{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.3,
  "eval_steps": 500,
  "global_step": 1875,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001,
      "loss": 3.7379,
      "step": 5
    },
    {
      "epoch": 0.0016,
      "grad_norm": 1.0,
      "learning_rate": 0.0001,
      "loss": 2.5903,
      "step": 10
    },
    {
      "epoch": 0.0024,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001,
      "loss": 1.3107,
      "step": 15
    },
    {
      "epoch": 0.0032,
      "grad_norm": 0.32421875,
      "learning_rate": 0.0001,
      "loss": 0.9727,
      "step": 20
    },
    {
      "epoch": 0.004,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001,
      "loss": 0.8397,
      "step": 25
    },
    {
      "epoch": 0.0048,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001,
      "loss": 0.606,
      "step": 30
    },
    {
      "epoch": 0.0056,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0001,
      "loss": 0.4789,
      "step": 35
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0001,
      "loss": 0.3708,
      "step": 40
    },
    {
      "epoch": 0.0072,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0001,
      "loss": 0.2913,
      "step": 45
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0001,
      "loss": 0.262,
      "step": 50
    },
    {
      "epoch": 0.0088,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001,
      "loss": 0.2648,
      "step": 55
    },
    {
      "epoch": 0.0096,
      "grad_norm": 1.7578125,
      "learning_rate": 0.0001,
      "loss": 0.2817,
      "step": 60
    },
    {
      "epoch": 0.0104,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0001,
      "loss": 0.3385,
      "step": 65
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0001,
      "loss": 0.3304,
      "step": 70
    },
    {
      "epoch": 0.012,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0001,
      "loss": 0.2756,
      "step": 75
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0001,
      "loss": 0.2459,
      "step": 80
    },
    {
      "epoch": 0.0136,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0001,
      "loss": 0.3307,
      "step": 85
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001,
      "loss": 0.3,
      "step": 90
    },
    {
      "epoch": 0.0152,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0001,
      "loss": 0.3219,
      "step": 95
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0001,
      "loss": 0.2893,
      "step": 100
    },
    {
      "epoch": 0.0168,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001,
      "loss": 0.1981,
      "step": 105
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0001,
      "loss": 0.262,
      "step": 110
    },
    {
      "epoch": 0.0184,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0001,
      "loss": 0.2604,
      "step": 115
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0001,
      "loss": 0.3104,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.203125,
      "learning_rate": 0.0001,
      "loss": 0.2566,
      "step": 125
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0001,
      "loss": 0.2938,
      "step": 130
    },
    {
      "epoch": 0.0216,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0001,
      "loss": 0.2879,
      "step": 135
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001,
      "loss": 0.3034,
      "step": 140
    },
    {
      "epoch": 0.0232,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.2789,
      "step": 145
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0001,
      "loss": 0.1826,
      "step": 150
    },
    {
      "epoch": 0.0248,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0001,
      "loss": 0.1382,
      "step": 155
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0001,
      "loss": 0.227,
      "step": 160
    },
    {
      "epoch": 0.0264,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0001,
      "loss": 0.2143,
      "step": 165
    },
    {
      "epoch": 0.0272,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0001,
      "loss": 0.2475,
      "step": 170
    },
    {
      "epoch": 0.028,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001,
      "loss": 0.2576,
      "step": 175
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0001,
      "loss": 0.2017,
      "step": 180
    },
    {
      "epoch": 0.0296,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001,
      "loss": 0.2451,
      "step": 185
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001,
      "loss": 0.2331,
      "step": 190
    },
    {
      "epoch": 0.0312,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0001,
      "loss": 0.2384,
      "step": 195
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0001,
      "loss": 0.2011,
      "step": 200
    },
    {
      "epoch": 0.0328,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0001,
      "loss": 0.2026,
      "step": 205
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0001,
      "loss": 0.2468,
      "step": 210
    },
    {
      "epoch": 0.0344,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0001,
      "loss": 0.2545,
      "step": 215
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0001,
      "loss": 0.254,
      "step": 220
    },
    {
      "epoch": 0.036,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0001,
      "loss": 0.2435,
      "step": 225
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0001,
      "loss": 0.2118,
      "step": 230
    },
    {
      "epoch": 0.0376,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.2569,
      "step": 235
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001,
      "loss": 0.2391,
      "step": 240
    },
    {
      "epoch": 0.0392,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0001,
      "loss": 0.2963,
      "step": 245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.193,
      "step": 250
    },
    {
      "epoch": 0.0408,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0001,
      "loss": 0.1782,
      "step": 255
    },
    {
      "epoch": 0.0416,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0001,
      "loss": 0.2331,
      "step": 260
    },
    {
      "epoch": 0.0424,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0001,
      "loss": 0.2288,
      "step": 265
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0001,
      "loss": 0.2077,
      "step": 270
    },
    {
      "epoch": 0.044,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0001,
      "loss": 0.2659,
      "step": 275
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0001,
      "loss": 0.2796,
      "step": 280
    },
    {
      "epoch": 0.0456,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001,
      "loss": 0.2486,
      "step": 285
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001,
      "loss": 0.2317,
      "step": 290
    },
    {
      "epoch": 0.0472,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001,
      "loss": 0.2234,
      "step": 295
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0001,
      "loss": 0.2471,
      "step": 300
    },
    {
      "epoch": 0.0488,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0001,
      "loss": 0.1701,
      "step": 305
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0001,
      "loss": 0.2577,
      "step": 310
    },
    {
      "epoch": 0.0504,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0001,
      "loss": 0.2537,
      "step": 315
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0001,
      "loss": 0.2345,
      "step": 320
    },
    {
      "epoch": 0.052,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0001,
      "loss": 0.2149,
      "step": 325
    },
    {
      "epoch": 0.0528,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0001,
      "loss": 0.2683,
      "step": 330
    },
    {
      "epoch": 0.0536,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0001,
      "loss": 0.2958,
      "step": 335
    },
    {
      "epoch": 0.0544,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0001,
      "loss": 0.2586,
      "step": 340
    },
    {
      "epoch": 0.0552,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.2361,
      "step": 345
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0001,
      "loss": 0.1915,
      "step": 350
    },
    {
      "epoch": 0.0568,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0001,
      "loss": 0.1489,
      "step": 355
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0001,
      "loss": 0.1857,
      "step": 360
    },
    {
      "epoch": 0.0584,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0001,
      "loss": 0.242,
      "step": 365
    },
    {
      "epoch": 0.0592,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0001,
      "loss": 0.3113,
      "step": 370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.2231,
      "step": 375
    },
    {
      "epoch": 0.0608,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0001,
      "loss": 0.22,
      "step": 380
    },
    {
      "epoch": 0.0616,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0001,
      "loss": 0.2711,
      "step": 385
    },
    {
      "epoch": 0.0624,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0001,
      "loss": 0.2025,
      "step": 390
    },
    {
      "epoch": 0.0632,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.2059,
      "step": 395
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.1581,
      "step": 400
    },
    {
      "epoch": 0.0648,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001,
      "loss": 0.1497,
      "step": 405
    },
    {
      "epoch": 0.0656,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0001,
      "loss": 0.2197,
      "step": 410
    },
    {
      "epoch": 0.0664,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0001,
      "loss": 0.2426,
      "step": 415
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0001,
      "loss": 0.1955,
      "step": 420
    },
    {
      "epoch": 0.068,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0001,
      "loss": 0.2446,
      "step": 425
    },
    {
      "epoch": 0.0688,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0001,
      "loss": 0.2525,
      "step": 430
    },
    {
      "epoch": 0.0696,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.1754,
      "step": 435
    },
    {
      "epoch": 0.0704,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0001,
      "loss": 0.2098,
      "step": 440
    },
    {
      "epoch": 0.0712,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0001,
      "loss": 0.2214,
      "step": 445
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0001,
      "loss": 0.2192,
      "step": 450
    },
    {
      "epoch": 0.0728,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001,
      "loss": 0.1476,
      "step": 455
    },
    {
      "epoch": 0.0736,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0001,
      "loss": 0.2621,
      "step": 460
    },
    {
      "epoch": 0.0744,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0001,
      "loss": 0.2053,
      "step": 465
    },
    {
      "epoch": 0.0752,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0001,
      "loss": 0.2702,
      "step": 470
    },
    {
      "epoch": 0.076,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0001,
      "loss": 0.2082,
      "step": 475
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0001,
      "loss": 0.1835,
      "step": 480
    },
    {
      "epoch": 0.0776,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0001,
      "loss": 0.2244,
      "step": 485
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0001,
      "loss": 0.2196,
      "step": 490
    },
    {
      "epoch": 0.0792,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0001,
      "loss": 0.2414,
      "step": 495
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0001,
      "loss": 0.1764,
      "step": 500
    },
    {
      "epoch": 0.0808,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0001,
      "loss": 0.1181,
      "step": 505
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0001,
      "loss": 0.2169,
      "step": 510
    },
    {
      "epoch": 0.0824,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.211,
      "step": 515
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0001,
      "loss": 0.236,
      "step": 520
    },
    {
      "epoch": 0.084,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001,
      "loss": 0.2106,
      "step": 525
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0001,
      "loss": 0.2374,
      "step": 530
    },
    {
      "epoch": 0.0856,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0001,
      "loss": 0.2146,
      "step": 535
    },
    {
      "epoch": 0.0864,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0001,
      "loss": 0.1817,
      "step": 540
    },
    {
      "epoch": 0.0872,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0001,
      "loss": 0.1903,
      "step": 545
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.11083984375,
      "learning_rate": 0.0001,
      "loss": 0.1787,
      "step": 550
    },
    {
      "epoch": 0.0888,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0001,
      "loss": 0.11,
      "step": 555
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0001,
      "loss": 0.2128,
      "step": 560
    },
    {
      "epoch": 0.0904,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0001,
      "loss": 0.2201,
      "step": 565
    },
    {
      "epoch": 0.0912,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0001,
      "loss": 0.2504,
      "step": 570
    },
    {
      "epoch": 0.092,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0001,
      "loss": 0.2184,
      "step": 575
    },
    {
      "epoch": 0.0928,
      "grad_norm": 0.08203125,
      "learning_rate": 0.0001,
      "loss": 0.2422,
      "step": 580
    },
    {
      "epoch": 0.0936,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0001,
      "loss": 0.2704,
      "step": 585
    },
    {
      "epoch": 0.0944,
      "grad_norm": 0.10302734375,
      "learning_rate": 0.0001,
      "loss": 0.2308,
      "step": 590
    },
    {
      "epoch": 0.0952,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0001,
      "loss": 0.2558,
      "step": 595
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001,
      "loss": 0.1417,
      "step": 600
    },
    {
      "epoch": 0.0968,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001,
      "loss": 0.1507,
      "step": 605
    },
    {
      "epoch": 0.0976,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0001,
      "loss": 0.1891,
      "step": 610
    },
    {
      "epoch": 0.0984,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.2159,
      "step": 615
    },
    {
      "epoch": 0.0992,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0001,
      "loss": 0.2256,
      "step": 620
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0001,
      "loss": 0.2233,
      "step": 625
    },
    {
      "epoch": 0.1008,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0001,
      "loss": 0.2395,
      "step": 630
    },
    {
      "epoch": 0.1016,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.2459,
      "step": 635
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0001,
      "loss": 0.203,
      "step": 640
    },
    {
      "epoch": 0.1032,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001,
      "loss": 0.1824,
      "step": 645
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0001,
      "loss": 0.1804,
      "step": 650
    },
    {
      "epoch": 0.1048,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0001,
      "loss": 0.1524,
      "step": 655
    },
    {
      "epoch": 0.1056,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0001,
      "loss": 0.2063,
      "step": 660
    },
    {
      "epoch": 0.1064,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.2021,
      "step": 665
    },
    {
      "epoch": 0.1072,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0001,
      "loss": 0.2325,
      "step": 670
    },
    {
      "epoch": 0.108,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0001,
      "loss": 0.2106,
      "step": 675
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.1781,
      "step": 680
    },
    {
      "epoch": 0.1096,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0001,
      "loss": 0.2596,
      "step": 685
    },
    {
      "epoch": 0.1104,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.2347,
      "step": 690
    },
    {
      "epoch": 0.1112,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0001,
      "loss": 0.2415,
      "step": 695
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0001,
      "loss": 0.2692,
      "step": 700
    },
    {
      "epoch": 0.1128,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0001,
      "loss": 0.1283,
      "step": 705
    },
    {
      "epoch": 0.1136,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.1653,
      "step": 710
    },
    {
      "epoch": 0.1144,
      "grad_norm": 0.171875,
      "learning_rate": 0.0001,
      "loss": 0.2332,
      "step": 715
    },
    {
      "epoch": 0.1152,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.2444,
      "step": 720
    },
    {
      "epoch": 0.116,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.205,
      "step": 725
    },
    {
      "epoch": 0.1168,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0001,
      "loss": 0.1439,
      "step": 730
    },
    {
      "epoch": 0.1176,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0001,
      "loss": 0.1817,
      "step": 735
    },
    {
      "epoch": 0.1184,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.0001,
      "loss": 0.237,
      "step": 740
    },
    {
      "epoch": 0.1192,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0001,
      "loss": 0.213,
      "step": 745
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001,
      "loss": 0.1995,
      "step": 750
    },
    {
      "epoch": 0.1208,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0001,
      "loss": 0.1177,
      "step": 755
    },
    {
      "epoch": 0.1216,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0001,
      "loss": 0.2337,
      "step": 760
    },
    {
      "epoch": 0.1224,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001,
      "loss": 0.2004,
      "step": 765
    },
    {
      "epoch": 0.1232,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0001,
      "loss": 0.1895,
      "step": 770
    },
    {
      "epoch": 0.124,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0001,
      "loss": 0.2361,
      "step": 775
    },
    {
      "epoch": 0.1248,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0001,
      "loss": 0.2351,
      "step": 780
    },
    {
      "epoch": 0.1256,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001,
      "loss": 0.1844,
      "step": 785
    },
    {
      "epoch": 0.1264,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0001,
      "loss": 0.2272,
      "step": 790
    },
    {
      "epoch": 0.1272,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0001,
      "loss": 0.2047,
      "step": 795
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.1463,
      "step": 800
    },
    {
      "epoch": 0.1288,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0001,
      "loss": 0.089,
      "step": 805
    },
    {
      "epoch": 0.1296,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.1946,
      "step": 810
    },
    {
      "epoch": 0.1304,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.2262,
      "step": 815
    },
    {
      "epoch": 0.1312,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001,
      "loss": 0.2472,
      "step": 820
    },
    {
      "epoch": 0.132,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0001,
      "loss": 0.1863,
      "step": 825
    },
    {
      "epoch": 0.1328,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0001,
      "loss": 0.1755,
      "step": 830
    },
    {
      "epoch": 0.1336,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0001,
      "loss": 0.2078,
      "step": 835
    },
    {
      "epoch": 0.1344,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.1884,
      "step": 840
    },
    {
      "epoch": 0.1352,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.2245,
      "step": 845
    },
    {
      "epoch": 0.136,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0001,
      "loss": 0.2148,
      "step": 850
    },
    {
      "epoch": 0.1368,
      "grad_norm": 0.1083984375,
      "learning_rate": 0.0001,
      "loss": 0.1205,
      "step": 855
    },
    {
      "epoch": 0.1376,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001,
      "loss": 0.2363,
      "step": 860
    },
    {
      "epoch": 0.1384,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0001,
      "loss": 0.2225,
      "step": 865
    },
    {
      "epoch": 0.1392,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0001,
      "loss": 0.1773,
      "step": 870
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001,
      "loss": 0.2088,
      "step": 875
    },
    {
      "epoch": 0.1408,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001,
      "loss": 0.2136,
      "step": 880
    },
    {
      "epoch": 0.1416,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0001,
      "loss": 0.1995,
      "step": 885
    },
    {
      "epoch": 0.1424,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001,
      "loss": 0.2461,
      "step": 890
    },
    {
      "epoch": 0.1432,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0001,
      "loss": 0.1903,
      "step": 895
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.11181640625,
      "learning_rate": 0.0001,
      "loss": 0.2139,
      "step": 900
    },
    {
      "epoch": 0.1448,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0001,
      "loss": 0.116,
      "step": 905
    },
    {
      "epoch": 0.1456,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0001,
      "loss": 0.2221,
      "step": 910
    },
    {
      "epoch": 0.1464,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.2111,
      "step": 915
    },
    {
      "epoch": 0.1472,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0001,
      "loss": 0.189,
      "step": 920
    },
    {
      "epoch": 0.148,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.2129,
      "step": 925
    },
    {
      "epoch": 0.1488,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0001,
      "loss": 0.2337,
      "step": 930
    },
    {
      "epoch": 0.1496,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0001,
      "loss": 0.1934,
      "step": 935
    },
    {
      "epoch": 0.1504,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0001,
      "loss": 0.2407,
      "step": 940
    },
    {
      "epoch": 0.1512,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0001,
      "loss": 0.1835,
      "step": 945
    },
    {
      "epoch": 0.152,
      "grad_norm": 0.07666015625,
      "learning_rate": 0.0001,
      "loss": 0.1069,
      "step": 950
    },
    {
      "epoch": 0.1528,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0001,
      "loss": 0.1285,
      "step": 955
    },
    {
      "epoch": 0.1536,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001,
      "loss": 0.1667,
      "step": 960
    },
    {
      "epoch": 0.1544,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0001,
      "loss": 0.2205,
      "step": 965
    },
    {
      "epoch": 0.1552,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0001,
      "loss": 0.1916,
      "step": 970
    },
    {
      "epoch": 0.156,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0001,
      "loss": 0.1922,
      "step": 975
    },
    {
      "epoch": 0.1568,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0001,
      "loss": 0.1807,
      "step": 980
    },
    {
      "epoch": 0.1576,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0001,
      "loss": 0.2092,
      "step": 985
    },
    {
      "epoch": 0.1584,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0001,
      "loss": 0.1886,
      "step": 990
    },
    {
      "epoch": 0.1592,
      "grad_norm": 0.08740234375,
      "learning_rate": 0.0001,
      "loss": 0.1895,
      "step": 995
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.1643,
      "step": 1000
    },
    {
      "epoch": 0.1608,
      "grad_norm": 0.099609375,
      "learning_rate": 0.0001,
      "loss": 0.14,
      "step": 1005
    },
    {
      "epoch": 0.1616,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0001,
      "loss": 0.1674,
      "step": 1010
    },
    {
      "epoch": 0.1624,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001,
      "loss": 0.2188,
      "step": 1015
    },
    {
      "epoch": 0.1632,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0001,
      "loss": 0.1951,
      "step": 1020
    },
    {
      "epoch": 0.164,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.2139,
      "step": 1025
    },
    {
      "epoch": 0.1648,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0001,
      "loss": 0.2094,
      "step": 1030
    },
    {
      "epoch": 0.1656,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001,
      "loss": 0.2431,
      "step": 1035
    },
    {
      "epoch": 0.1664,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0001,
      "loss": 0.2083,
      "step": 1040
    },
    {
      "epoch": 0.1672,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0001,
      "loss": 0.2161,
      "step": 1045
    },
    {
      "epoch": 0.168,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0001,
      "loss": 0.2044,
      "step": 1050
    },
    {
      "epoch": 0.1688,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.1478,
      "step": 1055
    },
    {
      "epoch": 0.1696,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0001,
      "loss": 0.1843,
      "step": 1060
    },
    {
      "epoch": 0.1704,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0001,
      "loss": 0.1918,
      "step": 1065
    },
    {
      "epoch": 0.1712,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001,
      "loss": 0.2203,
      "step": 1070
    },
    {
      "epoch": 0.172,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.2498,
      "step": 1075
    },
    {
      "epoch": 0.1728,
      "grad_norm": 0.09375,
      "learning_rate": 0.0001,
      "loss": 0.2503,
      "step": 1080
    },
    {
      "epoch": 0.1736,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0001,
      "loss": 0.17,
      "step": 1085
    },
    {
      "epoch": 0.1744,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0001,
      "loss": 0.2321,
      "step": 1090
    },
    {
      "epoch": 0.1752,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001,
      "loss": 0.2303,
      "step": 1095
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0001,
      "loss": 0.1653,
      "step": 1100
    },
    {
      "epoch": 0.1768,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001,
      "loss": 0.1229,
      "step": 1105
    },
    {
      "epoch": 0.1776,
      "grad_norm": 0.11376953125,
      "learning_rate": 0.0001,
      "loss": 0.1924,
      "step": 1110
    },
    {
      "epoch": 0.1784,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0001,
      "loss": 0.1645,
      "step": 1115
    },
    {
      "epoch": 0.1792,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0001,
      "loss": 0.2007,
      "step": 1120
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.0810546875,
      "learning_rate": 0.0001,
      "loss": 0.184,
      "step": 1125
    },
    {
      "epoch": 0.1808,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0001,
      "loss": 0.1889,
      "step": 1130
    },
    {
      "epoch": 0.1816,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.1633,
      "step": 1135
    },
    {
      "epoch": 0.1824,
      "grad_norm": 0.203125,
      "learning_rate": 0.0001,
      "loss": 0.2242,
      "step": 1140
    },
    {
      "epoch": 0.1832,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0001,
      "loss": 0.2061,
      "step": 1145
    },
    {
      "epoch": 0.184,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.1419,
      "step": 1150
    },
    {
      "epoch": 0.1848,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001,
      "loss": 0.1267,
      "step": 1155
    },
    {
      "epoch": 0.1856,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001,
      "loss": 0.1677,
      "step": 1160
    },
    {
      "epoch": 0.1864,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0001,
      "loss": 0.1936,
      "step": 1165
    },
    {
      "epoch": 0.1872,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0001,
      "loss": 0.1968,
      "step": 1170
    },
    {
      "epoch": 0.188,
      "grad_norm": 0.087890625,
      "learning_rate": 0.0001,
      "loss": 0.2463,
      "step": 1175
    },
    {
      "epoch": 0.1888,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.1803,
      "step": 1180
    },
    {
      "epoch": 0.1896,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0001,
      "loss": 0.2359,
      "step": 1185
    },
    {
      "epoch": 0.1904,
      "grad_norm": 0.09521484375,
      "learning_rate": 0.0001,
      "loss": 0.2117,
      "step": 1190
    },
    {
      "epoch": 0.1912,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0001,
      "loss": 0.1813,
      "step": 1195
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0001,
      "loss": 0.1761,
      "step": 1200
    },
    {
      "epoch": 0.1928,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001,
      "loss": 0.1248,
      "step": 1205
    },
    {
      "epoch": 0.1936,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0001,
      "loss": 0.2011,
      "step": 1210
    },
    {
      "epoch": 0.1944,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001,
      "loss": 0.1619,
      "step": 1215
    },
    {
      "epoch": 0.1952,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0001,
      "loss": 0.1991,
      "step": 1220
    },
    {
      "epoch": 0.196,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0001,
      "loss": 0.1931,
      "step": 1225
    },
    {
      "epoch": 0.1968,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001,
      "loss": 0.2103,
      "step": 1230
    },
    {
      "epoch": 0.1976,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0001,
      "loss": 0.1969,
      "step": 1235
    },
    {
      "epoch": 0.1984,
      "grad_norm": 0.083984375,
      "learning_rate": 0.0001,
      "loss": 0.1895,
      "step": 1240
    },
    {
      "epoch": 0.1992,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001,
      "loss": 0.2793,
      "step": 1245
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0001,
      "loss": 0.1517,
      "step": 1250
    },
    {
      "epoch": 0.2008,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001,
      "loss": 0.1037,
      "step": 1255
    },
    {
      "epoch": 0.2016,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001,
      "loss": 0.2392,
      "step": 1260
    },
    {
      "epoch": 0.2024,
      "grad_norm": 0.11328125,
      "learning_rate": 0.0001,
      "loss": 0.1847,
      "step": 1265
    },
    {
      "epoch": 0.2032,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0001,
      "loss": 0.2304,
      "step": 1270
    },
    {
      "epoch": 0.204,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0001,
      "loss": 0.2136,
      "step": 1275
    },
    {
      "epoch": 0.2048,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0001,
      "loss": 0.2052,
      "step": 1280
    },
    {
      "epoch": 0.2056,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0001,
      "loss": 0.1632,
      "step": 1285
    },
    {
      "epoch": 0.2064,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0001,
      "loss": 0.2417,
      "step": 1290
    },
    {
      "epoch": 0.2072,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0001,
      "loss": 0.2001,
      "step": 1295
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001,
      "loss": 0.1966,
      "step": 1300
    },
    {
      "epoch": 0.2088,
      "grad_norm": 0.1025390625,
      "learning_rate": 0.0001,
      "loss": 0.1319,
      "step": 1305
    },
    {
      "epoch": 0.2096,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0001,
      "loss": 0.1974,
      "step": 1310
    },
    {
      "epoch": 0.2104,
      "grad_norm": 0.1103515625,
      "learning_rate": 0.0001,
      "loss": 0.227,
      "step": 1315
    },
    {
      "epoch": 0.2112,
      "grad_norm": 0.1171875,
      "learning_rate": 0.0001,
      "loss": 0.1952,
      "step": 1320
    },
    {
      "epoch": 0.212,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0001,
      "loss": 0.1899,
      "step": 1325
    },
    {
      "epoch": 0.2128,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.2118,
      "step": 1330
    },
    {
      "epoch": 0.2136,
      "grad_norm": 0.10205078125,
      "learning_rate": 0.0001,
      "loss": 0.1975,
      "step": 1335
    },
    {
      "epoch": 0.2144,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0001,
      "loss": 0.2139,
      "step": 1340
    },
    {
      "epoch": 0.2152,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0001,
      "loss": 0.2306,
      "step": 1345
    },
    {
      "epoch": 0.216,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001,
      "loss": 0.1734,
      "step": 1350
    },
    {
      "epoch": 0.2168,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.1343,
      "step": 1355
    },
    {
      "epoch": 0.2176,
      "grad_norm": 0.1015625,
      "learning_rate": 0.0001,
      "loss": 0.1557,
      "step": 1360
    },
    {
      "epoch": 0.2184,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.2053,
      "step": 1365
    },
    {
      "epoch": 0.2192,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0001,
      "loss": 0.2094,
      "step": 1370
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0001,
      "loss": 0.2023,
      "step": 1375
    },
    {
      "epoch": 0.2208,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0001,
      "loss": 0.2249,
      "step": 1380
    },
    {
      "epoch": 0.2216,
      "grad_norm": 0.091796875,
      "learning_rate": 0.0001,
      "loss": 0.1874,
      "step": 1385
    },
    {
      "epoch": 0.2224,
      "grad_norm": 0.10107421875,
      "learning_rate": 0.0001,
      "loss": 0.153,
      "step": 1390
    },
    {
      "epoch": 0.2232,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0001,
      "loss": 0.1825,
      "step": 1395
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0001,
      "loss": 0.1931,
      "step": 1400
    },
    {
      "epoch": 0.2248,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.0736,
      "step": 1405
    },
    {
      "epoch": 0.2256,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0001,
      "loss": 0.1504,
      "step": 1410
    },
    {
      "epoch": 0.2264,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.2098,
      "step": 1415
    },
    {
      "epoch": 0.2272,
      "grad_norm": 0.11767578125,
      "learning_rate": 0.0001,
      "loss": 0.2176,
      "step": 1420
    },
    {
      "epoch": 0.228,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0001,
      "loss": 0.2271,
      "step": 1425
    },
    {
      "epoch": 0.2288,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0001,
      "loss": 0.2143,
      "step": 1430
    },
    {
      "epoch": 0.2296,
      "grad_norm": 0.10595703125,
      "learning_rate": 0.0001,
      "loss": 0.1632,
      "step": 1435
    },
    {
      "epoch": 0.2304,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0001,
      "loss": 0.2102,
      "step": 1440
    },
    {
      "epoch": 0.2312,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0001,
      "loss": 0.2016,
      "step": 1445
    },
    {
      "epoch": 0.232,
      "grad_norm": 0.08984375,
      "learning_rate": 0.0001,
      "loss": 0.1495,
      "step": 1450
    },
    {
      "epoch": 0.2328,
      "grad_norm": 0.109375,
      "learning_rate": 0.0001,
      "loss": 0.1122,
      "step": 1455
    },
    {
      "epoch": 0.2336,
      "grad_norm": 0.11279296875,
      "learning_rate": 0.0001,
      "loss": 0.1675,
      "step": 1460
    },
    {
      "epoch": 0.2344,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0001,
      "loss": 0.2246,
      "step": 1465
    },
    {
      "epoch": 0.2352,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0001,
      "loss": 0.2061,
      "step": 1470
    },
    {
      "epoch": 0.236,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001,
      "loss": 0.2036,
      "step": 1475
    },
    {
      "epoch": 0.2368,
      "grad_norm": 0.1044921875,
      "learning_rate": 0.0001,
      "loss": 0.1993,
      "step": 1480
    },
    {
      "epoch": 0.2376,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0001,
      "loss": 0.1856,
      "step": 1485
    },
    {
      "epoch": 0.2384,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0001,
      "loss": 0.1948,
      "step": 1490
    },
    {
      "epoch": 0.2392,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0001,
      "loss": 0.2294,
      "step": 1495
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0001,
      "loss": 0.1499,
      "step": 1500
    },
    {
      "epoch": 0.2408,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0001,
      "loss": 0.0981,
      "step": 1505
    },
    {
      "epoch": 0.2416,
      "grad_norm": 0.115234375,
      "learning_rate": 0.0001,
      "loss": 0.1912,
      "step": 1510
    },
    {
      "epoch": 0.2424,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001,
      "loss": 0.2121,
      "step": 1515
    },
    {
      "epoch": 0.2432,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001,
      "loss": 0.2284,
      "step": 1520
    },
    {
      "epoch": 0.244,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0001,
      "loss": 0.1754,
      "step": 1525
    },
    {
      "epoch": 0.2448,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.1756,
      "step": 1530
    },
    {
      "epoch": 0.2456,
      "grad_norm": 0.1064453125,
      "learning_rate": 0.0001,
      "loss": 0.2191,
      "step": 1535
    },
    {
      "epoch": 0.2464,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.1815,
      "step": 1540
    },
    {
      "epoch": 0.2472,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001,
      "loss": 0.1914,
      "step": 1545
    },
    {
      "epoch": 0.248,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0001,
      "loss": 0.1681,
      "step": 1550
    },
    {
      "epoch": 0.2488,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0001,
      "loss": 0.119,
      "step": 1555
    },
    {
      "epoch": 0.2496,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0001,
      "loss": 0.2339,
      "step": 1560
    },
    {
      "epoch": 0.2504,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0001,
      "loss": 0.1289,
      "step": 1565
    },
    {
      "epoch": 0.2512,
      "grad_norm": 0.10498046875,
      "learning_rate": 0.0001,
      "loss": 0.1783,
      "step": 1570
    },
    {
      "epoch": 0.252,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0001,
      "loss": 0.2125,
      "step": 1575
    },
    {
      "epoch": 0.2528,
      "grad_norm": 0.062255859375,
      "learning_rate": 0.0001,
      "loss": 0.2068,
      "step": 1580
    },
    {
      "epoch": 0.2536,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0001,
      "loss": 0.2065,
      "step": 1585
    },
    {
      "epoch": 0.2544,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0001,
      "loss": 0.1683,
      "step": 1590
    },
    {
      "epoch": 0.2552,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0001,
      "loss": 0.1879,
      "step": 1595
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0001,
      "loss": 0.1538,
      "step": 1600
    },
    {
      "epoch": 0.2568,
      "grad_norm": 0.0986328125,
      "learning_rate": 0.0001,
      "loss": 0.1126,
      "step": 1605
    },
    {
      "epoch": 0.2576,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0001,
      "loss": 0.2294,
      "step": 1610
    },
    {
      "epoch": 0.2584,
      "grad_norm": 0.10888671875,
      "learning_rate": 0.0001,
      "loss": 0.1845,
      "step": 1615
    },
    {
      "epoch": 0.2592,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001,
      "loss": 0.1437,
      "step": 1620
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0001,
      "loss": 0.1944,
      "step": 1625
    },
    {
      "epoch": 0.2608,
      "grad_norm": 0.107421875,
      "learning_rate": 0.0001,
      "loss": 0.1914,
      "step": 1630
    },
    {
      "epoch": 0.2616,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001,
      "loss": 0.25,
      "step": 1635
    },
    {
      "epoch": 0.2624,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001,
      "loss": 0.2547,
      "step": 1640
    },
    {
      "epoch": 0.2632,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0001,
      "loss": 0.252,
      "step": 1645
    },
    {
      "epoch": 0.264,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0001,
      "loss": 0.1843,
      "step": 1650
    },
    {
      "epoch": 0.2648,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0001,
      "loss": 0.0962,
      "step": 1655
    },
    {
      "epoch": 0.2656,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0001,
      "loss": 0.1539,
      "step": 1660
    },
    {
      "epoch": 0.2664,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0001,
      "loss": 0.2248,
      "step": 1665
    },
    {
      "epoch": 0.2672,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0001,
      "loss": 0.204,
      "step": 1670
    },
    {
      "epoch": 0.268,
      "grad_norm": 0.0966796875,
      "learning_rate": 0.0001,
      "loss": 0.2077,
      "step": 1675
    },
    {
      "epoch": 0.2688,
      "grad_norm": 0.109375,
      "learning_rate": 0.0001,
      "loss": 0.2036,
      "step": 1680
    },
    {
      "epoch": 0.2696,
      "grad_norm": 0.0947265625,
      "learning_rate": 0.0001,
      "loss": 0.1975,
      "step": 1685
    },
    {
      "epoch": 0.2704,
      "grad_norm": 0.1142578125,
      "learning_rate": 0.0001,
      "loss": 0.2118,
      "step": 1690
    },
    {
      "epoch": 0.2712,
      "grad_norm": 0.09375,
      "learning_rate": 0.0001,
      "loss": 0.2242,
      "step": 1695
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0001,
      "loss": 0.1279,
      "step": 1700
    },
    {
      "epoch": 0.2728,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0001,
      "loss": 0.1496,
      "step": 1705
    },
    {
      "epoch": 0.2736,
      "grad_norm": 0.1162109375,
      "learning_rate": 0.0001,
      "loss": 0.177,
      "step": 1710
    },
    {
      "epoch": 0.2744,
      "grad_norm": 0.1875,
      "learning_rate": 0.0001,
      "loss": 0.1661,
      "step": 1715
    },
    {
      "epoch": 0.2752,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.1542,
      "step": 1720
    },
    {
      "epoch": 0.276,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0001,
      "loss": 0.2083,
      "step": 1725
    },
    {
      "epoch": 0.2768,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0001,
      "loss": 0.1717,
      "step": 1730
    },
    {
      "epoch": 0.2776,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0001,
      "loss": 0.2311,
      "step": 1735
    },
    {
      "epoch": 0.2784,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0001,
      "loss": 0.2232,
      "step": 1740
    },
    {
      "epoch": 0.2792,
      "grad_norm": 0.140625,
      "learning_rate": 0.0001,
      "loss": 0.206,
      "step": 1745
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0001,
      "loss": 0.1293,
      "step": 1750
    },
    {
      "epoch": 0.2808,
      "grad_norm": 0.15625,
      "learning_rate": 0.0001,
      "loss": 0.1173,
      "step": 1755
    },
    {
      "epoch": 0.2816,
      "grad_norm": 0.1123046875,
      "learning_rate": 0.0001,
      "loss": 0.1973,
      "step": 1760
    },
    {
      "epoch": 0.2824,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001,
      "loss": 0.1764,
      "step": 1765
    },
    {
      "epoch": 0.2832,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0001,
      "loss": 0.1942,
      "step": 1770
    },
    {
      "epoch": 0.284,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0001,
      "loss": 0.2328,
      "step": 1775
    },
    {
      "epoch": 0.2848,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0001,
      "loss": 0.1835,
      "step": 1780
    },
    {
      "epoch": 0.2856,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0001,
      "loss": 0.2094,
      "step": 1785
    },
    {
      "epoch": 0.2864,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.2041,
      "step": 1790
    },
    {
      "epoch": 0.2872,
      "grad_norm": 0.10986328125,
      "learning_rate": 0.0001,
      "loss": 0.2374,
      "step": 1795
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.0869140625,
      "learning_rate": 0.0001,
      "loss": 0.1292,
      "step": 1800
    },
    {
      "epoch": 0.2888,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0001,
      "loss": 0.1566,
      "step": 1805
    },
    {
      "epoch": 0.2896,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0001,
      "loss": 0.1787,
      "step": 1810
    },
    {
      "epoch": 0.2904,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.2196,
      "step": 1815
    },
    {
      "epoch": 0.2912,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0001,
      "loss": 0.2175,
      "step": 1820
    },
    {
      "epoch": 0.292,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0001,
      "loss": 0.2285,
      "step": 1825
    },
    {
      "epoch": 0.2928,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0001,
      "loss": 0.2181,
      "step": 1830
    },
    {
      "epoch": 0.2936,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0001,
      "loss": 0.2023,
      "step": 1835
    },
    {
      "epoch": 0.2944,
      "grad_norm": 0.111328125,
      "learning_rate": 0.0001,
      "loss": 0.1639,
      "step": 1840
    },
    {
      "epoch": 0.2952,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0001,
      "loss": 0.1527,
      "step": 1845
    },
    {
      "epoch": 0.296,
      "grad_norm": 0.054931640625,
      "learning_rate": 0.0001,
      "loss": 0.1626,
      "step": 1850
    },
    {
      "epoch": 0.2968,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0001,
      "loss": 0.1281,
      "step": 1855
    },
    {
      "epoch": 0.2976,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0001,
      "loss": 0.1688,
      "step": 1860
    },
    {
      "epoch": 0.2984,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0001,
      "loss": 0.206,
      "step": 1865
    },
    {
      "epoch": 0.2992,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0001,
      "loss": 0.1935,
      "step": 1870
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0001,
      "loss": 0.1631,
      "step": 1875
    },
    {
      "epoch": 0.3,
      "step": 1875,
      "total_flos": 7.012531902575002e+17,
      "train_loss": 0.22927836108207703,
      "train_runtime": 39002.362,
      "train_samples_per_second": 0.769,
      "train_steps_per_second": 0.048
    }
  ],
  "logging_steps": 5,
  "max_steps": 1875,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 90,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.012531902575002e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}