{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002,
      "grad_norm": 3.1993721005818543e-09,
      "learning_rate": 9.984e-06,
      "loss": 0.027,
      "step": 10
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.0,
      "learning_rate": 9.966e-06,
      "loss": 2.1595,
      "step": 20
    },
    {
      "epoch": 0.006,
      "grad_norm": 0.006726053077727556,
      "learning_rate": 9.946000000000001e-06,
      "loss": 0.0,
      "step": 30
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.0,
      "learning_rate": 9.928e-06,
      "loss": 2.6688,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1408715838333592e-05,
      "learning_rate": 9.908e-06,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 0.012,
      "grad_norm": 7.529888934598716e-10,
      "learning_rate": 9.888000000000001e-06,
      "loss": 0.0,
      "step": 60
    },
    {
      "epoch": 0.014,
      "grad_norm": NaN,
      "learning_rate": 9.872e-06,
      "loss": 2.55,
      "step": 70
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.0,
      "learning_rate": 9.852e-06,
      "loss": 1.2953,
      "step": 80
    },
    {
      "epoch": 0.018,
      "grad_norm": 9.098942399532461e-08,
      "learning_rate": 9.832e-06,
      "loss": 0.0848,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 608.9567260742188,
      "learning_rate": 9.812e-06,
      "loss": 2.9031,
      "step": 100
    },
    {
      "epoch": 0.022,
      "grad_norm": 260.31805419921875,
      "learning_rate": 9.792e-06,
      "loss": 0.1289,
      "step": 110
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.005177302286028862,
      "learning_rate": 9.772e-06,
      "loss": 0.522,
      "step": 120
    },
    {
      "epoch": 0.026,
      "grad_norm": 0.0,
      "learning_rate": 9.752e-06,
      "loss": 0.0,
      "step": 130
    },
    {
      "epoch": 0.028,
      "grad_norm": 1.4419997418357866e-09,
      "learning_rate": 9.732000000000001e-06,
      "loss": 0.0,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.03419796586968e-05,
      "learning_rate": 9.712e-06,
      "loss": 0.0,
      "step": 150
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.0,
      "learning_rate": 9.692e-06,
      "loss": 0.0,
      "step": 160
    },
    {
      "epoch": 0.034,
      "grad_norm": 0.0,
      "learning_rate": 9.672e-06,
      "loss": 0.0,
      "step": 170
    },
    {
      "epoch": 0.036,
      "grad_norm": 8.001513833733043e-07,
      "learning_rate": 9.652e-06,
      "loss": 0.0,
      "step": 180
    },
    {
      "epoch": 0.038,
      "grad_norm": 0.0,
      "learning_rate": 9.632e-06,
      "loss": 0.0007,
      "step": 190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.005184683483093977,
      "learning_rate": 9.612000000000002e-06,
      "loss": 0.0009,
      "step": 200
    },
    {
      "epoch": 0.042,
      "grad_norm": 0.0,
      "learning_rate": 9.592e-06,
      "loss": 0.0,
      "step": 210
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.0,
      "learning_rate": 9.572000000000001e-06,
      "loss": 0.0005,
      "step": 220
    },
    {
      "epoch": 0.046,
      "grad_norm": 0.0,
      "learning_rate": 9.552000000000001e-06,
      "loss": 0.0007,
      "step": 230
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.0,
      "learning_rate": 9.532000000000001e-06,
      "loss": 0.0,
      "step": 240
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.448915999295423e-06,
      "learning_rate": 9.512000000000001e-06,
      "loss": 0.0,
      "step": 250
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.0,
      "learning_rate": 9.492000000000002e-06,
      "loss": 0.0018,
      "step": 260
    },
    {
      "epoch": 0.054,
      "grad_norm": 0.0,
      "learning_rate": 9.472000000000002e-06,
      "loss": 0.0,
      "step": 270
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.0,
      "learning_rate": 9.452000000000002e-06,
      "loss": 0.0,
      "step": 280
    },
    {
      "epoch": 0.058,
      "grad_norm": 0.0,
      "learning_rate": 9.432e-06,
      "loss": 0.9313,
      "step": 290
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.0,
      "learning_rate": 9.412e-06,
      "loss": 0.0,
      "step": 300
    },
    {
      "epoch": 0.062,
      "grad_norm": 0.0,
      "learning_rate": 9.392000000000001e-06,
      "loss": 0.0,
      "step": 310
    },
    {
      "epoch": 0.064,
      "grad_norm": 2.930929213107447e-06,
      "learning_rate": 9.372000000000001e-06,
      "loss": 0.0,
      "step": 320
    },
    {
      "epoch": 0.066,
      "grad_norm": 0.0,
      "learning_rate": 9.352000000000001e-06,
      "loss": 0.0,
      "step": 330
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.0,
      "learning_rate": 9.332000000000001e-06,
      "loss": 0.0,
      "step": 340
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.370261758092965e-07,
      "learning_rate": 9.312000000000002e-06,
      "loss": 0.0,
      "step": 350
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.0,
      "learning_rate": 9.292000000000002e-06,
      "loss": 0.0001,
      "step": 360
    },
    {
      "epoch": 0.074,
      "grad_norm": 0.0,
      "learning_rate": 9.272e-06,
      "loss": 0.0,
      "step": 370
    },
    {
      "epoch": 0.076,
      "grad_norm": 3.5792377275356557e-06,
      "learning_rate": 9.252e-06,
      "loss": 0.0,
      "step": 380
    },
    {
      "epoch": 0.078,
      "grad_norm": 0.0,
      "learning_rate": 9.232e-06,
      "loss": 0.0,
      "step": 390
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.0,
      "learning_rate": 9.212000000000001e-06,
      "loss": 0.0,
      "step": 400
    },
    {
      "epoch": 0.082,
      "grad_norm": 5.986302009652889e-10,
      "learning_rate": 9.192000000000001e-06,
      "loss": 0.0019,
      "step": 410
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.0,
      "learning_rate": 9.172000000000001e-06,
      "loss": 0.0,
      "step": 420
    },
    {
      "epoch": 0.086,
      "grad_norm": 0.0,
      "learning_rate": 9.152000000000001e-06,
      "loss": 0.0037,
      "step": 430
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.0,
      "learning_rate": 9.132000000000002e-06,
      "loss": 0.0,
      "step": 440
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6021938620269793e-07,
      "learning_rate": 9.112e-06,
      "loss": 0.0,
      "step": 450
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.0,
      "learning_rate": 9.092e-06,
      "loss": 0.0001,
      "step": 460
    },
    {
      "epoch": 0.094,
      "grad_norm": 0.0,
      "learning_rate": 9.072e-06,
      "loss": 0.0,
      "step": 470
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.0,
      "learning_rate": 9.052e-06,
      "loss": 0.0,
      "step": 480
    },
    {
      "epoch": 0.098,
      "grad_norm": 0.0,
      "learning_rate": 9.032000000000001e-06,
      "loss": 0.0,
      "step": 490
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.0,
      "learning_rate": 9.012000000000001e-06,
      "loss": 0.0,
      "step": 500
    },
    {
      "epoch": 0.102,
      "grad_norm": 0.0,
      "learning_rate": 8.992000000000001e-06,
      "loss": 0.0,
      "step": 510
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.0,
      "learning_rate": 8.972000000000002e-06,
      "loss": 0.0,
      "step": 520
    },
    {
      "epoch": 0.106,
      "grad_norm": 0.0,
      "learning_rate": 8.952e-06,
      "loss": 0.0,
      "step": 530
    },
    {
      "epoch": 0.108,
      "grad_norm": 0.0,
      "learning_rate": 8.932e-06,
      "loss": 0.0,
      "step": 540
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.0,
      "learning_rate": 8.912e-06,
      "loss": 0.0,
      "step": 550
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.0,
      "learning_rate": 8.892e-06,
      "loss": 0.0,
      "step": 560
    },
    {
      "epoch": 0.114,
      "grad_norm": 1.527065978734754e-05,
      "learning_rate": 8.872e-06,
      "loss": 0.0,
      "step": 570
    },
    {
      "epoch": 0.116,
      "grad_norm": 0.0,
      "learning_rate": 8.852000000000001e-06,
      "loss": 0.0,
      "step": 580
    },
    {
      "epoch": 0.118,
      "grad_norm": 0.0,
      "learning_rate": 8.832000000000001e-06,
      "loss": 0.0,
      "step": 590
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.0,
      "learning_rate": 8.812000000000001e-06,
      "loss": 0.0,
      "step": 600
    },
    {
      "epoch": 0.122,
      "grad_norm": 0.0,
      "learning_rate": 8.792e-06,
      "loss": 0.0,
      "step": 610
    },
    {
      "epoch": 0.124,
      "grad_norm": 0.0,
      "learning_rate": 8.772e-06,
      "loss": 0.0,
      "step": 620
    },
    {
      "epoch": 0.126,
      "grad_norm": 0.0,
      "learning_rate": 8.752e-06,
      "loss": 0.0,
      "step": 630
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.0,
      "learning_rate": 8.732e-06,
      "loss": 0.0,
      "step": 640
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.0,
      "learning_rate": 8.712e-06,
      "loss": 0.0,
      "step": 650
    },
    {
      "epoch": 0.132,
      "grad_norm": 0.0,
      "learning_rate": 8.692e-06,
      "loss": 0.0,
      "step": 660
    },
    {
      "epoch": 0.134,
      "grad_norm": 0.0,
      "learning_rate": 8.672000000000001e-06,
      "loss": 0.0,
      "step": 670
    },
    {
      "epoch": 0.136,
      "grad_norm": 0.0,
      "learning_rate": 8.652000000000001e-06,
      "loss": 0.0,
      "step": 680
    },
    {
      "epoch": 0.138,
      "grad_norm": 3.734453173365182e-07,
      "learning_rate": 8.632e-06,
      "loss": 0.0,
      "step": 690
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.0,
      "learning_rate": 8.612e-06,
      "loss": 0.0,
      "step": 700
    },
    {
      "epoch": 0.142,
      "grad_norm": 0.0,
      "learning_rate": 8.592e-06,
      "loss": 0.0,
      "step": 710
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.0,
      "learning_rate": 8.572e-06,
      "loss": 0.0,
      "step": 720
    },
    {
      "epoch": 0.146,
      "grad_norm": 0.0,
      "learning_rate": 8.552e-06,
      "loss": 0.0,
      "step": 730
    },
    {
      "epoch": 0.148,
      "grad_norm": 5.163356964210664e-10,
      "learning_rate": 8.532e-06,
      "loss": 0.0,
      "step": 740
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.0,
      "learning_rate": 8.512e-06,
      "loss": 0.0,
      "step": 750
    },
    {
      "epoch": 0.152,
      "grad_norm": 0.0,
      "learning_rate": 8.492000000000001e-06,
      "loss": 0.0,
      "step": 760
    },
    {
      "epoch": 0.154,
      "grad_norm": 0.0,
      "learning_rate": 8.472e-06,
      "loss": 0.0,
      "step": 770
    },
    {
      "epoch": 0.156,
      "grad_norm": 0.0,
      "learning_rate": 8.452e-06,
      "loss": 0.0,
      "step": 780
    },
    {
      "epoch": 0.158,
      "grad_norm": 0.0,
      "learning_rate": 8.432e-06,
      "loss": 0.0,
      "step": 790
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.0,
      "learning_rate": 8.412e-06,
      "loss": 0.0,
      "step": 800
    },
    {
      "epoch": 0.162,
      "grad_norm": 0.0,
      "learning_rate": 8.392e-06,
      "loss": 0.0,
      "step": 810
    },
    {
      "epoch": 0.164,
      "grad_norm": 0.0,
      "learning_rate": 8.372e-06,
      "loss": 0.0,
      "step": 820
    },
    {
      "epoch": 0.166,
      "grad_norm": 0.0,
      "learning_rate": 8.352e-06,
      "loss": 0.0,
      "step": 830
    },
    {
      "epoch": 0.168,
      "grad_norm": 0.0,
      "learning_rate": 8.332000000000001e-06,
      "loss": 0.0,
      "step": 840
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.0,
      "learning_rate": 8.312000000000001e-06,
      "loss": 0.0,
      "step": 850
    },
    {
      "epoch": 0.172,
      "grad_norm": 0.0,
      "learning_rate": 8.292000000000001e-06,
      "loss": 0.0,
      "step": 860
    },
    {
      "epoch": 0.174,
      "grad_norm": 1.004443447527592e-06,
      "learning_rate": 8.272000000000001e-06,
      "loss": 0.0,
      "step": 870
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.0,
      "learning_rate": 8.252000000000002e-06,
      "loss": 0.0,
      "step": 880
    },
    {
      "epoch": 0.178,
      "grad_norm": 0.08667419105768204,
      "learning_rate": 8.232000000000002e-06,
      "loss": 0.0,
      "step": 890
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.0,
      "learning_rate": 8.212e-06,
      "loss": 0.0,
      "step": 900
    },
    {
      "epoch": 0.182,
      "grad_norm": 0.0,
      "learning_rate": 8.192e-06,
      "loss": 0.0,
      "step": 910
    },
    {
      "epoch": 0.184,
      "grad_norm": 0.0,
      "learning_rate": 8.172e-06,
      "loss": 0.0,
      "step": 920
    },
    {
      "epoch": 0.186,
      "grad_norm": 0.0,
      "learning_rate": 8.152000000000001e-06,
      "loss": 0.0,
      "step": 930
    },
    {
      "epoch": 0.188,
      "grad_norm": 0.0,
      "learning_rate": 8.132000000000001e-06,
      "loss": 0.0,
      "step": 940
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.0,
      "learning_rate": 8.112000000000001e-06,
      "loss": 0.0,
      "step": 950
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.0,
      "learning_rate": 8.092000000000001e-06,
      "loss": 0.0,
      "step": 960
    },
    {
      "epoch": 0.194,
      "grad_norm": 0.0,
      "learning_rate": 8.072000000000002e-06,
      "loss": 0.0,
      "step": 970
    },
    {
      "epoch": 0.196,
      "grad_norm": 0.0,
      "learning_rate": 8.052e-06,
      "loss": 0.0,
      "step": 980
    },
    {
      "epoch": 0.198,
      "grad_norm": 5.12321907120139e-10,
      "learning_rate": 8.032e-06,
      "loss": 0.0,
      "step": 990
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.0,
      "learning_rate": 8.012e-06,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 0.202,
      "grad_norm": 0.0,
      "learning_rate": 7.992e-06,
      "loss": 0.0,
      "step": 1010
    },
    {
      "epoch": 0.204,
      "grad_norm": 0.0,
      "learning_rate": 7.972000000000001e-06,
      "loss": 0.0,
      "step": 1020
    },
    {
      "epoch": 0.206,
      "grad_norm": 0.0,
      "learning_rate": 7.952000000000001e-06,
      "loss": 0.0,
      "step": 1030
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.0,
      "learning_rate": 7.932000000000001e-06,
      "loss": 0.0,
      "step": 1040
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.0,
      "learning_rate": 7.912000000000001e-06,
      "loss": 0.0,
      "step": 1050
    },
    {
      "epoch": 0.212,
      "grad_norm": 0.0,
      "learning_rate": 7.892e-06,
      "loss": 0.0,
      "step": 1060
    },
    {
      "epoch": 0.214,
      "grad_norm": 0.0,
      "learning_rate": 7.872e-06,
      "loss": 0.0,
      "step": 1070
    },
    {
      "epoch": 0.216,
      "grad_norm": 0.0,
      "learning_rate": 7.852e-06,
      "loss": 0.0,
      "step": 1080
    },
    {
      "epoch": 0.218,
      "grad_norm": 0.0,
      "learning_rate": 7.832e-06,
      "loss": 0.0,
      "step": 1090
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2012708339170786e-07,
      "learning_rate": 7.812e-06,
      "loss": 0.0,
      "step": 1100
    },
    {
      "epoch": 0.222,
      "grad_norm": 0.0,
      "learning_rate": 7.792000000000001e-06,
      "loss": 0.0,
      "step": 1110
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.0,
      "learning_rate": 7.772000000000001e-06,
      "loss": 0.0,
      "step": 1120
    },
    {
      "epoch": 0.226,
      "grad_norm": 0.0,
      "learning_rate": 7.752000000000001e-06,
      "loss": 0.0,
      "step": 1130
    },
    {
      "epoch": 0.228,
      "grad_norm": 8.950109418037755e-07,
      "learning_rate": 7.732e-06,
      "loss": 0.0,
      "step": 1140
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.0,
      "learning_rate": 7.712e-06,
      "loss": 0.0,
      "step": 1150
    },
    {
      "epoch": 0.232,
      "grad_norm": 0.0,
      "learning_rate": 7.692e-06,
      "loss": 0.0,
      "step": 1160
    },
    {
      "epoch": 0.234,
      "grad_norm": 0.0,
      "learning_rate": 7.672e-06,
      "loss": 0.0,
      "step": 1170
    },
    {
      "epoch": 0.236,
      "grad_norm": 0.0,
      "learning_rate": 7.652e-06,
      "loss": 0.0,
      "step": 1180
    },
    {
      "epoch": 0.238,
      "grad_norm": 0.0,
      "learning_rate": 7.632e-06,
      "loss": 0.0,
      "step": 1190
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.0,
      "learning_rate": 7.612e-06,
      "loss": 0.0,
      "step": 1200
    },
    {
      "epoch": 0.242,
      "grad_norm": 0.0,
      "learning_rate": 7.592e-06,
      "loss": 0.0,
      "step": 1210
    },
    {
      "epoch": 0.244,
      "grad_norm": 0.0,
      "learning_rate": 7.5720000000000005e-06,
      "loss": 0.0,
      "step": 1220
    },
    {
      "epoch": 0.246,
      "grad_norm": 0.0,
      "learning_rate": 7.552000000000001e-06,
      "loss": 0.0,
      "step": 1230
    },
    {
      "epoch": 0.248,
      "grad_norm": 0.0,
      "learning_rate": 7.532e-06,
      "loss": 0.0,
      "step": 1240
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.0,
      "learning_rate": 7.512e-06,
      "loss": 0.0,
      "step": 1250
    },
    {
      "epoch": 0.252,
      "grad_norm": 0.0,
      "learning_rate": 7.4920000000000004e-06,
      "loss": 0.0,
      "step": 1260
    },
    {
      "epoch": 0.254,
      "grad_norm": 0.0,
      "learning_rate": 7.472000000000001e-06,
      "loss": 0.0,
      "step": 1270
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.0,
      "learning_rate": 7.452e-06,
      "loss": 0.0,
      "step": 1280
    },
    {
      "epoch": 0.258,
      "grad_norm": 0.0,
      "learning_rate": 7.432e-06,
      "loss": 0.0,
      "step": 1290
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.0,
      "learning_rate": 7.412e-06,
      "loss": 0.0,
      "step": 1300
    },
    {
      "epoch": 0.262,
      "grad_norm": 0.0,
      "learning_rate": 7.3920000000000005e-06,
      "loss": 0.0,
      "step": 1310
    },
    {
      "epoch": 0.264,
      "grad_norm": 0.0,
      "learning_rate": 7.372e-06,
      "loss": 0.0,
      "step": 1320
    },
    {
      "epoch": 0.266,
      "grad_norm": 0.0,
      "learning_rate": 7.352e-06,
      "loss": 0.0,
      "step": 1330
    },
    {
      "epoch": 0.268,
      "grad_norm": 0.0,
      "learning_rate": 7.332e-06,
      "loss": 0.0,
      "step": 1340
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.134319636113105e-10,
      "learning_rate": 7.3120000000000005e-06,
      "loss": 0.0,
      "step": 1350
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.0,
      "learning_rate": 7.292e-06,
      "loss": 0.0,
      "step": 1360
    },
    {
      "epoch": 0.274,
      "grad_norm": 0.0,
      "learning_rate": 7.272e-06,
      "loss": 0.0,
      "step": 1370
    },
    {
      "epoch": 0.276,
      "grad_norm": 2.098096363312152e-09,
      "learning_rate": 7.252e-06,
      "loss": 0.0,
      "step": 1380
    },
    {
      "epoch": 0.278,
      "grad_norm": 0.0,
      "learning_rate": 7.232e-06,
      "loss": 0.0,
      "step": 1390
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.007792504038661718,
      "learning_rate": 7.212e-06,
      "loss": 0.0,
      "step": 1400
    },
    {
      "epoch": 0.282,
      "grad_norm": 0.0,
      "learning_rate": 7.192e-06,
      "loss": 0.0,
      "step": 1410
    },
    {
      "epoch": 0.284,
      "grad_norm": 0.0,
      "learning_rate": 7.172e-06,
      "loss": 0.0,
      "step": 1420
    },
    {
      "epoch": 0.286,
      "grad_norm": 0.0,
      "learning_rate": 7.152e-06,
      "loss": 0.0,
      "step": 1430
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.0,
      "learning_rate": 7.132e-06,
      "loss": 0.0,
      "step": 1440
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.0,
      "learning_rate": 7.1120000000000015e-06,
      "loss": 0.0,
      "step": 1450
    },
    {
      "epoch": 0.292,
      "grad_norm": 0.0,
      "learning_rate": 7.092000000000001e-06,
      "loss": 0.0,
      "step": 1460
    },
    {
      "epoch": 0.294,
      "grad_norm": 0.0,
      "learning_rate": 7.072000000000001e-06,
      "loss": 0.0,
      "step": 1470
    },
    {
      "epoch": 0.296,
      "grad_norm": 0.0,
      "learning_rate": 7.052000000000001e-06,
      "loss": 0.0,
      "step": 1480
    },
    {
      "epoch": 0.298,
      "grad_norm": 1.3290431866153085e-07,
      "learning_rate": 7.0320000000000015e-06,
      "loss": 0.0,
      "step": 1490
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.0,
      "learning_rate": 7.012000000000001e-06,
      "loss": 0.0,
      "step": 1500
    },
    {
      "epoch": 0.302,
      "grad_norm": 0.0,
      "learning_rate": 6.992000000000001e-06,
      "loss": 0.0,
      "step": 1510
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.0,
      "learning_rate": 6.972000000000001e-06,
      "loss": 0.0,
      "step": 1520
    },
    {
      "epoch": 0.306,
      "grad_norm": 0.0,
      "learning_rate": 6.952000000000001e-06,
      "loss": 0.0,
      "step": 1530
    },
    {
      "epoch": 0.308,
      "grad_norm": 0.0,
      "learning_rate": 6.932000000000001e-06,
      "loss": 0.0,
      "step": 1540
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.0,
      "learning_rate": 6.912000000000001e-06,
      "loss": 0.0,
      "step": 1550
    },
    {
      "epoch": 0.312,
      "grad_norm": 0.0,
      "learning_rate": 6.892000000000001e-06,
      "loss": 0.0,
      "step": 1560
    },
    {
      "epoch": 0.314,
      "grad_norm": 0.0,
      "learning_rate": 6.872000000000001e-06,
      "loss": 0.0,
      "step": 1570
    },
    {
      "epoch": 0.316,
      "grad_norm": 0.0,
      "learning_rate": 6.852000000000001e-06,
      "loss": 0.0,
      "step": 1580
    },
    {
      "epoch": 0.318,
      "grad_norm": 0.0,
      "learning_rate": 6.832000000000001e-06,
      "loss": 0.0,
      "step": 1590
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.0,
      "learning_rate": 6.812000000000001e-06,
      "loss": 0.0,
      "step": 1600
    },
    {
      "epoch": 0.322,
      "grad_norm": 0.0,
      "learning_rate": 6.792000000000001e-06,
      "loss": 0.0,
      "step": 1610
    },
    {
      "epoch": 0.324,
      "grad_norm": 0.0,
      "learning_rate": 6.7720000000000006e-06,
      "loss": 0.0,
      "step": 1620
    },
    {
      "epoch": 0.326,
      "grad_norm": 0.0,
      "learning_rate": 6.752000000000001e-06,
      "loss": 0.0,
      "step": 1630
    },
    {
      "epoch": 0.328,
      "grad_norm": 0.0,
      "learning_rate": 6.732000000000001e-06,
      "loss": 0.0,
      "step": 1640
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.0,
      "learning_rate": 6.712000000000001e-06,
      "loss": 0.0,
      "step": 1650
    },
    {
      "epoch": 0.332,
      "grad_norm": 1.0833507531060604e-06,
      "learning_rate": 6.6920000000000005e-06,
      "loss": 0.0,
      "step": 1660
    },
    {
      "epoch": 0.334,
      "grad_norm": 0.0,
      "learning_rate": 6.672000000000001e-06,
      "loss": 0.0,
      "step": 1670
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.0,
      "learning_rate": 6.652000000000001e-06,
      "loss": 0.0,
      "step": 1680
    },
    {
      "epoch": 0.338,
      "grad_norm": 0.0,
      "learning_rate": 6.632000000000001e-06,
      "loss": 0.0,
      "step": 1690
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.0,
      "learning_rate": 6.612e-06,
      "loss": 0.0,
      "step": 1700
    },
    {
      "epoch": 0.342,
      "grad_norm": 0.0,
      "learning_rate": 6.592000000000001e-06,
      "loss": 0.0,
      "step": 1710
    },
    {
      "epoch": 0.344,
      "grad_norm": 0.0,
      "learning_rate": 6.572000000000001e-06,
      "loss": 0.0,
      "step": 1720
    },
    {
      "epoch": 0.346,
      "grad_norm": 0.0,
      "learning_rate": 6.552000000000001e-06,
      "loss": 0.0,
      "step": 1730
    },
    {
      "epoch": 0.348,
      "grad_norm": 0.0,
      "learning_rate": 6.532e-06,
      "loss": 0.0,
      "step": 1740
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.0,
      "learning_rate": 6.5120000000000005e-06,
      "loss": 0.0,
      "step": 1750
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.0,
      "learning_rate": 6.492000000000001e-06,
      "loss": 0.0,
      "step": 1760
    },
    {
      "epoch": 0.354,
      "grad_norm": 4.521739811025327e-06,
      "learning_rate": 6.472000000000001e-06,
      "loss": 0.0,
      "step": 1770
    },
    {
      "epoch": 0.356,
      "grad_norm": 0.0,
      "learning_rate": 6.452e-06,
      "loss": 0.0,
      "step": 1780
    },
    {
      "epoch": 0.358,
      "grad_norm": 0.0,
      "learning_rate": 6.432e-06,
      "loss": 0.0,
      "step": 1790
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.0,
      "learning_rate": 6.412000000000001e-06,
      "loss": 0.0,
      "step": 1800
    },
    {
      "epoch": 0.362,
      "grad_norm": 0.0,
      "learning_rate": 6.392000000000001e-06,
      "loss": 0.0,
      "step": 1810
    },
    {
      "epoch": 0.364,
      "grad_norm": 0.0,
      "learning_rate": 6.372e-06,
      "loss": 0.0,
      "step": 1820
    },
    {
      "epoch": 0.366,
      "grad_norm": 0.0,
      "learning_rate": 6.352e-06,
      "loss": 0.0,
      "step": 1830
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.0,
      "learning_rate": 6.3320000000000005e-06,
      "loss": 0.0,
      "step": 1840
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.0,
      "learning_rate": 6.312000000000001e-06,
      "loss": 0.0,
      "step": 1850
    },
    {
      "epoch": 0.372,
      "grad_norm": 0.0,
      "learning_rate": 6.292e-06,
      "loss": 0.0,
      "step": 1860
    },
    {
      "epoch": 0.374,
      "grad_norm": 0.0,
      "learning_rate": 6.272e-06,
      "loss": 0.0,
      "step": 1870
    },
    {
      "epoch": 0.376,
      "grad_norm": 0.0,
      "learning_rate": 6.2520000000000004e-06,
      "loss": 0.0,
      "step": 1880
    },
    {
      "epoch": 0.378,
      "grad_norm": 0.0,
      "learning_rate": 6.232000000000001e-06,
      "loss": 0.0,
      "step": 1890
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.0,
      "learning_rate": 6.212e-06,
      "loss": 0.0,
      "step": 1900
    },
    {
      "epoch": 0.382,
      "grad_norm": 0.0,
      "learning_rate": 6.192e-06,
      "loss": 0.0,
      "step": 1910
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.0,
      "learning_rate": 6.172e-06,
      "loss": 0.0,
      "step": 1920
    },
    {
      "epoch": 0.386,
      "grad_norm": 0.0,
      "learning_rate": 6.1520000000000006e-06,
      "loss": 0.0,
      "step": 1930
    },
    {
      "epoch": 0.388,
      "grad_norm": 0.0,
      "learning_rate": 6.132e-06,
      "loss": 0.0,
      "step": 1940
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.0,
      "learning_rate": 6.112e-06,
      "loss": 0.0,
      "step": 1950
    },
    {
      "epoch": 0.392,
      "grad_norm": 0.0,
      "learning_rate": 6.092e-06,
      "loss": 0.0,
      "step": 1960
    },
    {
      "epoch": 0.394,
      "grad_norm": 0.0,
      "learning_rate": 6.0720000000000005e-06,
      "loss": 0.0,
      "step": 1970
    },
    {
      "epoch": 0.396,
      "grad_norm": 0.0,
      "learning_rate": 6.052e-06,
      "loss": 0.0,
      "step": 1980
    },
    {
      "epoch": 0.398,
      "grad_norm": 5.1261139777381e-10,
      "learning_rate": 6.032e-06,
      "loss": 0.0,
      "step": 1990
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.0,
      "learning_rate": 6.012e-06,
      "loss": 0.0,
      "step": 2000
    },
    {
      "epoch": 0.402,
      "grad_norm": 0.0,
      "learning_rate": 5.992e-06,
      "loss": 0.0,
      "step": 2010
    },
    {
      "epoch": 0.404,
      "grad_norm": 0.0,
      "learning_rate": 5.972e-06,
      "loss": 0.0,
      "step": 2020
    },
    {
      "epoch": 0.406,
      "grad_norm": 0.0,
      "learning_rate": 5.952e-06,
      "loss": 0.0,
      "step": 2030
    },
    {
      "epoch": 0.408,
      "grad_norm": 0.0,
      "learning_rate": 5.932e-06,
      "loss": 0.0,
      "step": 2040
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.0,
      "learning_rate": 5.912e-06,
      "loss": 0.0,
      "step": 2050
    },
    {
      "epoch": 0.412,
      "grad_norm": 2.2953676737103024e-09,
      "learning_rate": 5.892e-06,
      "loss": 0.0,
      "step": 2060
    },
    {
      "epoch": 0.414,
      "grad_norm": 0.0,
      "learning_rate": 5.872000000000001e-06,
      "loss": 0.0,
      "step": 2070
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.0,
      "learning_rate": 5.852000000000001e-06,
      "loss": 0.0,
      "step": 2080
    },
    {
      "epoch": 0.418,
      "grad_norm": 3.5478933568811044e-07,
      "learning_rate": 5.832000000000001e-06,
      "loss": 0.0,
      "step": 2090
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.0,
      "learning_rate": 5.812000000000001e-06,
      "loss": 0.0,
      "step": 2100
    },
    {
      "epoch": 0.422,
      "grad_norm": 0.0,
      "learning_rate": 5.792000000000001e-06,
      "loss": 0.0,
      "step": 2110
    },
    {
      "epoch": 0.424,
      "grad_norm": 0.0,
      "learning_rate": 5.772000000000001e-06,
      "loss": 0.0,
      "step": 2120
    },
    {
      "epoch": 0.426,
      "grad_norm": 0.0,
      "learning_rate": 5.752000000000001e-06,
      "loss": 0.0,
      "step": 2130
    },
    {
      "epoch": 0.428,
      "grad_norm": 0.0,
      "learning_rate": 5.732000000000001e-06,
      "loss": 0.0,
      "step": 2140
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.0,
      "learning_rate": 5.7120000000000005e-06,
      "loss": 0.0,
      "step": 2150
    },
    {
      "epoch": 0.432,
      "grad_norm": 0.0,
      "learning_rate": 5.692000000000001e-06,
      "loss": 0.0,
      "step": 2160
    },
    {
      "epoch": 0.434,
      "grad_norm": 0.0,
      "learning_rate": 5.672000000000001e-06,
      "loss": 0.0,
      "step": 2170
    },
    {
      "epoch": 0.436,
      "grad_norm": 0.0,
      "learning_rate": 5.652000000000001e-06,
      "loss": 0.0,
      "step": 2180
    },
    {
      "epoch": 0.438,
      "grad_norm": 0.0,
      "learning_rate": 5.6320000000000005e-06,
      "loss": 0.0,
      "step": 2190
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.0,
      "learning_rate": 5.612000000000001e-06,
      "loss": 0.0,
      "step": 2200
    },
    {
      "epoch": 0.442,
      "grad_norm": 0.0,
      "learning_rate": 5.592000000000001e-06,
      "loss": 0.0,
      "step": 2210
    },
    {
      "epoch": 0.444,
      "grad_norm": 0.0,
      "learning_rate": 5.572000000000001e-06,
      "loss": 0.0,
      "step": 2220
    },
    {
      "epoch": 0.446,
      "grad_norm": 0.0,
      "learning_rate": 5.552e-06,
      "loss": 0.0,
      "step": 2230
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.0,
      "learning_rate": 5.5320000000000006e-06,
      "loss": 0.0,
      "step": 2240
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.0,
      "learning_rate": 5.512000000000001e-06,
      "loss": 0.0,
      "step": 2250
    },
    {
      "epoch": 0.452,
      "grad_norm": 0.0,
      "learning_rate": 5.492000000000001e-06,
      "loss": 0.0,
      "step": 2260
    },
    {
      "epoch": 0.454,
      "grad_norm": 0.0,
      "learning_rate": 5.472e-06,
      "loss": 0.0,
      "step": 2270
    },
    {
      "epoch": 0.456,
      "grad_norm": 0.0,
      "learning_rate": 5.4520000000000005e-06,
      "loss": 0.0,
      "step": 2280
    },
    {
      "epoch": 0.458,
      "grad_norm": 0.0,
      "learning_rate": 5.432000000000001e-06,
      "loss": 0.0,
      "step": 2290
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.0,
      "learning_rate": 5.412000000000001e-06,
      "loss": 0.0,
      "step": 2300
    },
    {
      "epoch": 0.462,
      "grad_norm": 0.0,
      "learning_rate": 5.392e-06,
      "loss": 0.0,
      "step": 2310
    },
    {
      "epoch": 0.464,
      "grad_norm": 0.0,
      "learning_rate": 5.372e-06,
      "loss": 0.0,
      "step": 2320
    },
    {
      "epoch": 0.466,
      "grad_norm": 0.0,
      "learning_rate": 5.352000000000001e-06,
      "loss": 0.0,
      "step": 2330
    },
    {
      "epoch": 0.468,
      "grad_norm": 0.0,
      "learning_rate": 5.332000000000001e-06,
      "loss": 0.0,
      "step": 2340
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.0,
      "learning_rate": 5.312e-06,
      "loss": 0.0,
      "step": 2350
    },
    {
      "epoch": 0.472,
      "grad_norm": 0.0,
      "learning_rate": 5.292e-06,
      "loss": 0.0,
      "step": 2360
    },
    {
      "epoch": 0.474,
      "grad_norm": 0.0,
      "learning_rate": 5.2720000000000005e-06,
      "loss": 0.0,
      "step": 2370
    },
    {
      "epoch": 0.476,
      "grad_norm": 0.0,
      "learning_rate": 5.252000000000001e-06,
      "loss": 0.0,
      "step": 2380
    },
    {
      "epoch": 0.478,
      "grad_norm": 0.0,
      "learning_rate": 5.232e-06,
      "loss": 0.0,
      "step": 2390
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.0,
      "learning_rate": 5.212e-06,
      "loss": 0.0,
      "step": 2400
    },
    {
      "epoch": 0.482,
      "grad_norm": 0.0,
      "learning_rate": 5.1920000000000004e-06,
      "loss": 0.0001,
      "step": 2410
    },
    {
      "epoch": 0.484,
      "grad_norm": 0.00044666591566056013,
      "learning_rate": 5.172000000000001e-06,
      "loss": 0.0,
      "step": 2420
    },
    {
      "epoch": 0.486,
      "grad_norm": 0.0007828622474335134,
      "learning_rate": 5.152e-06,
      "loss": 0.0,
      "step": 2430
    },
    {
      "epoch": 0.488,
      "grad_norm": 0.0,
      "learning_rate": 5.134e-06,
      "loss": 1.3219,
      "step": 2440
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.0,
      "learning_rate": 5.114e-06,
      "loss": 0.0,
      "step": 2450
    },
    {
      "epoch": 0.492,
      "grad_norm": 7.122468446141283e-07,
      "learning_rate": 5.094e-06,
      "loss": 0.0,
      "step": 2460
    },
    {
      "epoch": 0.494,
      "grad_norm": 1.69379859471519e-07,
      "learning_rate": 5.074e-06,
      "loss": 0.0,
      "step": 2470
    },
    {
      "epoch": 0.496,
      "grad_norm": 0.020741702988743782,
      "learning_rate": 5.054e-06,
      "loss": 0.0,
      "step": 2480
    },
    {
      "epoch": 0.498,
      "grad_norm": 5.789766444053157e-10,
      "learning_rate": 5.0339999999999996e-06,
      "loss": 0.0,
      "step": 2490
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.0,
      "learning_rate": 5.014e-06,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 0.502,
      "grad_norm": 0.0,
      "learning_rate": 4.994000000000001e-06,
      "loss": 0.0,
      "step": 2510
    },
    {
      "epoch": 0.504,
      "grad_norm": 7.884563224536123e-09,
      "learning_rate": 4.974e-06,
      "loss": 0.0,
      "step": 2520
    },
    {
      "epoch": 0.506,
      "grad_norm": 0.0,
      "learning_rate": 4.954e-06,
      "loss": 0.0,
      "step": 2530
    },
    {
      "epoch": 0.508,
      "grad_norm": 0.0,
      "learning_rate": 4.9340000000000005e-06,
      "loss": 0.0,
      "step": 2540
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.0,
      "learning_rate": 4.914000000000001e-06,
      "loss": 0.0,
      "step": 2550
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.0,
      "learning_rate": 4.894e-06,
      "loss": 0.0,
      "step": 2560
    },
    {
      "epoch": 0.514,
      "grad_norm": 0.0,
      "learning_rate": 4.874e-06,
      "loss": 0.0,
      "step": 2570
    },
    {
      "epoch": 0.516,
      "grad_norm": 0.0,
      "learning_rate": 4.8540000000000005e-06,
      "loss": 0.0,
      "step": 2580
    },
    {
      "epoch": 0.518,
      "grad_norm": 0.0,
      "learning_rate": 4.834000000000001e-06,
      "loss": 0.0,
      "step": 2590
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.0,
      "learning_rate": 4.814e-06,
      "loss": 0.0,
      "step": 2600
    },
    {
      "epoch": 0.522,
      "grad_norm": 0.0,
      "learning_rate": 4.794e-06,
      "loss": 0.0,
      "step": 2610
    },
    {
      "epoch": 0.524,
      "grad_norm": 0.0,
      "learning_rate": 4.774e-06,
      "loss": 0.0,
      "step": 2620
    },
    {
      "epoch": 0.526,
      "grad_norm": 0.0,
      "learning_rate": 4.7540000000000006e-06,
      "loss": 0.0,
      "step": 2630
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.0,
      "learning_rate": 4.734e-06,
      "loss": 0.0,
      "step": 2640
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.0,
      "learning_rate": 4.714e-06,
      "loss": 0.0,
      "step": 2650
    },
    {
      "epoch": 0.532,
      "grad_norm": 0.0,
      "learning_rate": 4.694e-06,
      "loss": 0.0,
      "step": 2660
    },
    {
      "epoch": 0.534,
      "grad_norm": 0.0,
      "learning_rate": 4.6740000000000005e-06,
      "loss": 0.0,
      "step": 2670
    },
    {
      "epoch": 0.536,
      "grad_norm": 0.0,
      "learning_rate": 4.654e-06,
      "loss": 0.0,
      "step": 2680
    },
    {
      "epoch": 0.538,
      "grad_norm": 0.0,
      "learning_rate": 4.634e-06,
      "loss": 0.0,
      "step": 2690
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.0,
      "learning_rate": 4.614e-06,
      "loss": 0.0,
      "step": 2700
    },
    {
      "epoch": 0.542,
      "grad_norm": 0.0,
      "learning_rate": 4.594e-06,
      "loss": 0.0,
      "step": 2710
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.0,
      "learning_rate": 4.574e-06,
      "loss": 0.0,
      "step": 2720
    },
    {
      "epoch": 0.546,
      "grad_norm": 0.0,
      "learning_rate": 4.554000000000001e-06,
      "loss": 0.0,
      "step": 2730
    },
    {
      "epoch": 0.548,
      "grad_norm": 0.0,
      "learning_rate": 4.534000000000001e-06,
      "loss": 0.0431,
      "step": 2740
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0969406494254486e-09,
      "learning_rate": 4.514e-06,
      "loss": 0.0,
      "step": 2750
    },
    {
      "epoch": 0.552,
      "grad_norm": 0.09688814729452133,
      "learning_rate": 4.4940000000000005e-06,
      "loss": 0.0,
      "step": 2760
    },
    {
      "epoch": 0.554,
      "grad_norm": 0.0,
      "learning_rate": 4.474000000000001e-06,
      "loss": 0.0,
      "step": 2770
    },
    {
      "epoch": 0.556,
      "grad_norm": 0.0,
      "learning_rate": 4.454000000000001e-06,
      "loss": 0.0,
      "step": 2780
    },
    {
      "epoch": 0.558,
      "grad_norm": 0.0,
      "learning_rate": 4.434e-06,
      "loss": 0.0,
      "step": 2790
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.0,
      "learning_rate": 4.4140000000000004e-06,
      "loss": 0.0,
      "step": 2800
    },
    {
      "epoch": 0.562,
      "grad_norm": 0.0,
      "learning_rate": 4.394000000000001e-06,
      "loss": 0.0,
      "step": 2810
    },
    {
      "epoch": 0.564,
      "grad_norm": 0.0,
      "learning_rate": 4.374000000000001e-06,
      "loss": 0.0,
      "step": 2820
    },
    {
      "epoch": 0.566,
      "grad_norm": 7.974545468414362e-08,
      "learning_rate": 4.354e-06,
      "loss": 0.0,
      "step": 2830
    },
    {
      "epoch": 0.568,
      "grad_norm": 0.0,
      "learning_rate": 4.334e-06,
      "loss": 0.0,
      "step": 2840
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.0,
      "learning_rate": 4.3140000000000005e-06,
      "loss": 0.0,
      "step": 2850
    },
    {
      "epoch": 0.572,
      "grad_norm": 0.0,
      "learning_rate": 4.294000000000001e-06,
      "loss": 0.0,
      "step": 2860
    },
    {
      "epoch": 0.574,
      "grad_norm": 0.0,
      "learning_rate": 4.274e-06,
      "loss": 0.0,
      "step": 2870
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.0,
      "learning_rate": 4.254e-06,
      "loss": 0.0,
      "step": 2880
    },
    {
      "epoch": 0.578,
      "grad_norm": 0.0,
      "learning_rate": 4.2340000000000005e-06,
      "loss": 0.2195,
      "step": 2890
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.0,
      "learning_rate": 4.214000000000001e-06,
      "loss": 0.0038,
      "step": 2900
    },
    {
      "epoch": 0.582,
      "grad_norm": 0.0,
      "learning_rate": 4.194e-06,
      "loss": 0.0,
      "step": 2910
    },
    {
      "epoch": 0.584,
      "grad_norm": 0.0,
      "learning_rate": 4.174e-06,
      "loss": 0.0,
      "step": 2920
    },
    {
      "epoch": 0.586,
      "grad_norm": 1.3934201206211583e-06,
      "learning_rate": 4.154e-06,
      "loss": 0.0,
      "step": 2930
    },
    {
      "epoch": 0.588,
      "grad_norm": 0.0,
      "learning_rate": 4.1340000000000006e-06,
      "loss": 0.0,
      "step": 2940
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.0,
      "learning_rate": 4.114e-06,
      "loss": 0.0,
      "step": 2950
    },
    {
      "epoch": 0.592,
      "grad_norm": 0.0,
      "learning_rate": 4.094e-06,
      "loss": 0.0,
      "step": 2960
    },
    {
      "epoch": 0.594,
      "grad_norm": 0.0,
      "learning_rate": 4.074e-06,
      "loss": 0.0,
      "step": 2970
    },
    {
      "epoch": 0.596,
      "grad_norm": 0.0,
      "learning_rate": 4.0540000000000005e-06,
      "loss": 0.0,
      "step": 2980
    },
    {
      "epoch": 0.598,
      "grad_norm": 0.0,
      "learning_rate": 4.034e-06,
      "loss": 0.0,
      "step": 2990
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.0,
      "learning_rate": 4.014e-06,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 0.602,
      "grad_norm": 0.0,
      "learning_rate": 3.994e-06,
      "loss": 0.0,
      "step": 3010
    },
    {
      "epoch": 0.604,
      "grad_norm": 0.0,
      "learning_rate": 3.974e-06,
      "loss": 0.0,
      "step": 3020
    },
    {
      "epoch": 0.606,
      "grad_norm": 0.0034688322339206934,
      "learning_rate": 3.954e-06,
      "loss": 0.0,
      "step": 3030
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.0,
      "learning_rate": 3.934000000000001e-06,
      "loss": 0.0,
      "step": 3040
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.0,
      "learning_rate": 3.914000000000001e-06,
      "loss": 0.0,
      "step": 3050
    },
    {
      "epoch": 0.612,
      "grad_norm": 0.0,
      "learning_rate": 3.894e-06,
      "loss": 0.0,
      "step": 3060
    },
    {
      "epoch": 0.614,
      "grad_norm": 0.0,
      "learning_rate": 3.8740000000000005e-06,
      "loss": 0.0,
      "step": 3070
    },
    {
      "epoch": 0.616,
      "grad_norm": 0.0,
      "learning_rate": 3.854000000000001e-06,
      "loss": 0.0,
      "step": 3080
    },
    {
      "epoch": 0.618,
      "grad_norm": 0.0,
      "learning_rate": 3.834000000000001e-06,
      "loss": 0.0,
      "step": 3090
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.0,
      "learning_rate": 3.8140000000000007e-06,
      "loss": 0.0,
      "step": 3100
    },
    {
      "epoch": 0.622,
      "grad_norm": 0.0,
      "learning_rate": 3.7940000000000004e-06,
      "loss": 0.0,
      "step": 3110
    },
    {
      "epoch": 0.624,
      "grad_norm": 0.0,
      "learning_rate": 3.7740000000000006e-06,
      "loss": 0.0,
      "step": 3120
    },
    {
      "epoch": 0.626,
      "grad_norm": 0.0,
      "learning_rate": 3.7540000000000004e-06,
      "loss": 0.0,
      "step": 3130
    },
    {
      "epoch": 0.628,
      "grad_norm": 0.0,
      "learning_rate": 3.7340000000000006e-06,
      "loss": 0.0,
      "step": 3140
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.3274219933664426e-05,
      "learning_rate": 3.7140000000000004e-06,
      "loss": 0.0,
      "step": 3150
    },
    {
      "epoch": 0.632,
      "grad_norm": 0.0,
      "learning_rate": 3.6940000000000005e-06,
      "loss": 0.0,
      "step": 3160
    },
    {
      "epoch": 0.634,
      "grad_norm": 0.0,
      "learning_rate": 3.6740000000000003e-06,
      "loss": 0.0,
      "step": 3170
    },
    {
      "epoch": 0.636,
      "grad_norm": 0.0,
      "learning_rate": 3.6540000000000005e-06,
      "loss": 0.0,
      "step": 3180
    },
    {
      "epoch": 0.638,
      "grad_norm": 0.0,
      "learning_rate": 3.6340000000000003e-06,
      "loss": 0.0,
      "step": 3190
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.0,
      "learning_rate": 3.6140000000000005e-06,
      "loss": 0.0,
      "step": 3200
    },
    {
      "epoch": 0.642,
      "grad_norm": 0.0,
      "learning_rate": 3.5940000000000002e-06,
      "loss": 0.0,
      "step": 3210
    },
    {
      "epoch": 0.644,
      "grad_norm": 0.0,
      "learning_rate": 3.5740000000000004e-06,
      "loss": 0.0,
      "step": 3220
    },
    {
      "epoch": 0.646,
      "grad_norm": 0.0,
      "learning_rate": 3.554e-06,
      "loss": 0.0,
      "step": 3230
    },
    {
      "epoch": 0.648,
      "grad_norm": 0.0,
      "learning_rate": 3.5340000000000004e-06,
      "loss": 0.0,
      "step": 3240
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.0,
      "learning_rate": 3.514e-06,
      "loss": 0.0,
      "step": 3250
    },
    {
      "epoch": 0.652,
      "grad_norm": 0.0,
      "learning_rate": 3.4940000000000003e-06,
      "loss": 0.0,
      "step": 3260
    },
    {
      "epoch": 0.654,
      "grad_norm": 0.0,
      "learning_rate": 3.474e-06,
      "loss": 0.0,
      "step": 3270
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.0,
      "learning_rate": 3.4540000000000003e-06,
      "loss": 0.0,
      "step": 3280
    },
    {
      "epoch": 0.658,
      "grad_norm": 1.2282844075173216e-08,
      "learning_rate": 3.434e-06,
      "loss": 0.0,
      "step": 3290
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.0,
      "learning_rate": 3.4140000000000003e-06,
      "loss": 0.0,
      "step": 3300
    },
    {
      "epoch": 0.662,
      "grad_norm": 0.0,
      "learning_rate": 3.394e-06,
      "loss": 0.0,
      "step": 3310
    },
    {
      "epoch": 0.664,
      "grad_norm": 0.0,
      "learning_rate": 3.3740000000000002e-06,
      "loss": 0.0,
      "step": 3320
    },
    {
      "epoch": 0.666,
      "grad_norm": 0.0,
      "learning_rate": 3.354e-06,
      "loss": 0.0,
      "step": 3330
    },
    {
      "epoch": 0.668,
      "grad_norm": 0.0,
      "learning_rate": 3.334e-06,
      "loss": 0.0,
      "step": 3340
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.0,
      "learning_rate": 3.314e-06,
      "loss": 0.0,
      "step": 3350
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.0,
      "learning_rate": 3.2940000000000006e-06,
      "loss": 0.0,
      "step": 3360
    },
    {
      "epoch": 0.674,
      "grad_norm": 0.0,
      "learning_rate": 3.2740000000000003e-06,
      "loss": 0.0,
      "step": 3370
    },
    {
      "epoch": 0.676,
      "grad_norm": 0.0,
      "learning_rate": 3.2540000000000005e-06,
      "loss": 0.0,
      "step": 3380
    },
    {
      "epoch": 0.678,
      "grad_norm": 0.0,
      "learning_rate": 3.2340000000000003e-06,
      "loss": 0.0,
      "step": 3390
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.0,
      "learning_rate": 3.2140000000000005e-06,
      "loss": 0.0,
      "step": 3400
    },
    {
      "epoch": 0.682,
      "grad_norm": 0.0,
      "learning_rate": 3.1940000000000003e-06,
      "loss": 0.0,
      "step": 3410
    },
    {
      "epoch": 0.684,
      "grad_norm": 0.0,
      "learning_rate": 3.1740000000000004e-06,
      "loss": 0.0,
      "step": 3420
    },
    {
      "epoch": 0.686,
      "grad_norm": 0.0,
      "learning_rate": 3.154e-06,
      "loss": 0.0,
      "step": 3430
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.0,
      "learning_rate": 3.1340000000000004e-06,
      "loss": 0.0,
      "step": 3440
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.0,
      "learning_rate": 3.114e-06,
      "loss": 0.0,
      "step": 3450
    },
    {
      "epoch": 0.692,
      "grad_norm": 0.0,
      "learning_rate": 3.0940000000000004e-06,
      "loss": 0.0,
      "step": 3460
    },
    {
      "epoch": 0.694,
      "grad_norm": 0.0,
      "learning_rate": 3.074e-06,
      "loss": 0.0,
      "step": 3470
    },
    {
      "epoch": 0.696,
      "grad_norm": 0.0,
      "learning_rate": 3.0540000000000003e-06,
      "loss": 0.0,
      "step": 3480
    },
    {
      "epoch": 0.698,
      "grad_norm": 0.0,
      "learning_rate": 3.034e-06,
      "loss": 0.0,
      "step": 3490
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.0,
      "learning_rate": 3.0140000000000003e-06,
      "loss": 0.0,
      "step": 3500
    },
    {
      "epoch": 0.702,
      "grad_norm": 0.0,
      "learning_rate": 2.994e-06,
      "loss": 0.0,
      "step": 3510
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.0,
      "learning_rate": 2.9740000000000002e-06,
      "loss": 0.0,
      "step": 3520
    },
    {
      "epoch": 0.706,
      "grad_norm": 0.0,
      "learning_rate": 2.954e-06,
      "loss": 0.0,
      "step": 3530
    },
    {
      "epoch": 0.708,
      "grad_norm": 0.0,
      "learning_rate": 2.934e-06,
      "loss": 0.0,
      "step": 3540
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.0,
      "learning_rate": 2.914e-06,
      "loss": 0.0,
      "step": 3550
    },
    {
      "epoch": 0.712,
      "grad_norm": 0.0,
      "learning_rate": 2.894e-06,
      "loss": 0.0,
      "step": 3560
    },
    {
      "epoch": 0.714,
      "grad_norm": 0.0,
      "learning_rate": 2.874e-06,
      "loss": 0.0,
      "step": 3570
    },
    {
      "epoch": 0.716,
      "grad_norm": 0.0,
      "learning_rate": 2.854e-06,
      "loss": 0.0,
      "step": 3580
    },
    {
      "epoch": 0.718,
      "grad_norm": 1.1484501127867475e-09,
      "learning_rate": 2.834e-06,
      "loss": 0.0,
      "step": 3590
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.0,
      "learning_rate": 2.814e-06,
      "loss": 0.0,
      "step": 3600
    },
    {
      "epoch": 0.722,
      "grad_norm": 0.0,
      "learning_rate": 2.794e-06,
      "loss": 0.0,
      "step": 3610
    },
    {
      "epoch": 0.724,
      "grad_norm": 0.0,
      "learning_rate": 2.774e-06,
      "loss": 0.0,
      "step": 3620
    },
    {
      "epoch": 0.726,
      "grad_norm": 0.0,
      "learning_rate": 2.754e-06,
      "loss": 0.0,
      "step": 3630
    },
    {
      "epoch": 0.728,
      "grad_norm": 0.0,
      "learning_rate": 2.734e-06,
      "loss": 0.0,
      "step": 3640
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.0,
      "learning_rate": 2.7139999999999998e-06,
      "loss": 0.0,
      "step": 3650
    },
    {
      "epoch": 0.732,
      "grad_norm": 0.0,
      "learning_rate": 2.694e-06,
      "loss": 0.0,
      "step": 3660
    },
    {
      "epoch": 0.734,
      "grad_norm": 0.0,
      "learning_rate": 2.6740000000000006e-06,
      "loss": 0.0,
      "step": 3670
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.0,
      "learning_rate": 2.6540000000000003e-06,
      "loss": 0.0,
      "step": 3680
    },
    {
      "epoch": 0.738,
      "grad_norm": 0.0,
      "learning_rate": 2.6340000000000005e-06,
      "loss": 0.0,
      "step": 3690
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.0,
      "learning_rate": 2.6140000000000003e-06,
      "loss": 0.0,
      "step": 3700
    },
    {
      "epoch": 0.742,
      "grad_norm": 0.0,
      "learning_rate": 2.5940000000000005e-06,
      "loss": 0.0,
      "step": 3710
    },
    {
      "epoch": 0.744,
      "grad_norm": 0.0,
      "learning_rate": 2.5740000000000003e-06,
      "loss": 0.0,
      "step": 3720
    },
    {
      "epoch": 0.746,
      "grad_norm": 0.0,
      "learning_rate": 2.5540000000000004e-06,
      "loss": 0.0,
      "step": 3730
    },
    {
      "epoch": 0.748,
      "grad_norm": 0.0,
      "learning_rate": 2.5340000000000002e-06,
      "loss": 0.0,
      "step": 3740
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.0,
      "learning_rate": 2.5140000000000004e-06,
      "loss": 0.0,
      "step": 3750
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.0,
      "learning_rate": 2.494e-06,
      "loss": 0.0,
      "step": 3760
    },
    {
      "epoch": 0.754,
      "grad_norm": 0.0,
      "learning_rate": 2.4740000000000004e-06,
      "loss": 0.0,
      "step": 3770
    },
    {
      "epoch": 0.756,
      "grad_norm": 0.0,
      "learning_rate": 2.454e-06,
      "loss": 0.0,
      "step": 3780
    },
    {
      "epoch": 0.758,
      "grad_norm": 0.0,
      "learning_rate": 2.4340000000000003e-06,
      "loss": 0.0,
      "step": 3790
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.0,
      "learning_rate": 2.414e-06,
      "loss": 0.0,
      "step": 3800
    },
    {
      "epoch": 0.762,
      "grad_norm": 0.0,
      "learning_rate": 2.3940000000000003e-06,
      "loss": 0.0001,
      "step": 3810
    },
    {
      "epoch": 0.764,
      "grad_norm": 0.0,
      "learning_rate": 2.374e-06,
      "loss": 0.0,
      "step": 3820
    },
    {
      "epoch": 0.766,
      "grad_norm": 0.0,
      "learning_rate": 2.3540000000000002e-06,
      "loss": 0.0,
      "step": 3830
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.0,
      "learning_rate": 2.334e-06,
      "loss": 0.0,
      "step": 3840
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.0,
      "learning_rate": 2.314e-06,
      "loss": 0.0,
      "step": 3850
    },
    {
      "epoch": 0.772,
      "grad_norm": 0.0,
      "learning_rate": 2.294e-06,
      "loss": 0.0,
      "step": 3860
    },
    {
      "epoch": 0.774,
      "grad_norm": 0.0,
      "learning_rate": 2.274e-06,
      "loss": 0.0,
      "step": 3870
    },
    {
      "epoch": 0.776,
      "grad_norm": 0.0,
      "learning_rate": 2.254e-06,
      "loss": 0.0,
      "step": 3880
    },
    {
      "epoch": 0.778,
      "grad_norm": 0.0,
      "learning_rate": 2.234e-06,
      "loss": 0.0,
      "step": 3890
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.0,
      "learning_rate": 2.2140000000000003e-06,
      "loss": 0.0,
      "step": 3900
    },
    {
      "epoch": 0.782,
      "grad_norm": 0.0,
      "learning_rate": 2.194e-06,
      "loss": 0.0,
      "step": 3910
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.0,
      "learning_rate": 2.1740000000000003e-06,
      "loss": 0.0,
      "step": 3920
    },
    {
      "epoch": 0.786,
      "grad_norm": 0.0,
      "learning_rate": 2.1540000000000005e-06,
      "loss": 0.0,
      "step": 3930
    },
    {
      "epoch": 0.788,
      "grad_norm": 0.0,
      "learning_rate": 2.1340000000000002e-06,
      "loss": 0.0,
      "step": 3940
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.0,
      "learning_rate": 2.1140000000000004e-06,
      "loss": 0.0,
      "step": 3950
    },
    {
      "epoch": 0.792,
      "grad_norm": 0.0013283728621900082,
      "learning_rate": 2.094e-06,
      "loss": 0.0,
      "step": 3960
    },
    {
      "epoch": 0.794,
      "grad_norm": 0.0,
      "learning_rate": 2.0740000000000004e-06,
      "loss": 0.0,
      "step": 3970
    },
    {
      "epoch": 0.796,
      "grad_norm": 0.0,
      "learning_rate": 2.054e-06,
      "loss": 0.0,
      "step": 3980
    },
    {
      "epoch": 0.798,
      "grad_norm": 0.0,
      "learning_rate": 2.0340000000000003e-06,
      "loss": 0.0,
      "step": 3990
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.0,
      "learning_rate": 2.014e-06,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 0.802,
      "grad_norm": 0.0,
      "learning_rate": 1.9940000000000003e-06,
      "loss": 0.0,
      "step": 4010
    },
    {
      "epoch": 0.804,
      "grad_norm": 0.0,
      "learning_rate": 1.974e-06,
      "loss": 0.0,
      "step": 4020
    },
    {
      "epoch": 0.806,
      "grad_norm": 0.0,
      "learning_rate": 1.9540000000000003e-06,
      "loss": 0.0,
      "step": 4030
    },
    {
      "epoch": 0.808,
      "grad_norm": 0.0,
      "learning_rate": 1.934e-06,
      "loss": 0.0,
      "step": 4040
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.0,
      "learning_rate": 1.9140000000000002e-06,
      "loss": 0.0,
      "step": 4050
    },
    {
      "epoch": 0.812,
      "grad_norm": 0.0,
      "learning_rate": 1.8940000000000002e-06,
      "loss": 0.0,
      "step": 4060
    },
    {
      "epoch": 0.814,
      "grad_norm": 0.0,
      "learning_rate": 1.8740000000000002e-06,
      "loss": 0.0,
      "step": 4070
    },
    {
      "epoch": 0.816,
      "grad_norm": 0.0,
      "learning_rate": 1.8540000000000002e-06,
      "loss": 0.0,
      "step": 4080
    },
    {
      "epoch": 0.818,
      "grad_norm": 0.0,
      "learning_rate": 1.8340000000000001e-06,
      "loss": 0.0,
      "step": 4090
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.0,
      "learning_rate": 1.8140000000000001e-06,
      "loss": 0.0,
      "step": 4100
    },
    {
      "epoch": 0.822,
      "grad_norm": 0.0,
      "learning_rate": 1.794e-06,
      "loss": 0.0,
      "step": 4110
    },
    {
      "epoch": 0.824,
      "grad_norm": 0.0,
      "learning_rate": 1.774e-06,
      "loss": 0.0,
      "step": 4120
    },
    {
      "epoch": 0.826,
      "grad_norm": 0.0,
      "learning_rate": 1.754e-06,
      "loss": 0.0,
      "step": 4130
    },
    {
      "epoch": 0.828,
      "grad_norm": 0.0,
      "learning_rate": 1.734e-06,
      "loss": 0.0,
      "step": 4140
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.0,
      "learning_rate": 1.714e-06,
      "loss": 0.0,
      "step": 4150
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.0,
      "learning_rate": 1.694e-06,
      "loss": 0.0,
      "step": 4160
    },
    {
      "epoch": 0.834,
      "grad_norm": 0.0,
      "learning_rate": 1.674e-06,
      "loss": 0.0,
      "step": 4170
    },
    {
      "epoch": 0.836,
      "grad_norm": 0.0,
      "learning_rate": 1.654e-06,
      "loss": 0.0007,
      "step": 4180
    },
    {
      "epoch": 0.838,
      "grad_norm": 0.0,
      "learning_rate": 1.634e-06,
      "loss": 0.0,
      "step": 4190
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.0,
      "learning_rate": 1.614e-06,
      "loss": 0.0,
      "step": 4200
    },
    {
      "epoch": 0.842,
      "grad_norm": 0.0,
      "learning_rate": 1.594e-06,
      "loss": 0.0,
      "step": 4210
    },
    {
      "epoch": 0.844,
      "grad_norm": 0.0,
      "learning_rate": 1.5740000000000003e-06,
      "loss": 0.0,
      "step": 4220
    },
    {
      "epoch": 0.846,
      "grad_norm": 0.0,
      "learning_rate": 1.5540000000000003e-06,
      "loss": 0.0,
      "step": 4230
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.0,
      "learning_rate": 1.5340000000000003e-06,
      "loss": 0.0,
      "step": 4240
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.0,
      "learning_rate": 1.5140000000000002e-06,
      "loss": 0.0,
      "step": 4250
    },
    {
      "epoch": 0.852,
      "grad_norm": 0.0,
      "learning_rate": 1.4940000000000002e-06,
      "loss": 0.0,
      "step": 4260
    },
    {
      "epoch": 0.854,
      "grad_norm": 0.0,
      "learning_rate": 1.4740000000000002e-06,
      "loss": 0.0,
      "step": 4270
    },
    {
      "epoch": 0.856,
      "grad_norm": 0.0,
      "learning_rate": 1.4540000000000002e-06,
      "loss": 0.0,
      "step": 4280
    },
    {
      "epoch": 0.858,
      "grad_norm": 0.0,
      "learning_rate": 1.4340000000000002e-06,
      "loss": 0.0,
      "step": 4290
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.0,
      "learning_rate": 1.4140000000000001e-06,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 0.862,
      "grad_norm": 0.0,
      "learning_rate": 1.3940000000000001e-06,
      "loss": 0.0,
      "step": 4310
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.0,
      "learning_rate": 1.374e-06,
      "loss": 0.0,
      "step": 4320
    },
    {
      "epoch": 0.866,
      "grad_norm": 0.0,
      "learning_rate": 1.354e-06,
      "loss": 0.0,
      "step": 4330
    },
    {
      "epoch": 0.868,
      "grad_norm": 0.0,
      "learning_rate": 1.334e-06,
      "loss": 0.0,
      "step": 4340
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.0,
      "learning_rate": 1.314e-06,
      "loss": 0.0,
      "step": 4350
    },
    {
      "epoch": 0.872,
      "grad_norm": 0.0,
      "learning_rate": 1.294e-06,
      "loss": 0.0,
      "step": 4360
    },
    {
      "epoch": 0.874,
      "grad_norm": 0.0,
      "learning_rate": 1.2740000000000002e-06,
      "loss": 0.0,
      "step": 4370
    },
    {
      "epoch": 0.876,
      "grad_norm": 0.0,
      "learning_rate": 1.2540000000000002e-06,
      "loss": 0.0,
      "step": 4380
    },
    {
      "epoch": 0.878,
      "grad_norm": 0.0,
      "learning_rate": 1.234e-06,
      "loss": 0.0,
      "step": 4390
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.0,
      "learning_rate": 1.214e-06,
      "loss": 0.0,
      "step": 4400
    },
    {
      "epoch": 0.882,
      "grad_norm": 0.0,
      "learning_rate": 1.1940000000000001e-06,
      "loss": 0.0,
      "step": 4410
    },
    {
      "epoch": 0.884,
      "grad_norm": 0.0,
      "learning_rate": 1.1740000000000001e-06,
      "loss": 0.0,
      "step": 4420
    },
    {
      "epoch": 0.886,
      "grad_norm": 0.0,
      "learning_rate": 1.154e-06,
      "loss": 0.0,
      "step": 4430
    },
    {
      "epoch": 0.888,
      "grad_norm": 0.0,
      "learning_rate": 1.134e-06,
      "loss": 0.0,
      "step": 4440
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.0,
      "learning_rate": 1.114e-06,
      "loss": 0.0,
      "step": 4450
    },
    {
      "epoch": 0.892,
      "grad_norm": 0.0,
      "learning_rate": 1.094e-06,
      "loss": 0.0,
      "step": 4460
    },
    {
      "epoch": 0.894,
      "grad_norm": 0.0,
      "learning_rate": 1.074e-06,
      "loss": 0.0,
      "step": 4470
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.0,
      "learning_rate": 1.054e-06,
      "loss": 0.0,
      "step": 4480
    },
    {
      "epoch": 0.898,
      "grad_norm": 0.0,
      "learning_rate": 1.0340000000000002e-06,
      "loss": 0.0,
      "step": 4490
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.0,
      "learning_rate": 1.0140000000000002e-06,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 0.902,
      "grad_norm": 0.0,
      "learning_rate": 9.940000000000001e-07,
      "loss": 0.0,
      "step": 4510
    },
    {
      "epoch": 0.904,
      "grad_norm": 0.0,
      "learning_rate": 9.740000000000001e-07,
      "loss": 0.0,
      "step": 4520
    },
    {
      "epoch": 0.906,
      "grad_norm": 0.0,
      "learning_rate": 9.54e-07,
      "loss": 0.0,
      "step": 4530
    },
    {
      "epoch": 0.908,
      "grad_norm": 0.0,
      "learning_rate": 9.340000000000001e-07,
      "loss": 0.0,
      "step": 4540
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.0,
      "learning_rate": 9.140000000000001e-07,
      "loss": 0.0,
      "step": 4550
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.0,
      "learning_rate": 8.94e-07,
      "loss": 0.0,
      "step": 4560
    },
    {
      "epoch": 0.914,
      "grad_norm": 0.0,
      "learning_rate": 8.740000000000001e-07,
      "loss": 0.0,
      "step": 4570
    },
    {
      "epoch": 0.916,
      "grad_norm": 0.0,
      "learning_rate": 8.540000000000001e-07,
      "loss": 0.0,
      "step": 4580
    },
    {
      "epoch": 0.918,
      "grad_norm": 0.0,
      "learning_rate": 8.36e-07,
      "loss": 0.7887,
      "step": 4590
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.0,
      "learning_rate": 8.160000000000001e-07,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 0.922,
      "grad_norm": 0.0,
      "learning_rate": 7.960000000000001e-07,
      "loss": 0.0,
      "step": 4610
    },
    {
      "epoch": 0.924,
      "grad_norm": 0.0,
      "learning_rate": 7.760000000000001e-07,
      "loss": 0.0,
      "step": 4620
    },
    {
      "epoch": 0.926,
      "grad_norm": 0.0,
      "learning_rate": 7.56e-07,
      "loss": 0.0,
      "step": 4630
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.0,
      "learning_rate": 7.36e-07,
      "loss": 0.0,
      "step": 4640
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.0,
      "learning_rate": 7.16e-07,
      "loss": 0.0,
      "step": 4650
    },
    {
      "epoch": 0.932,
      "grad_norm": 0.0,
      "learning_rate": 6.96e-07,
      "loss": 0.0,
      "step": 4660
    },
    {
      "epoch": 0.934,
      "grad_norm": 0.0,
      "learning_rate": 6.76e-07,
      "loss": 0.0,
      "step": 4670
    },
    {
      "epoch": 0.936,
      "grad_norm": 0.0,
      "learning_rate": 6.560000000000002e-07,
      "loss": 0.0,
      "step": 4680
    },
    {
      "epoch": 0.938,
      "grad_norm": 0.0,
      "learning_rate": 6.360000000000001e-07,
      "loss": 0.0,
      "step": 4690
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.0,
      "learning_rate": 6.160000000000001e-07,
      "loss": 0.0,
      "step": 4700
    },
    {
      "epoch": 0.942,
      "grad_norm": 0.0,
      "learning_rate": 5.960000000000001e-07,
      "loss": 0.0,
      "step": 4710
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.0,
      "learning_rate": 5.760000000000001e-07,
      "loss": 0.0,
      "step": 4720
    },
    {
      "epoch": 0.946,
      "grad_norm": 0.0,
      "learning_rate": 5.560000000000001e-07,
      "loss": 0.0,
      "step": 4730
    },
    {
      "epoch": 0.948,
      "grad_norm": 0.0,
      "learning_rate": 5.36e-07,
      "loss": 0.0,
      "step": 4740
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.0,
      "learning_rate": 5.16e-07,
      "loss": 0.0,
      "step": 4750
    },
    {
      "epoch": 0.952,
      "grad_norm": 0.0,
      "learning_rate": 4.96e-07,
      "loss": 0.0,
      "step": 4760
    },
    {
      "epoch": 0.954,
      "grad_norm": 0.0,
      "learning_rate": 4.760000000000001e-07,
      "loss": 0.0,
      "step": 4770
    },
    {
      "epoch": 0.956,
      "grad_norm": 0.0,
      "learning_rate": 4.5600000000000006e-07,
      "loss": 0.0,
      "step": 4780
    },
    {
      "epoch": 0.958,
      "grad_norm": 0.0,
      "learning_rate": 4.3600000000000004e-07,
      "loss": 0.0,
      "step": 4790
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.0,
      "learning_rate": 4.16e-07,
      "loss": 0.0,
      "step": 4800
    },
    {
      "epoch": 0.962,
      "grad_norm": 0.0,
      "learning_rate": 3.9600000000000005e-07,
      "loss": 0.0,
      "step": 4810
    },
    {
      "epoch": 0.964,
      "grad_norm": 0.0,
      "learning_rate": 3.7600000000000003e-07,
      "loss": 0.0,
      "step": 4820
    },
    {
      "epoch": 0.966,
      "grad_norm": 0.0,
      "learning_rate": 3.56e-07,
      "loss": 0.0,
      "step": 4830
    },
    {
      "epoch": 0.968,
      "grad_norm": 0.0,
      "learning_rate": 3.36e-07,
      "loss": 0.0,
      "step": 4840
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.0,
      "learning_rate": 3.160000000000001e-07,
      "loss": 0.0,
      "step": 4850
    },
    {
      "epoch": 0.972,
      "grad_norm": 0.0,
      "learning_rate": 2.9600000000000006e-07,
      "loss": 0.0,
      "step": 4860
    },
    {
      "epoch": 0.974,
      "grad_norm": 0.0,
      "learning_rate": 2.7600000000000004e-07,
      "loss": 0.0,
      "step": 4870
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.0,
      "learning_rate": 2.56e-07,
      "loss": 0.0,
      "step": 4880
    },
    {
      "epoch": 0.978,
      "grad_norm": 0.0,
      "learning_rate": 2.3600000000000002e-07,
      "loss": 0.0,
      "step": 4890
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.0,
      "learning_rate": 2.1600000000000003e-07,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 0.982,
      "grad_norm": 0.0,
      "learning_rate": 1.96e-07,
      "loss": 0.0,
      "step": 4910
    },
    {
      "epoch": 0.984,
      "grad_norm": 0.0,
      "learning_rate": 1.7600000000000001e-07,
      "loss": 0.0,
      "step": 4920
    },
    {
      "epoch": 0.986,
      "grad_norm": 0.0,
      "learning_rate": 1.56e-07,
      "loss": 0.0,
      "step": 4930
    },
    {
      "epoch": 0.988,
      "grad_norm": 0.0,
      "learning_rate": 1.36e-07,
      "loss": 0.0,
      "step": 4940
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.0,
      "learning_rate": 1.16e-07,
      "loss": 0.0,
      "step": 4950
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.0,
      "learning_rate": 9.6e-08,
      "loss": 0.0,
      "step": 4960
    },
    {
      "epoch": 0.994,
      "grad_norm": 0.0,
      "learning_rate": 7.6e-08,
      "loss": 0.0,
      "step": 4970
    },
    {
      "epoch": 0.996,
      "grad_norm": 0.0,
      "learning_rate": 5.6000000000000005e-08,
      "loss": 0.0,
      "step": 4980
    },
    {
      "epoch": 0.998,
      "grad_norm": 0.0,
      "learning_rate": 3.6000000000000005e-08,
      "loss": 0.0,
      "step": 4990
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.0,
      "learning_rate": 1.6e-08,
      "loss": 0.0,
      "step": 5000
    }
  ],
  "logging_steps": 10,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}