step": 3845
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.4641,
      "step": 3850
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.4163,
      "step": 3855
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.5641,
      "step": 3860
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.4459,
      "step": 3865
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.48,
      "step": 3870
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.3711,
      "step": 3875
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.4004,
      "step": 3880
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3915,
      "step": 3885
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3828125,
      "learning_rate": 0.001,
      "loss": 2.4762,
      "step": 3890
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.4994,
      "step": 3895
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4486,
      "step": 3900
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.4192,
      "step": 3905
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4359,
      "step": 3910
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.4987,
      "step": 3915
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.494,
      "step": 3920
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4081,
      "step": 3925
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4449,
      "step": 3930
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.4008,
      "step": 3935
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.4596,
      "step": 3940
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4376,
      "step": 3945
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4765,
      "step": 3950
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4726,
      "step": 3955
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4462,
      "step": 3960
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3776,
      "step": 3965
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3864,
      "step": 3970
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4396,
      "step": 3975
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.5275,
      "step": 3980
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.5021,
      "step": 3985
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.5053,
      "step": 3990
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4067,
      "step": 3995
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.3718,
      "step": 4000
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.396,
      "step": 4005
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4531,
      "step": 4010
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.4581,
      "step": 4015
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.4616,
      "step": 4020
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.4446,
      "step": 4025
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.3241,
      "step": 4030
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.5278,
      "step": 4035
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.4402,
      "step": 4040
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.375,
      "learning_rate": 0.001,
      "loss": 2.526,
      "step": 4045
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4471,
      "step": 4050
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.5027,
      "step": 4055
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4513,
      "step": 4060
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4194,
      "step": 4065
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4691,
      "step": 4070
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.4777,
      "step": 4075
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.3957,
      "step": 4080
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.4746,
      "step": 4085
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3968,
      "step": 4090
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.4137,
      "step": 4095
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.5123,
      "step": 4100
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4194,
      "step": 4105
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3577,
      "step": 4110
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.4128,
      "step": 4115
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4063,
      "step": 4120
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37890625,
      "learning_rate": 0.001,
      "loss": 2.4535,
      "step": 4125
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.3966,
      "step": 4130
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.4123,
      "step": 4135
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.45703125,
      "learning_rate": 0.001,
      "loss": 2.4634,
      "step": 4140
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.3939,
      "step": 4145
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3884,
      "step": 4150
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4213,
      "step": 4155
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.3715,
      "step": 4160
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.4166,
      "step": 4165
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.4946,
      "step": 4170
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.4918,
      "step": 4175
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.381,
      "step": 4180
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.3991,
      "step": 4185
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4139,
      "step": 4190
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.5107,
      "step": 4195
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.387,
      "step": 4200
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.447,
      "step": 4205
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3473,
      "step": 4210
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3907,
      "step": 4215
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.3656,
      "step": 4220
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.5121,
      "step": 4225
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4212,
      "step": 4230
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3811,
      "step": 4235
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.368,
      "step": 4240
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.4631,
      "step": 4245
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4267,
      "step": 4250
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.4453,
      "step": 4255
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3846,
      "step": 4260
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3578,
      "step": 4265
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.4811,
      "step": 4270
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.4718,
      "step": 4275
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4631,
      "step": 4280
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.428,
      "step": 4285
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.4018,
      "step": 4290
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3948,
      "step": 4295
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3939,
      "step": 4300
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3406,
      "step": 4305
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.427,
      "step": 4310
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4806,
      "step": 4315
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3565,
      "step": 4320
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4658,
      "step": 4325
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3383,
      "step": 4330
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4433,
      "step": 4335
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.4428,
      "step": 4340
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3914,
      "step": 4345
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.4283,
      "step": 4350
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.4238,
      "step": 4355
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3536,
      "step": 4360
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3306,
      "step": 4365
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.4716,
      "step": 4370
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3698,
      "step": 4375
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4516,
      "step": 4380
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3544,
      "step": 4385
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4579,
      "step": 4390
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4471,
      "step": 4395
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.4688,
      "step": 4400
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4272,
      "step": 4405
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3917,
      "step": 4410
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.4167,
      "step": 4415
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.4342,
      "step": 4420
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.5076,
      "step": 4425
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4897,
      "step": 4430
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4231,
      "step": 4435
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.358,
      "step": 4440
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3515,
      "step": 4445
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4139,
      "step": 4450
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.5031,
      "step": 4455
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3742,
      "step": 4460
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.467,
      "step": 4465
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3063,
      "step": 4470
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2742,
      "step": 4475
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.3993,
      "step": 4480
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.3711,
      "step": 4485
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3366,
      "step": 4490
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.3711,
      "step": 4495
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4112,
      "step": 4500
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.5269,
      "step": 4505
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4256,
      "step": 4510
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4714,
      "step": 4515
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4565,
      "step": 4520
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4071,
      "step": 4525
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4281,
      "step": 4530
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.4384,
      "step": 4535
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4587,
      "step": 4540
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.4184,
      "step": 4545
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3867,
      "step": 4550
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3673,
      "step": 4555
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.4,
      "step": 4560
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.4027,
      "step": 4565
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.4282,
      "step": 4570
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4324,
      "step": 4575
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4159,
      "step": 4580
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3382,
      "step": 4585
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3968,
      "step": 4590
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3331,
      "step": 4595
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.4229,
      "step": 4600
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37109375,
      "learning_rate": 0.001,
      "loss": 2.3908,
      "step": 4605
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4211,
      "step": 4610
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.4101,
      "step": 4615
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3875,
      "step": 4620
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3807,
      "step": 4625
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3716,
      "step": 4630
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.4305,
      "step": 4635
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.4326,
      "step": 4640
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4278,
      "step": 4645
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3428,
      "step": 4650
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3823,
      "step": 4655
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.4197,
      "step": 4660
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4177,
      "step": 4665
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2897,
      "step": 4670
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3867,
      "step": 4675
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2943,
      "step": 4680
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4413,
      "step": 4685
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3828,
      "step": 4690
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.451,
      "step": 4695
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3077,
      "step": 4700
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.4278,
      "step": 4705
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.371,
      "step": 4710
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.429,
      "step": 4715
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3383,
      "step": 4720
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.3408,
      "step": 4725
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.388671875,
      "learning_rate": 0.001,
      "loss": 2.4862,
      "step": 4730
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.5457,
      "step": 4735
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.4219,
      "step": 4740
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.277,
      "step": 4745
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.3562,
      "step": 4750
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4333,
      "step": 4755
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3555,
      "step": 4760
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.4844,
      "step": 4765
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.435546875,
      "learning_rate": 0.001,
      "loss": 2.4931,
      "step": 4770
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3959,
      "step": 4775
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3608,
      "step": 4780
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3661,
      "step": 4785
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4261,
      "step": 4790
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3362,
      "step": 4795
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4043,
      "step": 4800
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3257,
      "step": 4805
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.4065,
      "step": 4810
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.3205,
      "step": 4815
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.4849,
      "step": 4820
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.373046875,
      "learning_rate": 0.001,
      "loss": 2.4492,
      "step": 4825
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.3474,
      "step": 4830
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.3716,
      "step": 4835
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3561,
      "step": 4840
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3908,
      "step": 4845
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4253,
      "step": 4850
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.3876,
      "step": 4855
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3442,
      "step": 4860
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3322,
      "step": 4865
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3931,
      "step": 4870
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.4097,
      "step": 4875
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3871,
      "step": 4880
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.383,
      "step": 4885
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.375,
      "learning_rate": 0.001,
      "loss": 2.3096,
      "step": 4890
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3468,
      "step": 4895
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.339,
      "step": 4900
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.3719,
      "step": 4905
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3645,
      "step": 4910
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3295,
      "step": 4915
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.4452,
      "step": 4920
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3908,
      "step": 4925
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3639,
      "step": 4930
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3453,
      "step": 4935
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.4583,
      "step": 4940
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3326,
      "step": 4945
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3368,
      "step": 4950
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3238,
      "step": 4955
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3857,
      "step": 4960
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4269,
      "step": 4965
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.4069,
      "step": 4970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.4184,
      "step": 4975
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3375,
      "step": 4980
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3276,
      "step": 4985
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.387,
      "step": 4990
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4152,
      "step": 4995
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4344,
      "step": 5000
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3746,
      "step": 5005
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3692,
      "step": 5010
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3626,
      "step": 5015
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.4002,
      "step": 5020
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.4398,
      "step": 5025
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2824,
      "step": 5030
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3933,
      "step": 5035
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3513,
      "step": 5040
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3421,
      "step": 5045
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3887,
      "step": 5050
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3859,
      "step": 5055
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3026,
      "step": 5060
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.326,
      "step": 5065
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.3422,
      "step": 5070
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3424,
      "step": 5075
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.339,
      "step": 5080
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3462,
      "step": 5085
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3583,
      "step": 5090
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.4476,
      "step": 5095
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3964,
      "step": 5100
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3881,
      "step": 5105
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3989,
      "step": 5110
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3513,
      "step": 5115
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4226,
      "step": 5120
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.384765625,
      "learning_rate": 0.001,
      "loss": 2.318,
      "step": 5125
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.55078125,
      "learning_rate": 0.001,
      "loss": 2.3471,
      "step": 5130
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3504,
      "step": 5135
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3703,
      "step": 5140
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.356,
      "step": 5145
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3881,
      "step": 5150
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3372,
      "step": 5155
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2871,
      "step": 5160
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4192,
      "step": 5165
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3266,
      "step": 5170
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3311,
      "step": 5175
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3426,
      "step": 5180
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3464,
      "step": 5185
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4051,
      "step": 5190
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3669,
      "step": 5195
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.4088,
      "step": 5200
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3947,
      "step": 5205
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3573,
      "step": 5210
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4141,
      "step": 5215
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3164,
      "step": 5220
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.3358,
      "step": 5225
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.324,
      "step": 5230
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.4203,
      "step": 5235
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3093,
      "step": 5240
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3323,
      "step": 5245
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.3769,
      "step": 5250
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3383,
      "step": 5255
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2656,
      "step": 5260
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2729,
      "step": 5265
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.3279,
      "step": 5270
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.4237,
      "step": 5275
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3641,
      "step": 5280
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.3602,
      "step": 5285
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.385,
      "step": 5290
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3023,
      "step": 5295
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.3724,
      "step": 5300
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3638,
      "step": 5305
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3663,
      "step": 5310
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.365234375,
      "learning_rate": 0.001,
      "loss": 2.4862,
      "step": 5315
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3922,
      "step": 5320
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3677,
      "step": 5325
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3606,
      "step": 5330
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3088,
      "step": 5335
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4073,
      "step": 5340
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3641,
      "step": 5345
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3953,
      "step": 5350
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3133,
      "step": 5355
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3236,
      "step": 5360
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3984,
      "step": 5365
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3896,
      "step": 5370
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3215,
      "step": 5375
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.4614,
      "step": 5380
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3136,
      "step": 5385
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.3548,
      "step": 5390
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3609,
      "step": 5395
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3043,
      "step": 5400
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3837,
      "step": 5405
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.4236,
      "step": 5410
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.3492,
      "step": 5415
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3594,
      "step": 5420
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.4392,
      "step": 5425
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3757,
      "step": 5430
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3445,
      "step": 5435
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.4049,
      "step": 5440
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.4535,
      "step": 5445
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3116,
      "step": 5450
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3645,
      "step": 5455
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.3229,
      "step": 5460
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3648,
      "step": 5465
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.4006,
      "step": 5470
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.357421875,
      "learning_rate": 0.001,
      "loss": 2.3397,
      "step": 5475
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3543,
      "step": 5480
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3028,
      "step": 5485
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3388,
      "step": 5490
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3664,
      "step": 5495
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2125,
      "step": 5500
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4234,
      "step": 5505
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3316,
      "step": 5510
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.367,
      "step": 5515
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.4614,
      "step": 5520
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.2681,
      "step": 5525
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3821,
      "step": 5530
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3998,
      "step": 5535
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3163,
      "step": 5540
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3876,
      "step": 5545
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3965,
      "step": 5550
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.382,
      "step": 5555
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2978,
      "step": 5560
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3291,
      "step": 5565
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3339,
      "step": 5570
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3565,
      "step": 5575
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3237,
      "step": 5580
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4315,
      "step": 5585
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.231,
      "step": 5590
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3903,
      "step": 5595
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2526,
      "step": 5600
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3466,
      "step": 5605
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3012,
      "step": 5610
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.4253,
      "step": 5615
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2762,
      "step": 5620
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2793,
      "step": 5625
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.3468,
      "step": 5630
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.4084,
      "step": 5635
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.347,
      "step": 5640
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3839,
      "step": 5645
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3001,
      "step": 5650
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2755,
      "step": 5655
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2986,
      "step": 5660
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3897,
      "step": 5665
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3523,
      "step": 5670
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3523,
      "step": 5675
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3354,
      "step": 5680
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3529,
      "step": 5685
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.357,
      "step": 5690
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.368,
      "step": 5695
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3112,
      "step": 5700
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3848,
      "step": 5705
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3793,
      "step": 5710
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3272,
      "step": 5715
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.4029,
      "step": 5720
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3996,
      "step": 5725
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.329,
      "step": 5730
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.3392,
      "step": 5735
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3056,
      "step": 5740
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3413,
      "step": 5745
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3949,
      "step": 5750
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.438,
      "step": 5755
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.4453,
      "step": 5760
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3045,
      "step": 5765
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2906,
      "step": 5770
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2219,
      "step": 5775
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3613,
      "step": 5780
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3199,
      "step": 5785
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3443,
      "step": 5790
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2965,
      "step": 5795
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2862,
      "step": 5800
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3733,
      "step": 5805
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2991,
      "step": 5810
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.3591,
      "step": 5815
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3919,
      "step": 5820
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.2326,
      "step": 5825
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3035,
      "step": 5830
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3228,
      "step": 5835
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3327,
      "step": 5840
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.265,
      "step": 5845
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3723,
      "step": 5850
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2181,
      "step": 5855
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.314,
      "step": 5860
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.2494,
      "step": 5865
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38671875,
      "learning_rate": 0.001,
      "loss": 2.2576,
      "step": 5870
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.3208,
      "step": 5875
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3072,
      "step": 5880
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3374,
      "step": 5885
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3156,
      "step": 5890
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3659,
      "step": 5895
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3847,
      "step": 5900
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.29,
      "step": 5905
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.282,
      "step": 5910
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3172,
      "step": 5915
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2918,
      "step": 5920
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3004,
      "step": 5925
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3571,
      "step": 5930
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3381,
      "step": 5935
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2733,
      "step": 5940
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3277,
      "step": 5945
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4188,
      "step": 5950
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2578,
      "step": 5955
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.392578125,
      "learning_rate": 0.001,
      "loss": 2.2164,
      "step": 5960
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.361328125,
      "learning_rate": 0.001,
      "loss": 2.3124,
      "step": 5965
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2956,
      "step": 5970
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2811,
      "step": 5975
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3955,
      "step": 5980
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3116,
      "step": 5985
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3184,
      "step": 5990
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3773,
      "step": 5995
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3402,
      "step": 6000
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3409,
      "step": 6005
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3015,
      "step": 6010
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.2731,
      "step": 6015
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.352,
      "step": 6020
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3109,
      "step": 6025
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3099,
      "step": 6030
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2662,
      "step": 6035
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3723,
      "step": 6040
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3657,
      "step": 6045
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3337,
      "step": 6050
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.2631,
      "step": 6055
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3135,
      "step": 6060
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3438,
      "step": 6065
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2664,
      "step": 6070
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3946,
      "step": 6075
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2588,
      "step": 6080
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.2975,
      "step": 6085
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3101,
      "step": 6090
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3642,
      "step": 6095
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3494,
      "step": 6100
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2839,
      "step": 6105
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3428,
      "step": 6110
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2826,
      "step": 6115
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3481,
      "step": 6120
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3043,
      "step": 6125
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.2965,
      "step": 6130
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2599,
      "step": 6135
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2636,
      "step": 6140
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3423,
      "step": 6145
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2692,
      "step": 6150
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2963,
      "step": 6155
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2212,
      "step": 6160
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.3136,
      "step": 6165
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3659,
      "step": 6170
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2902,
      "step": 6175
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.3175,
      "step": 6180
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3527,
      "step": 6185
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3352,
      "step": 6190
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.458984375,
      "learning_rate": 0.001,
      "loss": 2.3264,
      "step": 6195
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3021,
      "step": 6200
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3128,
      "step": 6205
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3982,
      "step": 6210
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3052,
      "step": 6215
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.269,
      "step": 6220
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.357,
      "step": 6225
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3996,
      "step": 6230
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3695,
      "step": 6235
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3206,
      "step": 6240
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2325,
      "step": 6245
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.334,
      "step": 6250
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2761,
      "step": 6255
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3011,
      "step": 6260
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.3958,
      "step": 6265
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3143,
      "step": 6270
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3211,
      "step": 6275
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2919,
      "step": 6280
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3387,
      "step": 6285
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3049,
      "step": 6290
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3557,
      "step": 6295
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3384,
      "step": 6300
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.3372,
      "step": 6305
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2197,
      "step": 6310
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2533,
      "step": 6315
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2072,
      "step": 6320
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.2987,
      "step": 6325
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2634,
      "step": 6330
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.3597,
      "step": 6335
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2872,
      "step": 6340
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3196,
      "step": 6345
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2671,
      "step": 6350
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2474,
      "step": 6355
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.334,
      "step": 6360
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3749,
      "step": 6365
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2718,
      "step": 6370
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3395,
      "step": 6375
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3243,
      "step": 6380
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3901,
      "step": 6385
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2454,
      "step": 6390
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.287,
      "step": 6395
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.283,
      "step": 6400
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.3251,
      "step": 6405
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3293,
      "step": 6410
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.353515625,
      "learning_rate": 0.001,
      "loss": 2.2905,
      "step": 6415
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.2483,
      "step": 6420
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3314,
      "step": 6425
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3804,
      "step": 6430
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.3254,
      "step": 6435
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3487,
      "step": 6440
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3102,
      "step": 6445
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2659,
      "step": 6450
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.3119,
      "step": 6455
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.3058,
      "step": 6460
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2881,
      "step": 6465
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.2166,
      "step": 6470
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.3242,
      "step": 6475
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3057,
      "step": 6480
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2805,
      "step": 6485
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.2795,
      "step": 6490
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2733,
      "step": 6495
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2482,
      "step": 6500
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2718,
      "step": 6505
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2996,
      "step": 6510
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2118,
      "step": 6515
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.28515625,
      "learning_rate": 0.001,
      "loss": 2.3748,
      "step": 6520
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3675,
      "step": 6525
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2712,
      "step": 6530
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2544,
      "step": 6535
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3075,
      "step": 6540
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.299,
      "step": 6545
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.327,
      "step": 6550
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3941,
      "step": 6555
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2842,
      "step": 6560
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3262,
      "step": 6565
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.359375,
      "learning_rate": 0.001,
      "loss": 2.3961,
      "step": 6570
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3889,
      "step": 6575
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3397,
      "step": 6580
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.2399,
      "step": 6585
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3182,
      "step": 6590
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.3264,
      "step": 6595
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3071,
      "step": 6600
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.4024,
      "step": 6605
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2123,
      "step": 6610
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2309,
      "step": 6615
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.1942,
      "step": 6620
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2871,
      "step": 6625
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3449,
      "step": 6630
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3352,
      "step": 6635
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3952,
      "step": 6640
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2695,
      "step": 6645
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3078,
      "step": 6650
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3105,
      "step": 6655
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3317,
      "step": 6660
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3116,
      "step": 6665
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.2339,
      "step": 6670
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3095,
      "step": 6675
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.239,
      "step": 6680
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.4118,
      "step": 6685
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2235,
      "step": 6690
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.333984375,
      "learning_rate": 0.001,
      "loss": 2.3407,
      "step": 6695
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2263,
      "step": 6700
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2005,
      "step": 6705
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3457,
      "step": 6710
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1686,
      "step": 6715
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2701,
      "step": 6720
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2268,
      "step": 6725
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.232,
      "step": 6730
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2332,
      "step": 6735
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3156,
      "step": 6740
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2292,
      "step": 6745
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.267,
      "step": 6750
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3458,
      "step": 6755
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.2835,
      "step": 6760
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2314,
      "step": 6765
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2215,
      "step": 6770
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3393,
      "step": 6775
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2511,
      "step": 6780
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2021,
      "step": 6785
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2558,
      "step": 6790
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2729,
      "step": 6795
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2082,
      "step": 6800
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2801,
      "step": 6805
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.312,
      "step": 6810
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2518,
      "step": 6815
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2459,
      "step": 6820
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.1997,
      "step": 6825
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3033,
      "step": 6830
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2142,
      "step": 6835
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.3164,
      "step": 6840
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2225,
      "step": 6845
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3232,
      "step": 6850
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2207,
      "step": 6855
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3717,
      "step": 6860
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.2952,
      "step": 6865
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2302,
      "step": 6870
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3487,
      "step": 6875
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2465,
      "step": 6880
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3268,
      "step": 6885
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2749,
      "step": 6890
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3893,
      "step": 6895
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.3373,
      "step": 6900
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.2797,
      "step": 6905
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3495,
      "step": 6910
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.297,
      "step": 6915
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2722,
      "step": 6920
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2851,
      "step": 6925
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2873,
      "step": 6930
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3203,
      "step": 6935
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.3418,
      "step": 6940
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2656,
      "step": 6945
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2917,
      "step": 6950
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3157,
      "step": 6955
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3485,
      "step": 6960
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3457,
      "step": 6965
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3021,
      "step": 6970
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.3554,
      "step": 6975
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.214,
      "step": 6980
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2834,
      "step": 6985
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2508,
      "step": 6990
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3555,
      "step": 6995
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2949,
      "step": 7000
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3022,
      "step": 7005
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2868,
      "step": 7010
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3297,
      "step": 7015
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.257,
      "step": 7020
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2424,
      "step": 7025
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2686,
      "step": 7030
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2936,
      "step": 7035
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3123,
      "step": 7040
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2571,
      "step": 7045
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2647,
      "step": 7050
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.304,
      "step": 7055
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2541,
      "step": 7060
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.354,
      "step": 7065
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.336,
      "step": 7070
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3124,
      "step": 7075
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.3281,
      "step": 7080
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3212,
      "step": 7085
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3671875,
      "learning_rate": 0.001,
      "loss": 2.3689,
      "step": 7090
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.2572,
      "step": 7095
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.335,
      "step": 7100
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2974,
      "step": 7105
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.3526,
      "step": 7110
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2890625,
      "learning_rate": 0.001,
      "loss": 2.3439,
      "step": 7115
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3026,
      "step": 7120
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3288,
      "step": 7125
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29296875,
      "learning_rate": 0.001,
      "loss": 2.2952,
      "step": 7130
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2942,
      "step": 7135
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2384,
      "step": 7140
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2436,
      "step": 7145
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3166,
      "step": 7150
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2552,
      "step": 7155
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2395,
      "step": 7160
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2334,
      "step": 7165
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3261,
      "step": 7170
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2233,
      "step": 7175
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3239,
      "step": 7180
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.208,
      "step": 7185
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2938,
      "step": 7190
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3055,
      "step": 7195
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2853,
      "step": 7200
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.2827,
      "step": 7205
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.390625,
      "learning_rate": 0.001,
      "loss": 2.234,
      "step": 7210
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2621,
      "step": 7215
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2557,
      "step": 7220
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.1625,
      "step": 7225
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2853,
      "step": 7230
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1697,
      "step": 7235
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3439,
      "step": 7240
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2568,
      "step": 7245
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2698,
      "step": 7250
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3522,
      "step": 7255
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2786,
      "step": 7260
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.2758,
      "step": 7265
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2267,
      "step": 7270
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2619,
      "step": 7275
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.235,
      "step": 7280
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.1946,
      "step": 7285
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3306,
      "step": 7290
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3515625,
      "learning_rate": 0.001,
      "loss": 2.3197,
      "step": 7295
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3484,
      "step": 7300
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2644,
      "step": 7305
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3429,
      "step": 7310
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.248,
      "step": 7315
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2096,
      "step": 7320
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.376953125,
      "learning_rate": 0.001,
      "loss": 2.1851,
      "step": 7325
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2935,
      "step": 7330
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3234,
      "step": 7335
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2555,
      "step": 7340
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2168,
      "step": 7345
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.3627,
      "step": 7350
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2858,
      "step": 7355
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.3002,
      "step": 7360
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.2817,
      "step": 7365
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2995,
      "step": 7370
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2896,
      "step": 7375
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2691,
      "step": 7380
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.3571,
      "step": 7385
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2753,
      "step": 7390
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.1712,
      "step": 7395
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3137,
      "step": 7400
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3335,
      "step": 7405
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.1866,
      "step": 7410
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.219,
      "step": 7415
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3086,
      "step": 7420
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.2989,
      "step": 7425
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2209,
      "step": 7430
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2487,
      "step": 7435
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2269,
      "step": 7440
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.287109375,
      "learning_rate": 0.001,
      "loss": 2.2807,
      "step": 7445
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3287,
      "step": 7450
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2568,
      "step": 7455
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2662,
      "step": 7460
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.2783,
      "step": 7465
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.239,
      "step": 7470
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2978,
      "step": 7475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3329,
      "step": 7480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.1576,
      "step": 7485
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2521,
      "step": 7490
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2536,
      "step": 7495
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2902,
      "step": 7500
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2477,
      "step": 7505
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2292,
      "step": 7510
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.2579,
      "step": 7515
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2531,
      "step": 7520
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.3611,
      "step": 7525
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2527,
      "step": 7530
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3296,
      "step": 7535
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2596,
      "step": 7540
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2396,
      "step": 7545
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2573,
      "step": 7550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3043,
      "step": 7555
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2121,
      "step": 7560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3462,
      "step": 7565
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.287109375,
      "learning_rate": 0.001,
      "loss": 2.3078,
      "step": 7570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.316,
      "step": 7575
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.292,
      "step": 7580
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.3084,
      "step": 7585
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.235,
      "step": 7590
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2572,
      "step": 7595
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2612,
      "step": 7600
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2995,
      "step": 7605
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2694,
      "step": 7610
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2282,
      "step": 7615
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2096,
      "step": 7620
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2345,
      "step": 7625
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.2535,
      "step": 7630
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2774,
      "step": 7635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2084,
      "step": 7640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2479,
      "step": 7645
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2153,
      "step": 7650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2901,
      "step": 7655
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2653,
      "step": 7660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2731,
      "step": 7665
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.1943,
      "step": 7670
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.1499,
      "step": 7675
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2693,
      "step": 7680
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.2261,
      "step": 7685
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.1544,
      "step": 7690
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2456,
      "step": 7695
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2612,
      "step": 7700
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2848,
      "step": 7705
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2257,
      "step": 7710
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2658,
      "step": 7715
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2202,
      "step": 7720
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2599,
      "step": 7725
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.369140625,
      "learning_rate": 0.001,
      "loss": 2.3019,
      "step": 7730
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.3055,
      "step": 7735
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.1914,
      "step": 7740
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2936,
      "step": 7745
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2759,
      "step": 7750
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2421,
      "step": 7755
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3079,
      "step": 7760
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2364,
      "step": 7765
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2714,
      "step": 7770
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2086,
      "step": 7775
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2275,
      "step": 7780
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3385,
      "step": 7785
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2574,
      "step": 7790
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2743,
      "step": 7795
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2457,
      "step": 7800
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.3624,
      "step": 7805
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2603,
      "step": 7810
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2317,
      "step": 7815
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2466,
      "step": 7820
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.3946,
      "step": 7825
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2815,
      "step": 7830
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2996,
      "step": 7835
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2776,
      "step": 7840
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.184,
      "step": 7845
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2714,
      "step": 7850
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2512,
      "step": 7855
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2536,
      "step": 7860
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.1873,
      "step": 7865
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.337890625,
      "learning_rate": 0.001,
      "loss": 2.2324,
      "step": 7870
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.3213,
      "step": 7875
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2541,
      "step": 7880
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2901,
      "step": 7885
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3655,
      "step": 7890
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.1354,
      "step": 7895
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2691,
      "step": 7900
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.159,
      "step": 7905
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.1469,
      "step": 7910
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2344,
      "step": 7915
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2182,
      "step": 7920
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.3306,
      "step": 7925
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.1549,
      "step": 7930
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3377,
      "step": 7935
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2853,
      "step": 7940
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2024,
      "step": 7945
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2813,
      "step": 7950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.3001,
      "step": 7955
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2503,
      "step": 7960
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.1995,
      "step": 7965
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2804,
      "step": 7970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2591,
      "step": 7975
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2143,
      "step": 7980
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2275,
      "step": 7985
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3919,
      "step": 7990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.309,
      "step": 7995
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.2363,
      "step": 8000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.3152,
      "step": 8005
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2732,
      "step": 8010
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2561,
      "step": 8015
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.2615,
      "step": 8020
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2131,
      "step": 8025
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2894,
      "step": 8030
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3007,
      "step": 8035
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2376,
      "step": 8040
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3124,
      "step": 8045
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.1713,
      "step": 8050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3618,
      "step": 8055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3386,
      "step": 8060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2925,
      "step": 8065
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2123,
      "step": 8070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2016,
      "step": 8075
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2361,
      "step": 8080
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2677,
      "step": 8085
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2364,
      "step": 8090
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2845,
      "step": 8095
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2399,
      "step": 8100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.2638,
      "step": 8105
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2932,
      "step": 8110
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2401,
      "step": 8115
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2824,
      "step": 8120
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35546875,
      "learning_rate": 0.001,
      "loss": 2.2077,
      "step": 8125
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2046,
      "step": 8130
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.1893,
      "step": 8135
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2769,
      "step": 8140
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.217,
      "step": 8145
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2389,
      "step": 8150
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.266,
      "step": 8155
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2453,
      "step": 8160
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.281,
      "step": 8165
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2076,
      "step": 8170
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.1804,
      "step": 8175
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2309,
      "step": 8180
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2703,
      "step": 8185
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.3598,
      "step": 8190
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2038,
      "step": 8195
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2529,
      "step": 8200
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2816,
      "step": 8205
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.238,
      "step": 8210
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2579,
      "step": 8215
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.1992,
      "step": 8220
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.1818,
      "step": 8225
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.1506,
      "step": 8230
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2157,
      "step": 8235
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2456,
      "step": 8240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2597,
      "step": 8245
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2447,
      "step": 8250
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.1631,
      "step": 8255
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2394,
      "step": 8260
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2786,
      "step": 8265
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2592,
      "step": 8270
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2017,
      "step": 8275
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2654,
      "step": 8280
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2704,
      "step": 8285
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34375,
      "learning_rate": 0.001,
      "loss": 2.3055,
      "step": 8290
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.1985,
      "step": 8295
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.259,
      "step": 8300
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.276,
      "step": 8305
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1859,
      "step": 8310
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2696,
      "step": 8315
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2577,
      "step": 8320
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2533,
      "step": 8325
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.3143,
      "step": 8330
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2516,
      "step": 8335
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2624,
      "step": 8340
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2890625,
      "learning_rate": 0.001,
      "loss": 2.1764,
      "step": 8345
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2136,
      "step": 8350
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2297,
      "step": 8355
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2867,
      "step": 8360
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2896,
      "step": 8365
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2175,
      "step": 8370
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.19,
      "step": 8375
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.134,
      "step": 8380
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2358,
      "step": 8385
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.287109375,
      "learning_rate": 0.001,
      "loss": 2.1986,
      "step": 8390
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.203,
      "step": 8395
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.287109375,
      "learning_rate": 0.001,
      "loss": 2.2181,
      "step": 8400
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2655,
      "step": 8405
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2862,
      "step": 8410
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.3058,
      "step": 8415
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2694,
      "step": 8420
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2549,
      "step": 8425
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34765625,
      "learning_rate": 0.001,
      "loss": 2.2976,
      "step": 8430
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2406,
      "step": 8435
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2485,
      "step": 8440
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1618,
      "step": 8445
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.2759,
      "step": 8450
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2618,
      "step": 8455
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.203,
      "step": 8460
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.1972,
      "step": 8465
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.1704,
      "step": 8470
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2045,
      "step": 8475
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.188,
      "step": 8480
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2592,
      "step": 8485
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2913,
      "step": 8490
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.2592,
      "step": 8495
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2068,
      "step": 8500
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2432,
      "step": 8505
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.1883,
      "step": 8510
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2438,
      "step": 8515
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.1702,
      "step": 8520
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2469,
      "step": 8525
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3203125,
      "learning_rate": 0.001,
      "loss": 2.1156,
      "step": 8530
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.1934,
      "step": 8535
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.3777,
      "step": 8540
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2576,
      "step": 8545
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2462,
      "step": 8550
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.2936,
      "step": 8555
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3116,
      "step": 8560
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2689,
      "step": 8565
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36328125,
      "learning_rate": 0.001,
      "loss": 2.3067,
      "step": 8570
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.3342,
      "step": 8575
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2079,
      "step": 8580
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33984375,
      "learning_rate": 0.001,
      "loss": 2.243,
      "step": 8585
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2,
      "step": 8590
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2890625,
      "learning_rate": 0.001,
      "loss": 2.2437,
      "step": 8595
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.1937,
      "step": 8600
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2336,
      "step": 8605
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.314453125,
      "learning_rate": 0.001,
      "loss": 2.2758,
      "step": 8610
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.345703125,
      "learning_rate": 0.001,
      "loss": 2.2071,
      "step": 8615
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.149,
      "step": 8620
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.257,
      "step": 8625
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.207,
      "step": 8630
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2858,
      "step": 8635
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.328125,
      "learning_rate": 0.001,
      "loss": 2.2318,
      "step": 8640
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.295,
      "step": 8645
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.1429,
      "step": 8650
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.1635,
      "step": 8655
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.3268,
      "step": 8660
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2371,
      "step": 8665
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2604,
      "step": 8670
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2545,
      "step": 8675
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.1592,
      "step": 8680
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2776,
      "step": 8685
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.213,
      "step": 8690
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.287109375,
      "learning_rate": 0.001,
      "loss": 2.1203,
      "step": 8695
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3063,
      "step": 8700
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2479,
      "step": 8705
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2708,
      "step": 8710
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.326171875,
      "learning_rate": 0.001,
      "loss": 2.308,
      "step": 8715
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2732,
      "step": 8720
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.2556,
      "step": 8725
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3359375,
      "learning_rate": 0.001,
      "loss": 2.2082,
      "step": 8730
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2652,
      "step": 8735
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.296875,
      "learning_rate": 0.001,
      "loss": 2.3049,
      "step": 8740
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2438,
      "step": 8745
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.1741,
      "step": 8750
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2807,
      "step": 8755
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2431,
      "step": 8760
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2556,
      "step": 8765
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2498,
      "step": 8770
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2788,
      "step": 8775
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29296875,
      "learning_rate": 0.001,
      "loss": 2.2636,
      "step": 8780
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.330078125,
      "learning_rate": 0.001,
      "loss": 2.2968,
      "step": 8785
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2984,
      "step": 8790
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2495,
      "step": 8795
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.2131,
      "step": 8800
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33203125,
      "learning_rate": 0.001,
      "loss": 2.176,
      "step": 8805
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3043,
      "step": 8810
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.2409,
      "step": 8815
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.1861,
      "step": 8820
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.2472,
      "step": 8825
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.306640625,
      "learning_rate": 0.001,
      "loss": 2.2569,
      "step": 8830
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32421875,
      "learning_rate": 0.001,
      "loss": 2.2432,
      "step": 8835
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2889,
      "step": 8840
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1974,
      "step": 8845
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.3096,
      "step": 8850
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.1649,
      "step": 8855
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.298828125,
      "learning_rate": 0.001,
      "loss": 2.2163,
      "step": 8860
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.302734375,
      "learning_rate": 0.001,
      "loss": 2.2611,
      "step": 8865
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.245,
      "step": 8870
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30078125,
      "learning_rate": 0.001,
      "loss": 2.1904,
      "step": 8875
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.310546875,
      "learning_rate": 0.001,
      "loss": 2.2698,
      "step": 8880
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3046875,
      "learning_rate": 0.001,
      "loss": 2.2447,
      "step": 8885
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.349609375,
      "learning_rate": 0.001,
      "loss": 2.2229,
      "step": 8890
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.1953,
      "step": 8895
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.2675,
      "step": 8900
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.322265625,
      "learning_rate": 0.001,
      "loss": 2.3104,
      "step": 8905
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31640625,
      "learning_rate": 0.001,
      "loss": 2.3058,
      "step": 8910
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.294921875,
      "learning_rate": 0.001,
      "loss": 2.1981,
      "step": 8915
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1727,
      "step": 8920
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.1917,
      "step": 8925
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.291015625,
      "learning_rate": 0.001,
      "loss": 2.1705,
      "step": 8930
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3125,
      "learning_rate": 0.001,
      "loss": 2.2193,
      "step": 8935
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2683,
      "step": 8940
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.1673,
      "step": 8945
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2890625,
      "learning_rate": 0.001,
      "loss": 2.2151,
      "step": 8950
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.341796875,
      "learning_rate": 0.001,
      "loss": 2.2106,
      "step": 8955
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.318359375,
      "learning_rate": 0.001,
      "loss": 2.2439,
      "step": 8960
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30859375,
      "learning_rate": 0.001,
      "loss": 2.2279,
      "step": 8965
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.2348995208740234,
      "eval_runtime": 2639.8137,
      "eval_samples_per_second": 6.016,
      "eval_steps_per_second": 0.752,
      "step": 8969
    },
    {
      "epoch": 1.0,
      "step": 8969,
      "total_flos": 1.50079627163861e+17,
      "train_loss": 2.666482654486044,
      "train_runtime": 30048.5081,
      "train_samples_per_second": 4.775,
      "train_steps_per_second": 0.298
    }
  ],
  "logging_steps": 5,
  "max_steps": 8969,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 1.50079627163861e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}