{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 10780,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.361328125,
      "learning_rate": 3.710575139146568e-07,
      "loss": 2.6427,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.34765625,
      "learning_rate": 1.855287569573284e-06,
      "loss": 2.6431,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.380859375,
      "learning_rate": 3.710575139146568e-06,
      "loss": 2.6904,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.369140625,
      "learning_rate": 5.565862708719852e-06,
      "loss": 2.643,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.359375,
      "learning_rate": 7.421150278293136e-06,
      "loss": 2.6806,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.337890625,
      "learning_rate": 9.276437847866419e-06,
      "loss": 2.6886,
      "step": 25
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.40234375,
      "learning_rate": 1.1131725417439704e-05,
      "loss": 2.6287,
      "step": 30
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.36328125,
      "learning_rate": 1.2987012987012986e-05,
      "loss": 2.5784,
      "step": 35
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.376953125,
      "learning_rate": 1.4842300556586271e-05,
      "loss": 2.5996,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3046875,
      "learning_rate": 1.6697588126159555e-05,
      "loss": 2.5966,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.31640625,
      "learning_rate": 1.8552875695732837e-05,
      "loss": 2.5816,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.314453125,
      "learning_rate": 2.0408163265306123e-05,
      "loss": 2.5522,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.3046875,
      "learning_rate": 2.2263450834879408e-05,
      "loss": 2.5633,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.328125,
      "learning_rate": 2.4118738404452693e-05,
      "loss": 2.5333,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.26171875,
      "learning_rate": 2.5974025974025972e-05,
      "loss": 2.4653,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2431640625,
      "learning_rate": 2.782931354359926e-05,
      "loss": 2.4254,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.2333984375,
      "learning_rate": 2.9684601113172543e-05,
      "loss": 2.4339,
      "step": 80
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.248046875,
      "learning_rate": 3.1539888682745825e-05,
      "loss": 2.4517,
      "step": 85
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.339517625231911e-05,
      "loss": 2.4258,
      "step": 90
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.220703125,
      "learning_rate": 3.5250463821892396e-05,
      "loss": 2.4284,
      "step": 95
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.23046875,
      "learning_rate": 3.7105751391465674e-05,
      "loss": 2.3756,
      "step": 100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.21484375,
      "learning_rate": 3.8961038961038966e-05,
      "loss": 2.3719,
      "step": 105
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.0816326530612245e-05,
      "loss": 2.3452,
      "step": 110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.203125,
      "learning_rate": 4.267161410018553e-05,
      "loss": 2.389,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.2080078125,
      "learning_rate": 4.4526901669758816e-05,
      "loss": 2.3233,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.201171875,
      "learning_rate": 4.6382189239332094e-05,
      "loss": 2.3368,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.823747680890539e-05,
      "loss": 2.2901,
      "step": 130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.212890625,
      "learning_rate": 5.0092764378478665e-05,
      "loss": 2.3386,
      "step": 135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.1948051948051944e-05,
      "loss": 2.3245,
      "step": 140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.380333951762524e-05,
      "loss": 2.3099,
      "step": 145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1787109375,
      "learning_rate": 5.565862708719852e-05,
      "loss": 2.2899,
      "step": 150
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1796875,
      "learning_rate": 5.751391465677181e-05,
      "loss": 2.2857,
      "step": 155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.185546875,
      "learning_rate": 5.9369202226345086e-05,
      "loss": 2.2968,
      "step": 160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.197265625,
      "learning_rate": 6.122448979591838e-05,
      "loss": 2.2657,
      "step": 165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1884765625,
      "learning_rate": 6.307977736549165e-05,
      "loss": 2.3,
      "step": 170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1875,
      "learning_rate": 6.493506493506494e-05,
      "loss": 2.2807,
      "step": 175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.208984375,
      "learning_rate": 6.679035250463822e-05,
      "loss": 2.257,
      "step": 180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.1962890625,
      "learning_rate": 6.86456400742115e-05,
      "loss": 2.2708,
      "step": 185
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1787109375,
      "learning_rate": 7.050092764378479e-05,
      "loss": 2.273,
      "step": 190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1875,
      "learning_rate": 7.235621521335806e-05,
      "loss": 2.2603,
      "step": 195
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1904296875,
      "learning_rate": 7.421150278293135e-05,
      "loss": 2.2514,
      "step": 200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2119140625,
      "learning_rate": 7.606679035250465e-05,
      "loss": 2.2908,
      "step": 205
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1767578125,
      "learning_rate": 7.792207792207793e-05,
      "loss": 2.2708,
      "step": 210
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1982421875,
      "learning_rate": 7.977736549165122e-05,
      "loss": 2.2279,
      "step": 215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.19921875,
      "learning_rate": 8.163265306122449e-05,
      "loss": 2.2627,
      "step": 220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2099609375,
      "learning_rate": 8.348794063079778e-05,
      "loss": 2.2642,
      "step": 225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2041015625,
      "learning_rate": 8.534322820037106e-05,
      "loss": 2.2373,
      "step": 230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.2001953125,
      "learning_rate": 8.719851576994435e-05,
      "loss": 2.2572,
      "step": 235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.1953125,
      "learning_rate": 8.905380333951763e-05,
      "loss": 2.2105,
      "step": 240
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1953125,
      "learning_rate": 9.090909090909092e-05,
      "loss": 2.2398,
      "step": 245
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.203125,
      "learning_rate": 9.276437847866419e-05,
      "loss": 2.2448,
      "step": 250
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.203125,
      "learning_rate": 9.461966604823747e-05,
      "loss": 2.2113,
      "step": 255
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2021484375,
      "learning_rate": 9.647495361781077e-05,
      "loss": 2.2631,
      "step": 260
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.19921875,
      "learning_rate": 9.833024118738406e-05,
      "loss": 2.2716,
      "step": 265
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00010018552875695733,
      "loss": 2.2368,
      "step": 270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00010204081632653062,
      "loss": 2.2297,
      "step": 275
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00010389610389610389,
      "loss": 2.2706,
      "step": 280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00010575139146567719,
      "loss": 2.254,
      "step": 285
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00010760667903525049,
      "loss": 2.2576,
      "step": 290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00010946196660482376,
      "loss": 2.2456,
      "step": 295
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00011131725417439704,
      "loss": 2.2665,
      "step": 300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00011317254174397031,
      "loss": 2.2529,
      "step": 305
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00011502782931354361,
      "loss": 2.2427,
      "step": 310
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00011688311688311689,
      "loss": 2.2489,
      "step": 315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.203125,
      "learning_rate": 0.00011873840445269017,
      "loss": 2.2153,
      "step": 320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00012059369202226344,
      "loss": 2.2204,
      "step": 325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00012244897959183676,
      "loss": 2.2397,
      "step": 330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00012430426716141001,
      "loss": 2.2586,
      "step": 335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0001261595547309833,
      "loss": 2.2376,
      "step": 340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0001280148423005566,
      "loss": 2.1859,
      "step": 345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00012987012987012987,
      "loss": 2.2297,
      "step": 350
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.203125,
      "learning_rate": 0.00013172541743970318,
      "loss": 2.2543,
      "step": 355
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00013358070500927644,
      "loss": 2.2614,
      "step": 360
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00013543599257884973,
      "loss": 2.2152,
      "step": 365
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000137291280148423,
      "loss": 2.2712,
      "step": 370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0001391465677179963,
      "loss": 2.2363,
      "step": 375
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00014100185528756958,
      "loss": 2.2156,
      "step": 380
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00014285714285714287,
      "loss": 2.2721,
      "step": 385
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00014471243042671613,
      "loss": 2.2148,
      "step": 390
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00014656771799628944,
      "loss": 2.1739,
      "step": 395
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0001484230055658627,
      "loss": 2.2193,
      "step": 400
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.000150278293135436,
      "loss": 2.1901,
      "step": 405
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0001521335807050093,
      "loss": 2.2271,
      "step": 410
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00015398886827458255,
      "loss": 2.2502,
      "step": 415
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00015584415584415587,
      "loss": 2.2148,
      "step": 420
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00015769944341372912,
      "loss": 2.1976,
      "step": 425
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00015955473098330244,
      "loss": 2.1779,
      "step": 430
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0001614100185528757,
      "loss": 2.2163,
      "step": 435
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00016326530612244898,
      "loss": 2.2278,
      "step": 440
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00016512059369202227,
      "loss": 2.2018,
      "step": 445
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00016697588126159555,
      "loss": 2.2086,
      "step": 450
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00016883116883116884,
      "loss": 2.2486,
      "step": 455
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00017068645640074212,
      "loss": 2.2295,
      "step": 460
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0001725417439703154,
      "loss": 2.2318,
      "step": 465
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0001743970315398887,
      "loss": 2.2289,
      "step": 470
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00017625231910946198,
      "loss": 2.2378,
      "step": 475
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00017810760667903526,
      "loss": 2.1831,
      "step": 480
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00017996289424860855,
      "loss": 2.2105,
      "step": 485
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00018181818181818183,
      "loss": 2.2156,
      "step": 490
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00018367346938775512,
      "loss": 2.2045,
      "step": 495
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00018552875695732838,
      "loss": 2.2298,
      "step": 500
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0001873840445269017,
      "loss": 2.2342,
      "step": 505
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00018923933209647495,
      "loss": 2.2121,
      "step": 510
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00019109461966604823,
      "loss": 2.2169,
      "step": 515
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00019294990723562155,
      "loss": 2.2174,
      "step": 520
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0001948051948051948,
      "loss": 2.2328,
      "step": 525
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00019666048237476812,
      "loss": 2.2081,
      "step": 530
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00019851576994434138,
      "loss": 2.2411,
      "step": 535
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00020037105751391466,
      "loss": 2.208,
      "step": 540
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00020222634508348795,
      "loss": 2.1872,
      "step": 545
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00020408163265306123,
      "loss": 2.1829,
      "step": 550
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00020593692022263454,
      "loss": 2.2194,
      "step": 555
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00020779220779220778,
      "loss": 2.2283,
      "step": 560
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002096474953617811,
      "loss": 2.2119,
      "step": 565
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00021150278293135437,
      "loss": 2.2196,
      "step": 570
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00021335807050092766,
      "loss": 2.2378,
      "step": 575
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00021521335807050097,
      "loss": 2.2361,
      "step": 580
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002170686456400742,
      "loss": 2.2253,
      "step": 585
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00021892393320964752,
      "loss": 2.2111,
      "step": 590
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002207792207792208,
      "loss": 2.2417,
      "step": 595
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00022263450834879409,
      "loss": 2.1814,
      "step": 600
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00022448979591836734,
      "loss": 2.248,
      "step": 605
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00022634508348794063,
      "loss": 2.191,
      "step": 610
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00022820037105751392,
      "loss": 2.2258,
      "step": 615
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00023005565862708723,
      "loss": 2.226,
      "step": 620
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00023191094619666046,
      "loss": 2.231,
      "step": 625
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00023376623376623377,
      "loss": 2.2119,
      "step": 630
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00023562152133580706,
      "loss": 2.2206,
      "step": 635
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00023747680890538034,
      "loss": 2.2083,
      "step": 640
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00023933209647495365,
      "loss": 2.2206,
      "step": 645
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00024118738404452689,
      "loss": 2.1931,
      "step": 650
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002430426716141002,
      "loss": 2.1753,
      "step": 655
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002448979591836735,
      "loss": 2.1885,
      "step": 660
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002467532467532468,
      "loss": 2.2343,
      "step": 665
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00024860853432282003,
      "loss": 2.2425,
      "step": 670
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002504638218923933,
      "loss": 2.1929,
      "step": 675
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002523191094619666,
      "loss": 2.2569,
      "step": 680
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00025417439703153994,
      "loss": 2.2311,
      "step": 685
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0002560296846011132,
      "loss": 2.2218,
      "step": 690
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00025788497217068645,
      "loss": 2.2148,
      "step": 695
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00025974025974025974,
      "loss": 2.1973,
      "step": 700
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000261595547309833,
      "loss": 2.197,
      "step": 705
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00026345083487940637,
      "loss": 2.2127,
      "step": 710
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0002653061224489796,
      "loss": 2.1964,
      "step": 715
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.171875,
      "learning_rate": 0.0002671614100185529,
      "loss": 2.2008,
      "step": 720
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00026901669758812617,
      "loss": 2.1945,
      "step": 725
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00027087198515769945,
      "loss": 2.1954,
      "step": 730
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00027272727272727274,
      "loss": 2.2016,
      "step": 735
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000274582560296846,
      "loss": 2.257,
      "step": 740
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0002764378478664193,
      "loss": 2.2091,
      "step": 745
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0002782931354359926,
      "loss": 2.2242,
      "step": 750
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002801484230055659,
      "loss": 2.2002,
      "step": 755
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00028200371057513916,
      "loss": 2.2441,
      "step": 760
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.171875,
      "learning_rate": 0.00028385899814471245,
      "loss": 2.1937,
      "step": 765
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00028571428571428574,
      "loss": 2.2132,
      "step": 770
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000287569573283859,
      "loss": 2.2231,
      "step": 775
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00028942486085343225,
      "loss": 2.2246,
      "step": 780
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0002912801484230056,
      "loss": 2.2031,
      "step": 785
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0002931354359925789,
      "loss": 2.2084,
      "step": 790
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.171875,
      "learning_rate": 0.00029499072356215216,
      "loss": 2.2021,
      "step": 795
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0002968460111317254,
      "loss": 2.1771,
      "step": 800
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0002987012987012987,
      "loss": 2.2068,
      "step": 805
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000300556586270872,
      "loss": 2.1828,
      "step": 810
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003024118738404453,
      "loss": 2.2225,
      "step": 815
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003042671614100186,
      "loss": 2.2214,
      "step": 820
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0003061224489795918,
      "loss": 2.2285,
      "step": 825
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003079777365491651,
      "loss": 2.1673,
      "step": 830
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00030983302411873845,
      "loss": 2.2316,
      "step": 835
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00031168831168831173,
      "loss": 2.211,
      "step": 840
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1875,
      "learning_rate": 0.00031354359925788496,
      "loss": 2.2433,
      "step": 845
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00031539888682745825,
      "loss": 2.2117,
      "step": 850
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00031725417439703153,
      "loss": 2.2104,
      "step": 855
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003191094619666049,
      "loss": 2.2543,
      "step": 860
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00032096474953617816,
      "loss": 2.202,
      "step": 865
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003228200371057514,
      "loss": 2.2002,
      "step": 870
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003246753246753247,
      "loss": 2.2332,
      "step": 875
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00032653061224489796,
      "loss": 2.1782,
      "step": 880
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003283858998144713,
      "loss": 2.2428,
      "step": 885
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.171875,
      "learning_rate": 0.00033024118738404453,
      "loss": 2.2064,
      "step": 890
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003320964749536178,
      "loss": 2.2444,
      "step": 895
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003339517625231911,
      "loss": 2.2078,
      "step": 900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003358070500927644,
      "loss": 2.2068,
      "step": 905
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00033766233766233767,
      "loss": 2.1913,
      "step": 910
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00033951762523191096,
      "loss": 2.2534,
      "step": 915
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.17578125,
      "learning_rate": 0.00034137291280148424,
      "loss": 2.1718,
      "step": 920
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00034322820037105753,
      "loss": 2.2004,
      "step": 925
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003450834879406308,
      "loss": 2.2119,
      "step": 930
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0003469387755102041,
      "loss": 2.2186,
      "step": 935
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003487940630797774,
      "loss": 2.2306,
      "step": 940
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00035064935064935067,
      "loss": 2.234,
      "step": 945
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00035250463821892396,
      "loss": 2.1975,
      "step": 950
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00035435992578849724,
      "loss": 2.2149,
      "step": 955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003562152133580705,
      "loss": 2.1836,
      "step": 960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003580705009276438,
      "loss": 2.1872,
      "step": 965
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003599257884972171,
      "loss": 2.1991,
      "step": 970
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003617810760667904,
      "loss": 2.2465,
      "step": 975
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.166015625,
      "learning_rate": 0.00036363636363636367,
      "loss": 2.2186,
      "step": 980
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00036549165120593695,
      "loss": 2.2318,
      "step": 985
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00036734693877551024,
      "loss": 2.1858,
      "step": 990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0003692022263450835,
      "loss": 2.1947,
      "step": 995
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00037105751391465676,
      "loss": 2.2203,
      "step": 1000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00037291280148423004,
      "loss": 2.2217,
      "step": 1005
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003747680890538034,
      "loss": 2.2001,
      "step": 1010
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00037662337662337667,
      "loss": 2.2013,
      "step": 1015
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003784786641929499,
      "loss": 2.2198,
      "step": 1020
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003803339517625232,
      "loss": 2.2236,
      "step": 1025
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00038218923933209647,
      "loss": 2.2227,
      "step": 1030
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003840445269016698,
      "loss": 2.2275,
      "step": 1035
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003858998144712431,
      "loss": 2.2241,
      "step": 1040
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0003877551020408163,
      "loss": 2.2066,
      "step": 1045
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003896103896103896,
      "loss": 2.1842,
      "step": 1050
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003914656771799629,
      "loss": 2.1279,
      "step": 1055
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00039332096474953624,
      "loss": 2.1712,
      "step": 1060
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00039517625231910947,
      "loss": 2.1972,
      "step": 1065
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00039703153988868275,
      "loss": 2.2082,
      "step": 1070
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039888682745825604,
      "loss": 2.21,
      "step": 1075
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003999999580591546,
      "loss": 2.2126,
      "step": 1080
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039999948622484506,
      "loss": 2.1857,
      "step": 1085
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003999984901314102,
      "loss": 2.1972,
      "step": 1090
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003999969697814611,
      "loss": 2.1948,
      "step": 1095
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00039999492517898294,
      "loss": 2.2061,
      "step": 1100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00039999235632933523,
      "loss": 2.2225,
      "step": 1105
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003999892632392519,
      "loss": 2.2043,
      "step": 1110
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00039998564591684063,
      "loss": 2.1753,
      "step": 1115
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039998150437158366,
      "loss": 2.1665,
      "step": 1120
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039997683861433723,
      "loss": 2.2314,
      "step": 1125
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003999716486573316,
      "loss": 2.1757,
      "step": 1130
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0003999659345141714,
      "loss": 2.2133,
      "step": 1135
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003999596961998349,
      "loss": 2.204,
      "step": 1140
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003999529337306748,
      "loss": 2.1927,
      "step": 1145
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003999456471244174,
      "loss": 2.2271,
      "step": 1150
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039993783640016327,
      "loss": 2.1898,
      "step": 1155
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.171875,
      "learning_rate": 0.0003999295015783866,
      "loss": 2.1957,
      "step": 1160
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039992064268093544,
      "loss": 2.1883,
      "step": 1165
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039991125973103174,
      "loss": 2.2178,
      "step": 1170
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039990135275327096,
      "loss": 2.2018,
      "step": 1175
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003998909217736223,
      "loss": 2.2102,
      "step": 1180
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003998799668194285,
      "loss": 2.1726,
      "step": 1185
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003998684879194059,
      "loss": 2.2063,
      "step": 1190
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003998564851036441,
      "loss": 2.2208,
      "step": 1195
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039984395840360603,
      "loss": 2.1942,
      "step": 1200
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003998309078521281,
      "loss": 2.2295,
      "step": 1205
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.173828125,
      "learning_rate": 0.00039981733348341966,
      "loss": 2.1758,
      "step": 1210
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.17578125,
      "learning_rate": 0.00039980323533306327,
      "loss": 2.2272,
      "step": 1215
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039978861343801446,
      "loss": 2.1481,
      "step": 1220
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003997734678366016,
      "loss": 2.2496,
      "step": 1225
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00039975779856852596,
      "loss": 2.1578,
      "step": 1230
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003997416056748613,
      "loss": 2.1974,
      "step": 1235
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003997248891980542,
      "loss": 2.195,
      "step": 1240
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039970764918192356,
      "loss": 2.1602,
      "step": 1245
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003996898856716607,
      "loss": 2.1616,
      "step": 1250
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.17578125,
      "learning_rate": 0.00039967159871382915,
      "loss": 2.2104,
      "step": 1255
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003996527883563645,
      "loss": 2.2346,
      "step": 1260
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003996334546485744,
      "loss": 2.1678,
      "step": 1265
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039961359764113845,
      "loss": 2.2538,
      "step": 1270
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.171875,
      "learning_rate": 0.00039959321738610777,
      "loss": 2.1512,
      "step": 1275
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003995723139369052,
      "loss": 2.1887,
      "step": 1280
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039955088734832485,
      "loss": 2.2243,
      "step": 1285
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039952893767653257,
      "loss": 2.182,
      "step": 1290
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0003995064649790649,
      "loss": 2.2166,
      "step": 1295
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00039948346931482963,
      "loss": 2.2192,
      "step": 1300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003994599507441053,
      "loss": 2.2279,
      "step": 1305
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00039943590932854124,
      "loss": 2.1521,
      "step": 1310
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039941134513115734,
      "loss": 2.1969,
      "step": 1315
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039938625821634365,
      "loss": 2.1979,
      "step": 1320
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039936064864986063,
      "loss": 2.1844,
      "step": 1325
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039933451649883865,
      "loss": 2.2703,
      "step": 1330
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003993078618317781,
      "loss": 2.218,
      "step": 1335
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039928068471854875,
      "loss": 2.1867,
      "step": 1340
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039925298523039017,
      "loss": 2.1961,
      "step": 1345
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000399224763439911,
      "loss": 2.1803,
      "step": 1350
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003991960194210892,
      "loss": 2.2029,
      "step": 1355
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003991667532492714,
      "loss": 2.1686,
      "step": 1360
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003991369650011731,
      "loss": 2.18,
      "step": 1365
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003991066547548785,
      "loss": 2.1813,
      "step": 1370
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.17578125,
      "learning_rate": 0.00039907582258983965,
      "loss": 2.1894,
      "step": 1375
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039904446858687713,
      "loss": 2.2217,
      "step": 1380
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003990125928281793,
      "loss": 2.1899,
      "step": 1385
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039898019539730197,
      "loss": 2.202,
      "step": 1390
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003989472763791688,
      "loss": 2.1604,
      "step": 1395
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039891383586007043,
      "loss": 2.1963,
      "step": 1400
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00039887987392766454,
      "loss": 2.1794,
      "step": 1405
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003988453906709756,
      "loss": 2.1875,
      "step": 1410
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003988103861803948,
      "loss": 2.1734,
      "step": 1415
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0003987748605476793,
      "loss": 2.1798,
      "step": 1420
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0003987388138659526,
      "loss": 2.2137,
      "step": 1425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.000398702246229704,
      "loss": 2.1868,
      "step": 1430
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00039866515773478826,
      "loss": 2.1732,
      "step": 1435
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039862754847842563,
      "loss": 2.1935,
      "step": 1440
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003985894185592012,
      "loss": 2.1878,
      "step": 1445
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00039855076807706523,
      "loss": 2.1381,
      "step": 1450
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003985115971333321,
      "loss": 2.1796,
      "step": 1455
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003984719058306808,
      "loss": 2.2372,
      "step": 1460
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039843169427315425,
      "loss": 2.2127,
      "step": 1465
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003983909625661591,
      "loss": 2.1515,
      "step": 1470
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003983497108164654,
      "loss": 2.2013,
      "step": 1475
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0003983079391322065,
      "loss": 2.1768,
      "step": 1480
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003982656476228787,
      "loss": 2.2009,
      "step": 1485
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0003982228363993406,
      "loss": 2.24,
      "step": 1490
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003981795055738137,
      "loss": 2.2217,
      "step": 1495
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00039813565525988084,
      "loss": 2.1766,
      "step": 1500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1875,
      "learning_rate": 0.00039809128557248726,
      "loss": 2.1778,
      "step": 1505
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039804639662793914,
      "loss": 2.2053,
      "step": 1510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000398000988543904,
      "loss": 2.2288,
      "step": 1515
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039795506143941017,
      "loss": 2.1998,
      "step": 1520
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003979086154348465,
      "loss": 2.2061,
      "step": 1525
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039786165065196205,
      "loss": 2.1373,
      "step": 1530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039781416721386566,
      "loss": 2.2074,
      "step": 1535
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003977661652450257,
      "loss": 2.2157,
      "step": 1540
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00039771764487127,
      "loss": 2.2054,
      "step": 1545
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039766860621978504,
      "loss": 2.1908,
      "step": 1550
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039761904941911603,
      "loss": 2.1985,
      "step": 1555
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003975689745991662,
      "loss": 2.1675,
      "step": 1560
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003975183818911969,
      "loss": 2.178,
      "step": 1565
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039746727142782686,
      "loss": 2.1903,
      "step": 1570
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003974156433430321,
      "loss": 2.2173,
      "step": 1575
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0003973634977721454,
      "loss": 2.235,
      "step": 1580
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039731083485185605,
      "loss": 2.159,
      "step": 1585
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003972576547202096,
      "loss": 2.2094,
      "step": 1590
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003972039575166071,
      "loss": 2.2054,
      "step": 1595
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003971497433818053,
      "loss": 2.166,
      "step": 1600
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00039709501245791575,
      "loss": 2.216,
      "step": 1605
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003970397648884048,
      "loss": 2.1989,
      "step": 1610
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000396984000818093,
      "loss": 2.159,
      "step": 1615
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039692772039315484,
      "loss": 2.2085,
      "step": 1620
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003968709237611183,
      "loss": 2.1399,
      "step": 1625
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039681361107086463,
      "loss": 2.1798,
      "step": 1630
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003967557824726276,
      "loss": 2.1858,
      "step": 1635
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039669743811799354,
      "loss": 2.1729,
      "step": 1640
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003966385781599006,
      "loss": 2.2067,
      "step": 1645
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039657920275263856,
      "loss": 2.1663,
      "step": 1650
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00039651931205184824,
      "loss": 2.1732,
      "step": 1655
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1875,
      "learning_rate": 0.00039645890621452137,
      "loss": 2.1602,
      "step": 1660
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003963979853989999,
      "loss": 2.2557,
      "step": 1665
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00039633654976497563,
      "loss": 2.149,
      "step": 1670
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00039627459947349,
      "loss": 2.172,
      "step": 1675
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039621213468693343,
      "loss": 2.1722,
      "step": 1680
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1875,
      "learning_rate": 0.000396149155569045,
      "loss": 2.177,
      "step": 1685
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039608566228491204,
      "loss": 2.1349,
      "step": 1690
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00039602165500096973,
      "loss": 2.2292,
      "step": 1695
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039595713388500037,
      "loss": 2.1504,
      "step": 1700
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1875,
      "learning_rate": 0.00039589209910613336,
      "loss": 2.153,
      "step": 1705
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039582655083484454,
      "loss": 2.2195,
      "step": 1710
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00039576048924295576,
      "loss": 2.1511,
      "step": 1715
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003956939145036344,
      "loss": 2.1825,
      "step": 1720
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18359375,
      "learning_rate": 0.000395626826791393,
      "loss": 2.2075,
      "step": 1725
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039555922628208874,
      "loss": 2.2107,
      "step": 1730
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039549111315292294,
      "loss": 2.2175,
      "step": 1735
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039542248758244077,
      "loss": 2.1779,
      "step": 1740
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003953533497505306,
      "loss": 2.1802,
      "step": 1745
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.00039528369983842356,
      "loss": 2.1587,
      "step": 1750
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003952135380286931,
      "loss": 2.1503,
      "step": 1755
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039514286450525457,
      "loss": 2.2139,
      "step": 1760
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003950716794533647,
      "loss": 2.1928,
      "step": 1765
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000394999983059621,
      "loss": 2.1583,
      "step": 1770
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00039492777551196134,
      "loss": 2.2097,
      "step": 1775
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039485505699966356,
      "loss": 2.175,
      "step": 1780
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039478182771334494,
      "loss": 2.1728,
      "step": 1785
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003947080878449615,
      "loss": 2.1756,
      "step": 1790
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003946338375878078,
      "loss": 2.2138,
      "step": 1795
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00039455907713651614,
      "loss": 2.1444,
      "step": 1800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003944838066870563,
      "loss": 2.1863,
      "step": 1805
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039440802643673486,
      "loss": 2.1562,
      "step": 1810
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039433173658419483,
      "loss": 2.2186,
      "step": 1815
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003942549373294149,
      "loss": 2.1931,
      "step": 1820
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00039417762887370924,
      "loss": 2.1274,
      "step": 1825
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003940998114197266,
      "loss": 2.2013,
      "step": 1830
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0003940214851714501,
      "loss": 2.177,
      "step": 1835
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003939426503341965,
      "loss": 2.1978,
      "step": 1840
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003938633071146158,
      "loss": 2.2461,
      "step": 1845
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039378345572069044,
      "loss": 2.1986,
      "step": 1850
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00039370309636173513,
      "loss": 2.1906,
      "step": 1855
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039362222924839614,
      "loss": 2.1805,
      "step": 1860
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039354085459265055,
      "loss": 2.1797,
      "step": 1865
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003934589726078059,
      "loss": 2.1593,
      "step": 1870
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039337658350849973,
      "loss": 2.1646,
      "step": 1875
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003932936875106986,
      "loss": 2.215,
      "step": 1880
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00039321028483169817,
      "loss": 2.1932,
      "step": 1885
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00039312637569012207,
      "loss": 2.1654,
      "step": 1890
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003930419603059214,
      "loss": 2.1686,
      "step": 1895
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00039295703890037444,
      "loss": 2.1788,
      "step": 1900
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00039287161169608597,
      "loss": 2.2414,
      "step": 1905
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003927856789169865,
      "loss": 2.2244,
      "step": 1910
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003926992407883317,
      "loss": 2.1363,
      "step": 1915
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003926122975367022,
      "loss": 2.204,
      "step": 1920
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003925248493900024,
      "loss": 2.181,
      "step": 1925
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00039243689657746046,
      "loss": 2.1653,
      "step": 1930
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003923484393296273,
      "loss": 2.1717,
      "step": 1935
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000392259477878376,
      "loss": 2.1734,
      "step": 1940
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003921700124569015,
      "loss": 2.1973,
      "step": 1945
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003920800432997197,
      "loss": 2.189,
      "step": 1950
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000391989570642667,
      "loss": 2.2058,
      "step": 1955
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003918985947228995,
      "loss": 2.174,
      "step": 1960
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00039180711577889264,
      "loss": 2.2542,
      "step": 1965
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003917151340504405,
      "loss": 2.1754,
      "step": 1970
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003916226497786548,
      "loss": 2.2225,
      "step": 1975
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0003915296632059649,
      "loss": 2.1529,
      "step": 1980
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1875,
      "learning_rate": 0.00039143617457611674,
      "loss": 2.1837,
      "step": 1985
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003913421841341723,
      "loss": 2.1773,
      "step": 1990
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00039124769212650883,
      "loss": 2.1852,
      "step": 1995
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003911526988008185,
      "loss": 2.2199,
      "step": 2000
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00039105720440610765,
      "loss": 2.2312,
      "step": 2005
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00039096120919269577,
      "loss": 2.2002,
      "step": 2010
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003908647134122156,
      "loss": 2.1313,
      "step": 2015
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003907677173176115,
      "loss": 2.1836,
      "step": 2020
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00039067022116313964,
      "loss": 2.2171,
      "step": 2025
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000390572225204367,
      "loss": 2.1877,
      "step": 2030
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00039047372969817044,
      "loss": 2.1615,
      "step": 2035
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00039037473490273673,
      "loss": 2.1992,
      "step": 2040
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003902752410775609,
      "loss": 2.1588,
      "step": 2045
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00039017524848344653,
      "loss": 2.1624,
      "step": 2050
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003900747573825044,
      "loss": 2.1364,
      "step": 2055
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00038997376803815196,
      "loss": 2.1946,
      "step": 2060
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003898722807151129,
      "loss": 2.1941,
      "step": 2065
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003897702956794163,
      "loss": 2.217,
      "step": 2070
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003896678131983956,
      "loss": 2.1456,
      "step": 2075
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0003895648335406884,
      "loss": 2.1496,
      "step": 2080
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003894613569762356,
      "loss": 2.1576,
      "step": 2085
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00038935738377628045,
      "loss": 2.1978,
      "step": 2090
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00038925291421336824,
      "loss": 2.221,
      "step": 2095
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003891479485613452,
      "loss": 2.1936,
      "step": 2100
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00038904248709535817,
      "loss": 2.2115,
      "step": 2105
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003889365300918534,
      "loss": 2.1994,
      "step": 2110
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00038883007782857627,
      "loss": 2.2152,
      "step": 2115
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00038872313058457044,
      "loss": 2.1625,
      "step": 2120
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003886156886401768,
      "loss": 2.1987,
      "step": 2125
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003885077522770334,
      "loss": 2.1692,
      "step": 2130
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1875,
      "learning_rate": 0.00038839932177807385,
      "loss": 2.1872,
      "step": 2135
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003882903974275275,
      "loss": 2.2051,
      "step": 2140
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00038818097951091776,
      "loss": 2.2039,
      "step": 2145
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003880710683150622,
      "loss": 2.1854,
      "step": 2150
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0003879606641280714,
      "loss": 2.2216,
      "step": 2155
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00038784976723934796,
      "loss": 2.169,
      "step": 2160
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00038773837793958625,
      "loss": 2.202,
      "step": 2165
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003876264965207712,
      "loss": 2.1729,
      "step": 2170
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00038751412327617794,
      "loss": 2.1744,
      "step": 2175
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003874012585003707,
      "loss": 2.1703,
      "step": 2180
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003872879024892021,
      "loss": 2.2233,
      "step": 2185
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00038717405553981266,
      "loss": 2.2082,
      "step": 2190
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00038705971795062954,
      "loss": 2.1604,
      "step": 2195
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00038694489002136625,
      "loss": 2.1782,
      "step": 2200
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.00038682957205302137,
      "loss": 2.1704,
      "step": 2205
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00038671376434787824,
      "loss": 2.1866,
      "step": 2210
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003865974672095039,
      "loss": 2.1453,
      "step": 2215
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00038648068094274823,
      "loss": 2.183,
      "step": 2220
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003863634058537434,
      "loss": 2.1609,
      "step": 2225
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1875,
      "learning_rate": 0.00038624564224990285,
      "loss": 2.2011,
      "step": 2230
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003861273904399207,
      "loss": 2.1987,
      "step": 2235
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003860086507337705,
      "loss": 2.147,
      "step": 2240
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00038588942344270504,
      "loss": 2.2012,
      "step": 2245
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00038576970887925515,
      "loss": 2.1498,
      "step": 2250
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003856495073572289,
      "loss": 2.1827,
      "step": 2255
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003855288191917106,
      "loss": 2.2141,
      "step": 2260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00038540764469906073,
      "loss": 2.1551,
      "step": 2265
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00038528598419691404,
      "loss": 2.1988,
      "step": 2270
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003851638380041796,
      "loss": 2.2131,
      "step": 2275
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003850412064410396,
      "loss": 2.2368,
      "step": 2280
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003849180898289482,
      "loss": 2.2131,
      "step": 2285
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00038479448849063145,
      "loss": 2.1753,
      "step": 2290
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0003846704027500859,
      "loss": 2.125,
      "step": 2295
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00038454583293257754,
      "loss": 2.1785,
      "step": 2300
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003844207793646417,
      "loss": 2.1774,
      "step": 2305
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0003842952423740815,
      "loss": 2.201,
      "step": 2310
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003841692222899675,
      "loss": 2.1697,
      "step": 2315
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00038404271944263635,
      "loss": 2.2196,
      "step": 2320
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003839157341636903,
      "loss": 2.1834,
      "step": 2325
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003837882667859961,
      "loss": 2.168,
      "step": 2330
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003836603176436842,
      "loss": 2.2153,
      "step": 2335
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.00038353188707214826,
      "loss": 2.1761,
      "step": 2340
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003834029754080435,
      "loss": 2.1737,
      "step": 2345
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00038327358298928624,
      "loss": 2.152,
      "step": 2350
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00038314371015505327,
      "loss": 2.2155,
      "step": 2355
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00038301335724578057,
      "loss": 2.196,
      "step": 2360
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00038288252460316253,
      "loss": 2.1863,
      "step": 2365
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1875,
      "learning_rate": 0.000382751212570151,
      "loss": 2.1775,
      "step": 2370
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003826194214909545,
      "loss": 2.185,
      "step": 2375
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00038248715171103744,
      "loss": 2.1951,
      "step": 2380
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003823544035771187,
      "loss": 2.143,
      "step": 2385
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003822211774371715,
      "loss": 2.2101,
      "step": 2390
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00038208747364042167,
      "loss": 2.1411,
      "step": 2395
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00038195329253734735,
      "loss": 2.1732,
      "step": 2400
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003818186344796778,
      "loss": 2.1737,
      "step": 2405
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00038168349982039244,
      "loss": 2.185,
      "step": 2410
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003815478889137201,
      "loss": 2.1498,
      "step": 2415
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.000381411802115138,
      "loss": 2.1966,
      "step": 2420
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003812752397813708,
      "loss": 2.1869,
      "step": 2425
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00038113820227038967,
      "loss": 2.1542,
      "step": 2430
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003810006899414113,
      "loss": 2.1937,
      "step": 2435
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00038086270315489703,
      "loss": 2.178,
      "step": 2440
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003807242422725521,
      "loss": 2.199,
      "step": 2445
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003805853076573243,
      "loss": 2.1818,
      "step": 2450
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003804458996734032,
      "loss": 2.1872,
      "step": 2455
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003803060186862193,
      "loss": 2.1838,
      "step": 2460
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19140625,
      "learning_rate": 0.000380165665062443,
      "loss": 2.188,
      "step": 2465
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003800248391699836,
      "loss": 2.1769,
      "step": 2470
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003798835413779883,
      "loss": 2.1981,
      "step": 2475
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003797417720568413,
      "loss": 2.1709,
      "step": 2480
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003795995315781629,
      "loss": 2.1286,
      "step": 2485
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00037945682031480845,
      "loss": 2.174,
      "step": 2490
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003793136386408673,
      "loss": 2.2026,
      "step": 2495
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00037916998693166183,
      "loss": 2.1502,
      "step": 2500
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00037902586556374666,
      "loss": 2.2136,
      "step": 2505
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00037888127491490754,
      "loss": 2.1598,
      "step": 2510
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00037873621536416017,
      "loss": 2.1932,
      "step": 2515
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00037859068729174955,
      "loss": 2.1698,
      "step": 2520
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00037844469107914874,
      "loss": 2.1692,
      "step": 2525
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003782982271090579,
      "loss": 2.2203,
      "step": 2530
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00037815129576540356,
      "loss": 2.1922,
      "step": 2535
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.197265625,
      "learning_rate": 0.000378003897433337,
      "loss": 2.1406,
      "step": 2540
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1875,
      "learning_rate": 0.00037785603249923386,
      "loss": 2.1636,
      "step": 2545
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.185546875,
      "learning_rate": 0.00037770770135069293,
      "loss": 2.1878,
      "step": 2550
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000377558904376535,
      "loss": 2.1672,
      "step": 2555
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003774096419668018,
      "loss": 2.1447,
      "step": 2560
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003772599145127553,
      "loss": 2.1738,
      "step": 2565
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00037710972240687654,
      "loss": 2.127,
      "step": 2570
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00037695906604286427,
      "loss": 2.1835,
      "step": 2575
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003768079458156344,
      "loss": 2.2312,
      "step": 2580
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003766563621213189,
      "loss": 2.1855,
      "step": 2585
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003765043153572643,
      "loss": 2.1818,
      "step": 2590
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003763518059220311,
      "loss": 2.1717,
      "step": 2595
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003761988342153929,
      "loss": 2.1862,
      "step": 2600
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003760454006383345,
      "loss": 2.1539,
      "step": 2605
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003758915055930519,
      "loss": 2.1678,
      "step": 2610
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00037573714948295044,
      "loss": 2.1442,
      "step": 2615
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00037558233271264423,
      "loss": 2.2111,
      "step": 2620
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003754270556879547,
      "loss": 2.1631,
      "step": 2625
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003752713188159101,
      "loss": 2.1998,
      "step": 2630
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00037511512250474363,
      "loss": 2.1674,
      "step": 2635
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00037495846716389323,
      "loss": 2.1769,
      "step": 2640
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003748013532039998,
      "loss": 2.1797,
      "step": 2645
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00037464378103690656,
      "loss": 2.1771,
      "step": 2650
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00037448575107565786,
      "loss": 2.1965,
      "step": 2655
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.000374327263734498,
      "loss": 2.176,
      "step": 2660
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003741683194288701,
      "loss": 2.2076,
      "step": 2665
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003740089185754154,
      "loss": 2.1803,
      "step": 2670
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003738490615919716,
      "loss": 2.1576,
      "step": 2675
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003736887488975723,
      "loss": 2.2038,
      "step": 2680
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00037352798091244547,
      "loss": 2.1394,
      "step": 2685
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003733667580580127,
      "loss": 2.2167,
      "step": 2690
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00037320508075688776,
      "loss": 2.1758,
      "step": 2695
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003730429494328757,
      "loss": 2.1592,
      "step": 2700
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003728803645109719,
      "loss": 2.1819,
      "step": 2705
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00037271732641736043,
      "loss": 2.2038,
      "step": 2710
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003725538355794135,
      "loss": 2.191,
      "step": 2715
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00037238989242569003,
      "loss": 2.1797,
      "step": 2720
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003722254973859346,
      "loss": 2.1828,
      "step": 2725
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003720606508910763,
      "loss": 2.1772,
      "step": 2730
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00037189535337322767,
      "loss": 2.1698,
      "step": 2735
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0003717296052656835,
      "loss": 2.1274,
      "step": 2740
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003715634070029196,
      "loss": 2.1501,
      "step": 2745
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003713967590205919,
      "loss": 2.1519,
      "step": 2750
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00037122966175553524,
      "loss": 2.1676,
      "step": 2755
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.000371062115645762,
      "loss": 2.1744,
      "step": 2760
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00037089412113046116,
      "loss": 2.1807,
      "step": 2765
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00037072567864999723,
      "loss": 2.183,
      "step": 2770
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00037055678864590874,
      "loss": 2.1767,
      "step": 2775
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00037038745156090766,
      "loss": 2.1606,
      "step": 2780
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0003702176678388775,
      "loss": 2.149,
      "step": 2785
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0003700474379248728,
      "loss": 2.2374,
      "step": 2790
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003698767622651178,
      "loss": 2.1584,
      "step": 2795
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003697056413070047,
      "loss": 2.1679,
      "step": 2800
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003695340754990935,
      "loss": 2.1748,
      "step": 2805
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00036936206529110995,
      "loss": 2.2067,
      "step": 2810
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003691896111339449,
      "loss": 2.1876,
      "step": 2815
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00036901671347965275,
      "loss": 2.1719,
      "step": 2820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003688433727814506,
      "loss": 2.1687,
      "step": 2825
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00036866958949371677,
      "loss": 2.1702,
      "step": 2830
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003684953640719899,
      "loss": 2.1444,
      "step": 2835
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003683206969729673,
      "loss": 2.2099,
      "step": 2840
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003681455886545045,
      "loss": 2.1858,
      "step": 2845
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00036797003957561315,
      "loss": 2.1791,
      "step": 2850
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003677940501964606,
      "loss": 2.1658,
      "step": 2855
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003676176209783681,
      "loss": 2.1968,
      "step": 2860
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00036744075238381017,
      "loss": 2.1729,
      "step": 2865
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00036726344487641267,
      "loss": 2.1772,
      "step": 2870
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00036708569892095227,
      "loss": 2.1572,
      "step": 2875
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00036690751498335487,
      "loss": 2.1488,
      "step": 2880
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003667288935306944,
      "loss": 2.1995,
      "step": 2885
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003665498350311918,
      "loss": 2.1854,
      "step": 2890
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00036637033995421347,
      "loss": 2.1867,
      "step": 2895
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003661904087702702,
      "loss": 2.1902,
      "step": 2900
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003660100419510161,
      "loss": 2.1654,
      "step": 2905
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00036582923996924724,
      "loss": 2.1987,
      "step": 2910
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003656480032989001,
      "loss": 2.1442,
      "step": 2915
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00036546633241505094,
      "loss": 2.1855,
      "step": 2920
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.000365284227793914,
      "loss": 2.2243,
      "step": 2925
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003651016899128406,
      "loss": 2.1589,
      "step": 2930
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00036491871925031755,
      "loss": 2.1643,
      "step": 2935
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003647353162859666,
      "loss": 2.1953,
      "step": 2940
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003645514815005421,
      "loss": 2.1838,
      "step": 2945
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0003643672153759307,
      "loss": 2.2055,
      "step": 2950
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00036418251839514956,
      "loss": 2.1623,
      "step": 2955
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18359375,
      "learning_rate": 0.00036399739104234544,
      "loss": 2.1533,
      "step": 2960
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00036381183380279305,
      "loss": 2.1831,
      "step": 2965
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00036362584716289405,
      "loss": 2.1655,
      "step": 2970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003634394316101756,
      "loss": 2.1324,
      "step": 2975
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003632525876332892,
      "loss": 2.2055,
      "step": 2980
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00036306531572200944,
      "loss": 2.1625,
      "step": 2985
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00036287761636723275,
      "loss": 2.1718,
      "step": 2990
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00036268949006097566,
      "loss": 2.192,
      "step": 2995
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00036250093729637433,
      "loss": 2.1969,
      "step": 3000
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00036231195856768235,
      "loss": 2.1664,
      "step": 3005
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003621225543702703,
      "loss": 2.1199,
      "step": 3010
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00036193272520062376,
      "loss": 2.211,
      "step": 3015
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00036174247155634233,
      "loss": 2.1411,
      "step": 3020
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003615517939361385,
      "loss": 2.1303,
      "step": 3025
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00036136069283983577,
      "loss": 2.1731,
      "step": 3030
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00036116916876836804,
      "loss": 2.1537,
      "step": 3035
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00036097722222377775,
      "loss": 2.2165,
      "step": 3040
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00036078485370921476,
      "loss": 2.1514,
      "step": 3045
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00036059206372893523,
      "loss": 2.1642,
      "step": 3050
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003603988527883,
      "loss": 2.1888,
      "step": 3055
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00036020522139377327,
      "loss": 2.1869,
      "step": 3060
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00036001117005292154,
      "loss": 2.1677,
      "step": 3065
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.189453125,
      "learning_rate": 0.000359816699274412,
      "loss": 2.167,
      "step": 3070
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00035962180956801133,
      "loss": 2.1305,
      "step": 3075
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003594265014445845,
      "loss": 2.1672,
      "step": 3080
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00035923077541609314,
      "loss": 2.1662,
      "step": 3085
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003590346319955942,
      "loss": 2.2116,
      "step": 3090
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1875,
      "learning_rate": 0.000358838071697239,
      "loss": 2.1816,
      "step": 3095
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003586410950362715,
      "loss": 2.182,
      "step": 3100
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003584437025290271,
      "loss": 2.1423,
      "step": 3105
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00035824589469293127,
      "loss": 2.1577,
      "step": 3110
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00035804767204649805,
      "loss": 2.2046,
      "step": 3115
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00035784903510932905,
      "loss": 2.2101,
      "step": 3120
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00035764998440211167,
      "loss": 2.1624,
      "step": 3125
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00035745052044661803,
      "loss": 2.213,
      "step": 3130
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003572506437657035,
      "loss": 2.197,
      "step": 3135
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1875,
      "learning_rate": 0.00035705035488330523,
      "loss": 2.1877,
      "step": 3140
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00035684965432444094,
      "loss": 2.1717,
      "step": 3145
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00035664854261520753,
      "loss": 2.1887,
      "step": 3150
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00035644702028277955,
      "loss": 2.1668,
      "step": 3155
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00035624508785540805,
      "loss": 2.1618,
      "step": 3160
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00035604274586241886,
      "loss": 2.2285,
      "step": 3165
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003558399948342116,
      "loss": 2.2208,
      "step": 3170
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00035563683530225797,
      "loss": 2.197,
      "step": 3175
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003554332677991006,
      "loss": 2.2029,
      "step": 3180
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003552292928583514,
      "loss": 2.1402,
      "step": 3185
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003550249110146904,
      "loss": 2.2078,
      "step": 3190
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003548201228038642,
      "loss": 2.1611,
      "step": 3195
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003546149287626846,
      "loss": 2.1795,
      "step": 3200
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00035440932942902727,
      "loss": 2.1683,
      "step": 3205
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00035420332534183023,
      "loss": 2.1783,
      "step": 3210
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003539969170410924,
      "loss": 2.1756,
      "step": 3215
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003537901050678724,
      "loss": 2.1829,
      "step": 3220
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000353582889964287,
      "loss": 2.1947,
      "step": 3225
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003533752722735096,
      "loss": 2.1792,
      "step": 3230
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00035316725253976887,
      "loss": 2.1858,
      "step": 3235
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003529588313083474,
      "loss": 2.1128,
      "step": 3240
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003527500091255805,
      "loss": 2.2005,
      "step": 3245
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.185546875,
      "learning_rate": 0.000352540786538854,
      "loss": 2.1451,
      "step": 3250
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003523311640966037,
      "loss": 2.1474,
      "step": 3255
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003521211423483133,
      "loss": 2.1799,
      "step": 3260
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003519107218445134,
      "loss": 2.1957,
      "step": 3265
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00035169990313677974,
      "loss": 2.178,
      "step": 3270
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00035148868677773186,
      "loss": 2.2025,
      "step": 3275
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00035127707332103175,
      "loss": 2.1991,
      "step": 3280
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00035106506332138217,
      "loss": 2.1969,
      "step": 3285
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00035085265733452554,
      "loss": 2.1569,
      "step": 3290
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003506398559172421,
      "loss": 2.2137,
      "step": 3295
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003504266596273488,
      "loss": 2.1768,
      "step": 3300
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00035021306902369745,
      "loss": 2.2186,
      "step": 3305
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003499990846661737,
      "loss": 2.1161,
      "step": 3310
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003497847071156952,
      "loss": 2.1391,
      "step": 3315
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003495699369342104,
      "loss": 2.1398,
      "step": 3320
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003493547746846968,
      "loss": 2.1858,
      "step": 3325
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003491392209311597,
      "loss": 2.159,
      "step": 3330
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00034892327623863077,
      "loss": 2.1631,
      "step": 3335
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003487069411731663,
      "loss": 2.1614,
      "step": 3340
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00034849021630184587,
      "loss": 2.1665,
      "step": 3345
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003482731021927709,
      "loss": 2.1872,
      "step": 3350
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003480555994150631,
      "loss": 2.1781,
      "step": 3355
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003478377085388631,
      "loss": 2.1907,
      "step": 3360
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.00034761943013532874,
      "loss": 2.1636,
      "step": 3365
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003474007647766336,
      "loss": 2.1246,
      "step": 3370
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003471817130359659,
      "loss": 2.1438,
      "step": 3375
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003469622754875263,
      "loss": 2.1563,
      "step": 3380
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003467424527065271,
      "loss": 2.1809,
      "step": 3385
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00034652224526919014,
      "loss": 2.1731,
      "step": 3390
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003463016537527458,
      "loss": 2.1673,
      "step": 3395
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00034608067873543116,
      "loss": 2.1539,
      "step": 3400
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003458593207964885,
      "loss": 2.1642,
      "step": 3405
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003456375805161638,
      "loss": 2.1938,
      "step": 3410
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003454154584757056,
      "loss": 2.1743,
      "step": 3415
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.1875,
      "learning_rate": 0.0003451929552573629,
      "loss": 2.1621,
      "step": 3420
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00034497007144438367,
      "loss": 2.1694,
      "step": 3425
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00034474680762101406,
      "loss": 2.1555,
      "step": 3430
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003445231643724959,
      "loss": 2.1856,
      "step": 3435
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003442991422850658,
      "loss": 2.1824,
      "step": 3440
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003440747419459534,
      "loss": 2.1838,
      "step": 3445
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003438499639433798,
      "loss": 2.1467,
      "step": 3450
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00034362480886655615,
      "loss": 2.1592,
      "step": 3455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.000343399277305682,
      "loss": 2.1921,
      "step": 3460
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003431733698519437,
      "loss": 2.1951,
      "step": 3465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003429470870975131,
      "loss": 2.225,
      "step": 3470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00034272042963554554,
      "loss": 2.185,
      "step": 3475
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003424933980601789,
      "loss": 2.1617,
      "step": 3480
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003422659929665316,
      "loss": 2.1954,
      "step": 3485
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00034203821495070103,
      "loss": 2.1607,
      "step": 3490
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003418100646097624,
      "loss": 2.1595,
      "step": 3495
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00034158154254176654,
      "loss": 2.1473,
      "step": 3500
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003413526493457391,
      "loss": 2.2123,
      "step": 3505
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003411233856216781,
      "loss": 2.1788,
      "step": 3510
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00034089375197055336,
      "loss": 2.1447,
      "step": 3515
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003406637489943039,
      "loss": 2.1855,
      "step": 3520
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003404333772958372,
      "loss": 2.1916,
      "step": 3525
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00034020263747902715,
      "loss": 2.1985,
      "step": 3530
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00033997153014871237,
      "loss": 2.2015,
      "step": 3535
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003397400559106953,
      "loss": 2.1476,
      "step": 3540
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003395082153717397,
      "loss": 2.159,
      "step": 3545
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00033927600913956986,
      "loss": 2.19,
      "step": 3550
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003390434378228685,
      "loss": 2.1506,
      "step": 3555
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00033881050203127527,
      "loss": 2.1569,
      "step": 3560
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003385772023753855,
      "loss": 2.1574,
      "step": 3565
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000338343539466748,
      "loss": 2.1559,
      "step": 3570
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00033810951391786384,
      "loss": 2.2004,
      "step": 3575
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00033787512634218483,
      "loss": 2.1497,
      "step": 3580
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003376403773541117,
      "loss": 2.1442,
      "step": 3585
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003374052675689925,
      "loss": 2.1432,
      "step": 3590
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000337169797603121,
      "loss": 2.1708,
      "step": 3595
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00033693396807373536,
      "loss": 2.1471,
      "step": 3600
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00033669777959901583,
      "loss": 2.1216,
      "step": 3605
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.000336461232798084,
      "loss": 2.1942,
      "step": 3610
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003362243282910005,
      "loss": 2.198,
      "step": 3615
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0003359870666987637,
      "loss": 2.1837,
      "step": 3620
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003357494486433078,
      "loss": 2.1701,
      "step": 3625
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003355114747475019,
      "loss": 2.1814,
      "step": 3630
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00033527314563514725,
      "loss": 2.1656,
      "step": 3635
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003350344619309767,
      "loss": 2.196,
      "step": 3640
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003347954242606522,
      "loss": 2.1617,
      "step": 3645
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.000334556033250764,
      "loss": 2.1825,
      "step": 3650
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00033431628952882813,
      "loss": 2.132,
      "step": 3655
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00033407619372328545,
      "loss": 2.1444,
      "step": 3660
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00033383574646349973,
      "loss": 2.1785,
      "step": 3665
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003335949483797558,
      "loss": 2.117,
      "step": 3670
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003333538001032583,
      "loss": 2.1793,
      "step": 3675
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00033311230226612987,
      "loss": 2.1872,
      "step": 3680
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00033287045550140924,
      "loss": 2.1664,
      "step": 3685
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00033262826044304995,
      "loss": 2.1627,
      "step": 3690
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00033238571772591845,
      "loss": 2.1571,
      "step": 3695
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00033214282798579256,
      "loss": 2.1433,
      "step": 3700
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.189453125,
      "learning_rate": 0.00033189959185935973,
      "loss": 2.127,
      "step": 3705
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00033165600998421544,
      "loss": 2.1733,
      "step": 3710
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003314120829988613,
      "loss": 2.1758,
      "step": 3715
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003311678115427039,
      "loss": 2.17,
      "step": 3720
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0003309231962560524,
      "loss": 2.167,
      "step": 3725
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003306782377801175,
      "loss": 2.172,
      "step": 3730
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003304329367570094,
      "loss": 2.1651,
      "step": 3735
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003301872938297362,
      "loss": 2.153,
      "step": 3740
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003299413096422025,
      "loss": 2.1491,
      "step": 3745
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00032969498483920704,
      "loss": 2.1947,
      "step": 3750
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00032944832006644165,
      "loss": 2.1276,
      "step": 3755
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003292013159704893,
      "loss": 2.1548,
      "step": 3760
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00032895397319882236,
      "loss": 2.1466,
      "step": 3765
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003287062923998011,
      "loss": 2.1091,
      "step": 3770
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003284582742226717,
      "loss": 2.1843,
      "step": 3775
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00032820991931756493,
      "loss": 2.2426,
      "step": 3780
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00032796122833549394,
      "loss": 2.1489,
      "step": 3785
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.189453125,
      "learning_rate": 0.000327712201928353,
      "loss": 2.1309,
      "step": 3790
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.00032746284074891573,
      "loss": 2.1768,
      "step": 3795
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0003272131454508331,
      "loss": 2.1846,
      "step": 3800
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.000326963116688632,
      "loss": 2.2,
      "step": 3805
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00032671275511771337,
      "loss": 2.1608,
      "step": 3810
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00032646206139435067,
      "loss": 2.1848,
      "step": 3815
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00032621103617568785,
      "loss": 2.1613,
      "step": 3820
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00032595968011973805,
      "loss": 2.1394,
      "step": 3825
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00032570799388538133,
      "loss": 2.1516,
      "step": 3830
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00032545597813236356,
      "loss": 2.214,
      "step": 3835
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0003252036335212941,
      "loss": 2.1682,
      "step": 3840
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003249509607136446,
      "loss": 2.1702,
      "step": 3845
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00032469796037174674,
      "loss": 2.2195,
      "step": 3850
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00032444463315879103,
      "loss": 2.0988,
      "step": 3855
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003241909797388246,
      "loss": 2.1504,
      "step": 3860
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00032393700077674987,
      "loss": 2.1767,
      "step": 3865
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003236826969383224,
      "loss": 2.1628,
      "step": 3870
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003234280688901495,
      "loss": 2.2204,
      "step": 3875
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00032317311729968825,
      "loss": 2.1606,
      "step": 3880
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003229178428352438,
      "loss": 2.1721,
      "step": 3885
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00032266224616596785,
      "loss": 2.1741,
      "step": 3890
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0003224063279618564,
      "loss": 2.1784,
      "step": 3895
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00032215008889374853,
      "loss": 2.1944,
      "step": 3900
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00032189352963332425,
      "loss": 2.1608,
      "step": 3905
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003216366508531031,
      "loss": 2.1662,
      "step": 3910
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00032137945322644184,
      "loss": 2.1546,
      "step": 3915
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00032112193742753333,
      "loss": 2.1698,
      "step": 3920
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.203125,
      "learning_rate": 0.0003208641041314043,
      "loss": 2.2374,
      "step": 3925
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003206059540139139,
      "loss": 2.1985,
      "step": 3930
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0003203474877517514,
      "loss": 2.1759,
      "step": 3935
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00032008870602243523,
      "loss": 2.1489,
      "step": 3940
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003198296095043104,
      "loss": 2.1875,
      "step": 3945
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00031957019887654737,
      "loss": 2.1403,
      "step": 3950
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003193104748191398,
      "loss": 2.1555,
      "step": 3955
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003190504380129029,
      "loss": 2.178,
      "step": 3960
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003187900891394718,
      "loss": 2.1751,
      "step": 3965
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003185294288812996,
      "loss": 2.1864,
      "step": 3970
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003182684579216557,
      "loss": 2.1756,
      "step": 3975
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0003180071769446238,
      "loss": 2.1778,
      "step": 3980
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00031774558663510044,
      "loss": 2.1806,
      "step": 3985
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00031748368767879284,
      "loss": 2.1518,
      "step": 3990
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003172214807622175,
      "loss": 2.1643,
      "step": 3995
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00031695896657269785,
      "loss": 2.1502,
      "step": 4000
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00031669614579836307,
      "loss": 2.1605,
      "step": 4005
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00031643301912814576,
      "loss": 2.1922,
      "step": 4010
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00031616958725178056,
      "loss": 2.146,
      "step": 4015
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003159058508598021,
      "loss": 2.1565,
      "step": 4020
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00031564181064354315,
      "loss": 2.141,
      "step": 4025
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.000315377467295133,
      "loss": 2.1504,
      "step": 4030
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003151128215074954,
      "loss": 2.1534,
      "step": 4035
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003148478739743472,
      "loss": 2.1279,
      "step": 4040
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0003145826253901957,
      "loss": 2.1695,
      "step": 4045
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00031431707645033775,
      "loss": 2.2067,
      "step": 4050
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00031405122785085757,
      "loss": 2.1689,
      "step": 4055
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003137850802886245,
      "loss": 2.1577,
      "step": 4060
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00031351863446129187,
      "loss": 2.1237,
      "step": 4065
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003132518910672948,
      "loss": 2.247,
      "step": 4070
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00031298485080584823,
      "loss": 2.1987,
      "step": 4075
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003127175143769456,
      "loss": 2.1932,
      "step": 4080
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00031244988248135645,
      "loss": 2.1826,
      "step": 4085
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003121819558206249,
      "loss": 2.1388,
      "step": 4090
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00031191373509706767,
      "loss": 2.1869,
      "step": 4095
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00031164522101377254,
      "loss": 2.1403,
      "step": 4100
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.203125,
      "learning_rate": 0.000311376414274596,
      "loss": 2.176,
      "step": 4105
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00031110731558416175,
      "loss": 2.184,
      "step": 4110
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0003108379256478589,
      "loss": 2.1698,
      "step": 4115
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00031056824517183986,
      "loss": 2.1796,
      "step": 4120
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0003102982748630188,
      "loss": 2.196,
      "step": 4125
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00031002801542906943,
      "loss": 2.165,
      "step": 4130
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00030975746757842354,
      "loss": 2.1812,
      "step": 4135
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00030948663202026873,
      "loss": 2.2209,
      "step": 4140
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00030921550946454694,
      "loss": 2.1573,
      "step": 4145
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003089441006219524,
      "loss": 2.1966,
      "step": 4150
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003086724062039297,
      "loss": 2.1905,
      "step": 4155
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00030840042692267214,
      "loss": 2.1838,
      "step": 4160
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00030812816349111953,
      "loss": 2.1487,
      "step": 4165
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00030785561662295674,
      "loss": 2.1751,
      "step": 4170
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0003075827870326115,
      "loss": 2.1752,
      "step": 4175
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0003073096754352526,
      "loss": 2.1844,
      "step": 4180
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003070362825467882,
      "loss": 2.1858,
      "step": 4185
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00030676260908386367,
      "loss": 2.1701,
      "step": 4190
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00030648865576385993,
      "loss": 2.1694,
      "step": 4195
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00030621442330489143,
      "loss": 2.1362,
      "step": 4200
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003059399124258044,
      "loss": 2.1407,
      "step": 4205
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00030566512384617485,
      "loss": 2.1593,
      "step": 4210
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003053900582863068,
      "loss": 2.1579,
      "step": 4215
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003051147164672301,
      "loss": 2.2069,
      "step": 4220
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19921875,
      "learning_rate": 0.000304839099110699,
      "loss": 2.1785,
      "step": 4225
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003045632069391899,
      "loss": 2.1446,
      "step": 4230
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003042870406758996,
      "loss": 2.1558,
      "step": 4235
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0003040106010447435,
      "loss": 2.1839,
      "step": 4240
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003037338887703532,
      "loss": 2.1597,
      "step": 4245
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003034569045780755,
      "loss": 2.146,
      "step": 4250
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0003031796491939695,
      "loss": 2.1618,
      "step": 4255
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0003029021233448056,
      "loss": 2.1484,
      "step": 4260
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003026243277580628,
      "loss": 2.1625,
      "step": 4265
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0003023462631619274,
      "loss": 2.1474,
      "step": 4270
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0003020679302852908,
      "loss": 2.1733,
      "step": 4275
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.203125,
      "learning_rate": 0.00030178932985774757,
      "loss": 2.1369,
      "step": 4280
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00030151046260959376,
      "loss": 2.1609,
      "step": 4285
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003012313292718246,
      "loss": 2.1482,
      "step": 4290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00030095193057613306,
      "loss": 2.1889,
      "step": 4295
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0003006722672549076,
      "loss": 2.1691,
      "step": 4300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0003003923400412304,
      "loss": 2.1902,
      "step": 4305
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0003001121496688752,
      "loss": 2.1412,
      "step": 4310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00029983169687230576,
      "loss": 2.2059,
      "step": 4315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002995509823866736,
      "loss": 2.1333,
      "step": 4320
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00029927000694781637,
      "loss": 2.159,
      "step": 4325
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002989887712922555,
      "loss": 2.1684,
      "step": 4330
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002987072761571948,
      "loss": 2.1985,
      "step": 4335
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002984255222805182,
      "loss": 2.1664,
      "step": 4340
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002981435104007876,
      "loss": 2.1755,
      "step": 4345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002978612412572416,
      "loss": 2.1208,
      "step": 4350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002975787155897927,
      "loss": 2.1604,
      "step": 4355
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00029729593413902643,
      "loss": 2.1972,
      "step": 4360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00029701289764619824,
      "loss": 2.179,
      "step": 4365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002967296068532325,
      "loss": 2.1741,
      "step": 4370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002964460625027199,
      "loss": 2.1915,
      "step": 4375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00029616226533791607,
      "loss": 2.1776,
      "step": 4380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19921875,
      "learning_rate": 0.000295878216102739,
      "loss": 2.1422,
      "step": 4385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002955939155417678,
      "loss": 2.1374,
      "step": 4390
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002953093644002402,
      "loss": 2.1876,
      "step": 4395
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002950245634240506,
      "loss": 2.1888,
      "step": 4400
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00029473951335974856,
      "loss": 2.1611,
      "step": 4405
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002944542149545366,
      "loss": 2.1848,
      "step": 4410
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002941686689562679,
      "loss": 2.1544,
      "step": 4415
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00029388287611344506,
      "loss": 2.1715,
      "step": 4420
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002935968371752174,
      "loss": 2.1448,
      "step": 4425
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002933105528913795,
      "loss": 2.1661,
      "step": 4430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00029302402401236904,
      "loss": 2.1847,
      "step": 4435
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00029273725128926484,
      "loss": 2.1442,
      "step": 4440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00029245023547378493,
      "loss": 2.1806,
      "step": 4445
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002921629773182845,
      "loss": 2.1519,
      "step": 4450
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002918754775757541,
      "loss": 2.1223,
      "step": 4455
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0002915877369998174,
      "loss": 2.1435,
      "step": 4460
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0002912997563447296,
      "loss": 2.1613,
      "step": 4465
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002910115363653749,
      "loss": 2.1497,
      "step": 4470
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000290723077817265,
      "loss": 2.1734,
      "step": 4475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00029043438145653715,
      "loss": 2.1742,
      "step": 4480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002901454480399517,
      "loss": 2.1759,
      "step": 4485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00028985627832489044,
      "loss": 2.1722,
      "step": 4490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002895668730693548,
      "loss": 2.1574,
      "step": 4495
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002892772330319633,
      "loss": 2.1523,
      "step": 4500
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002889873589719501,
      "loss": 2.1746,
      "step": 4505
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002886972516491627,
      "loss": 2.1642,
      "step": 4510
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002884069118240602,
      "loss": 2.1204,
      "step": 4515
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002881163402577111,
      "loss": 2.1641,
      "step": 4520
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00028782553771179123,
      "loss": 2.1502,
      "step": 4525
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000287534504948582,
      "loss": 2.1846,
      "step": 4530
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00028724324273096837,
      "loss": 2.2048,
      "step": 4535
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002869517518224366,
      "loss": 2.1565,
      "step": 4540
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002866600329870725,
      "loss": 2.1622,
      "step": 4545
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00028636808698955933,
      "loss": 2.1589,
      "step": 4550
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00028607591459517596,
      "loss": 2.1666,
      "step": 4555
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002857835165697944,
      "loss": 2.1468,
      "step": 4560
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002854908936798783,
      "loss": 2.1915,
      "step": 4565
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00028519804669248084,
      "loss": 2.2123,
      "step": 4570
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002849049763752424,
      "loss": 2.1614,
      "step": 4575
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00028461168349638903,
      "loss": 2.1514,
      "step": 4580
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00028431816882473,
      "loss": 2.1957,
      "step": 4585
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00028402443312965596,
      "loss": 2.172,
      "step": 4590
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.19921875,
      "learning_rate": 0.000283730477181137,
      "loss": 2.1331,
      "step": 4595
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002834363017497205,
      "loss": 2.1357,
      "step": 4600
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002831419076065293,
      "loss": 2.1649,
      "step": 4605
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002828472955232595,
      "loss": 2.1565,
      "step": 4610
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00028255246627217824,
      "loss": 2.1964,
      "step": 4615
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00028225742062612237,
      "loss": 2.1533,
      "step": 4620
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00028196215935849555,
      "loss": 2.1384,
      "step": 4625
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00028166668324326695,
      "loss": 2.1632,
      "step": 4630
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0002813709930549688,
      "loss": 2.1184,
      "step": 4635
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002810750895686944,
      "loss": 2.1813,
      "step": 4640
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002807789735600964,
      "loss": 2.1603,
      "step": 4645
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00028048264580538435,
      "loss": 2.1534,
      "step": 4650
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00028018610708132274,
      "loss": 2.191,
      "step": 4655
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002798893581652295,
      "loss": 2.1827,
      "step": 4660
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002795923998349729,
      "loss": 2.1759,
      "step": 4665
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002792952328689709,
      "loss": 2.15,
      "step": 4670
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002789978580461877,
      "loss": 2.1464,
      "step": 4675
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002787002761461328,
      "loss": 2.1575,
      "step": 4680
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00027840248794885826,
      "loss": 2.1241,
      "step": 4685
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0002781044942349569,
      "loss": 2.1715,
      "step": 4690
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00027780629578556045,
      "loss": 2.1551,
      "step": 4695
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002775078933823372,
      "loss": 2.1945,
      "step": 4700
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00027720928780749,
      "loss": 2.1742,
      "step": 4705
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002769104798437546,
      "loss": 2.1535,
      "step": 4710
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00027661147027439664,
      "loss": 2.1172,
      "step": 4715
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00027631225988321084,
      "loss": 2.1983,
      "step": 4720
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002760128494545181,
      "loss": 2.1574,
      "step": 4725
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002757132397731636,
      "loss": 2.1594,
      "step": 4730
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00027541343162451495,
      "loss": 2.1266,
      "step": 4735
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00027511342579446,
      "loss": 2.1707,
      "step": 4740
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002748132230694047,
      "loss": 2.1271,
      "step": 4745
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002745128242362711,
      "loss": 2.1612,
      "step": 4750
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00027421223008249545,
      "loss": 2.1779,
      "step": 4755
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00027391144139602596,
      "loss": 2.1914,
      "step": 4760
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00027361045896532053,
      "loss": 2.1352,
      "step": 4765
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002733092835793454,
      "loss": 2.1487,
      "step": 4770
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002730079160275721,
      "loss": 2.1512,
      "step": 4775
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00027270635709997616,
      "loss": 2.1593,
      "step": 4780
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002724046075870348,
      "loss": 2.1361,
      "step": 4785
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002721026682797245,
      "loss": 2.1535,
      "step": 4790
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002718005399695197,
      "loss": 2.1602,
      "step": 4795
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00027149822344839006,
      "loss": 2.1593,
      "step": 4800
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.193359375,
      "learning_rate": 0.00027119571950879847,
      "loss": 2.1754,
      "step": 4805
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00027089302894369924,
      "loss": 2.1668,
      "step": 4810
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00027059015254653586,
      "loss": 2.1684,
      "step": 4815
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000270287091111239,
      "loss": 2.1496,
      "step": 4820
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00026998384543222434,
      "loss": 2.1385,
      "step": 4825
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002696804163043904,
      "loss": 2.1392,
      "step": 4830
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00026937680452311674,
      "loss": 2.2032,
      "step": 4835
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002690730108842615,
      "loss": 2.1713,
      "step": 4840
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002687690361841599,
      "loss": 2.1981,
      "step": 4845
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00026846488121962136,
      "loss": 2.1998,
      "step": 4850
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00026816054678792807,
      "loss": 2.1832,
      "step": 4855
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00026785603368683253,
      "loss": 2.137,
      "step": 4860
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002675513427145558,
      "loss": 2.1652,
      "step": 4865
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00026724647466978493,
      "loss": 2.164,
      "step": 4870
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002669414303516712,
      "loss": 2.1208,
      "step": 4875
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002666362105598281,
      "loss": 2.1669,
      "step": 4880
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00026633081609432895,
      "loss": 2.1483,
      "step": 4885
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000266025247755705,
      "loss": 2.1685,
      "step": 4890
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002657195063449432,
      "loss": 2.1387,
      "step": 4895
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00026541359266348436,
      "loss": 2.1927,
      "step": 4900
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002651075075132206,
      "loss": 2.1684,
      "step": 4905
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002648012516964937,
      "loss": 2.1658,
      "step": 4910
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002644948260160928,
      "loss": 2.1713,
      "step": 4915
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00026418823127525215,
      "loss": 2.1661,
      "step": 4920
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0002638814682776494,
      "loss": 2.2177,
      "step": 4925
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000263574537827403,
      "loss": 2.145,
      "step": 4930
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00026326744072907056,
      "loss": 2.1322,
      "step": 4935
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00026296017778764633,
      "loss": 2.1891,
      "step": 4940
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00026265274980855947,
      "loss": 2.2206,
      "step": 4945
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00026234515759767166,
      "loss": 2.1822,
      "step": 4950
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00026203740196127504,
      "loss": 2.1469,
      "step": 4955
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0002617294837060902,
      "loss": 2.1701,
      "step": 4960
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.201171875,
      "learning_rate": 0.000261421403639264,
      "loss": 2.1456,
      "step": 4965
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00026111316256836745,
      "loss": 2.1366,
      "step": 4970
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002608047613013936,
      "loss": 2.1981,
      "step": 4975
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002604962006467555,
      "loss": 2.1793,
      "step": 4980
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002601874814132837,
      "loss": 2.1473,
      "step": 4985
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0002598786044102249,
      "loss": 2.1915,
      "step": 4990
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002595695704472389,
      "loss": 2.1568,
      "step": 4995
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002592603803343973,
      "loss": 2.157,
      "step": 5000
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00025895103488218085,
      "loss": 2.1811,
      "step": 5005
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002586415349014775,
      "loss": 2.1387,
      "step": 5010
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0002583318812035803,
      "loss": 2.1382,
      "step": 5015
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002580220746001852,
      "loss": 2.1861,
      "step": 5020
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.000257712115903389,
      "loss": 2.1506,
      "step": 5025
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00025740200592568713,
      "loss": 2.1404,
      "step": 5030
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00025709174547997157,
      "loss": 2.1181,
      "step": 5035
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002567813353795288,
      "loss": 2.1433,
      "step": 5040
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00025647077643803763,
      "loss": 2.1527,
      "step": 5045
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00025616006946956683,
      "loss": 2.1853,
      "step": 5050
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002558492152885735,
      "loss": 2.1775,
      "step": 5055
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00025553821470990007,
      "loss": 2.1644,
      "step": 5060
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00025522706854877344,
      "loss": 2.1959,
      "step": 5065
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002549157776208016,
      "loss": 2.2071,
      "step": 5070
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002546043427419723,
      "loss": 2.1704,
      "step": 5075
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002542927647286505,
      "loss": 2.1489,
      "step": 5080
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00025398104439757624,
      "loss": 2.1703,
      "step": 5085
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002536691825658629,
      "loss": 2.1398,
      "step": 5090
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.00025335718005099474,
      "loss": 2.1478,
      "step": 5095
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0002530450376708244,
      "loss": 2.137,
      "step": 5100
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.19921875,
      "learning_rate": 0.00025273275624357165,
      "loss": 2.1979,
      "step": 5105
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00025242033658782044,
      "loss": 2.1232,
      "step": 5110
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00025210777952251723,
      "loss": 2.1871,
      "step": 5115
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00025179508586696856,
      "loss": 2.1422,
      "step": 5120
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.205078125,
      "learning_rate": 0.000251482256440839,
      "loss": 2.1773,
      "step": 5125
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002511692920641491,
      "loss": 2.167,
      "step": 5130
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002508561935572731,
      "loss": 2.1768,
      "step": 5135
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002505429617409369,
      "loss": 2.1799,
      "step": 5140
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0002502295974362158,
      "loss": 2.153,
      "step": 5145
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0002499161014645324,
      "loss": 2.1172,
      "step": 5150
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00024960247464765443,
      "loss": 2.206,
      "step": 5155
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002492887178076924,
      "loss": 2.1683,
      "step": 5160
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002489748317670982,
      "loss": 2.1484,
      "step": 5165
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002486608173486618,
      "loss": 2.1443,
      "step": 5170
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00024834667537550995,
      "loss": 2.16,
      "step": 5175
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002480324066711039,
      "loss": 2.1605,
      "step": 5180
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002477180120592368,
      "loss": 2.1522,
      "step": 5185
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002474034923640321,
      "loss": 2.1217,
      "step": 5190
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00024708884840994095,
      "loss": 2.2046,
      "step": 5195
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00024677408102174027,
      "loss": 2.1732,
      "step": 5200
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00024645919102453065,
      "loss": 2.1298,
      "step": 5205
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00024614417924373385,
      "loss": 2.1284,
      "step": 5210
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002458290465050911,
      "loss": 2.1746,
      "step": 5215
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0002455137936346606,
      "loss": 2.1432,
      "step": 5220
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002451984214588153,
      "loss": 2.1479,
      "step": 5225
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002448829308042412,
      "loss": 2.1422,
      "step": 5230
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002445673224979347,
      "loss": 2.1339,
      "step": 5235
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002442515973672005,
      "loss": 2.1753,
      "step": 5240
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002439357562396496,
      "loss": 2.1239,
      "step": 5245
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00024361979994319715,
      "loss": 2.1903,
      "step": 5250
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00024330372930606018,
      "loss": 2.1487,
      "step": 5255
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00024298754515675533,
      "loss": 2.1711,
      "step": 5260
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0002426712483240969,
      "loss": 2.1315,
      "step": 5265
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00024235483963719453,
      "loss": 2.191,
      "step": 5270
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00024203831992545096,
      "loss": 2.1441,
      "step": 5275
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0002417216900185602,
      "loss": 2.1895,
      "step": 5280
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00024140495074650487,
      "loss": 2.2274,
      "step": 5285
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00024108810293955438,
      "loss": 2.1568,
      "step": 5290
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002407711474282627,
      "loss": 2.1499,
      "step": 5295
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00024045408504346606,
      "loss": 2.1909,
      "step": 5300
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.00024013691661628074,
      "loss": 2.1499,
      "step": 5305
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00023981964297810118,
      "loss": 2.1664,
      "step": 5310
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00023950226496059743,
      "loss": 2.1702,
      "step": 5315
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00023918478339571335,
      "loss": 2.1252,
      "step": 5320
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00023886719911566404,
      "loss": 2.1412,
      "step": 5325
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002385495129529339,
      "loss": 2.1895,
      "step": 5330
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002382317257402745,
      "loss": 2.1669,
      "step": 5335
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0002379138383107021,
      "loss": 2.1341,
      "step": 5340
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0002375958514974959,
      "loss": 2.1688,
      "step": 5345
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00023727776613419543,
      "loss": 2.1949,
      "step": 5350
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1953125,
      "learning_rate": 0.00023695958305459854,
      "loss": 2.1693,
      "step": 5355
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002366413030927594,
      "loss": 2.1514,
      "step": 5360
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00023632292708298587,
      "loss": 2.1296,
      "step": 5365
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00023600445585983791,
      "loss": 2.1054,
      "step": 5370
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002356858902581248,
      "loss": 2.1498,
      "step": 5375
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00023536723111290328,
      "loss": 2.1251,
      "step": 5380
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0002350484792594754,
      "loss": 2.1655,
      "step": 5385
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.421875,
      "learning_rate": 0.00023472963553338613,
      "loss": 2.1002,
      "step": 5390
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.155019521713257,
      "eval_runtime": 171.4248,
      "eval_samples_per_second": 28.286,
      "eval_steps_per_second": 3.541,
      "step": 5390
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.00023441070077042118,
      "loss": 2.0676,
      "step": 5395
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002340916758066051,
      "loss": 2.0957,
      "step": 5400
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00023377256147819872,
      "loss": 2.0928,
      "step": 5405
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00023345335862169723,
      "loss": 2.1057,
      "step": 5410
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00023313406807382782,
      "loss": 2.0921,
      "step": 5415
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0002328146906715476,
      "loss": 2.0752,
      "step": 5420
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00023249522725204123,
      "loss": 2.0937,
      "step": 5425
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002321756786527189,
      "loss": 2.09,
      "step": 5430
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00023185604571121418,
      "loss": 2.1037,
      "step": 5435
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00023153632926538166,
      "loss": 2.1509,
      "step": 5440
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002312165301532948,
      "loss": 2.1312,
      "step": 5445
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00023089664921324373,
      "loss": 2.1174,
      "step": 5450
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00023057668728373315,
      "loss": 2.1352,
      "step": 5455
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00023025664520348005,
      "loss": 2.1088,
      "step": 5460
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.203125,
      "learning_rate": 0.00022993652381141138,
      "loss": 2.1271,
      "step": 5465
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00022961632394666222,
      "loss": 2.1281,
      "step": 5470
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002292960464485732,
      "loss": 2.1294,
      "step": 5475
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00022897569215668843,
      "loss": 2.1289,
      "step": 5480
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00022865526191075347,
      "loss": 2.0997,
      "step": 5485
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00022833475655071274,
      "loss": 2.1106,
      "step": 5490
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002280141769167078,
      "loss": 2.1005,
      "step": 5495
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002276935238490748,
      "loss": 2.0904,
      "step": 5500
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.203125,
      "learning_rate": 0.00022737279818834237,
      "loss": 2.114,
      "step": 5505
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002270520007752294,
      "loss": 2.1366,
      "step": 5510
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00022673113245064296,
      "loss": 2.1823,
      "step": 5515
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0002264101940556759,
      "loss": 2.1347,
      "step": 5520
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00022608918643160486,
      "loss": 2.1263,
      "step": 5525
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002257681104198878,
      "loss": 2.132,
      "step": 5530
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00022544696686216208,
      "loss": 2.1122,
      "step": 5535
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00022512575660024205,
      "loss": 2.1077,
      "step": 5540
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00022480448047611695,
      "loss": 2.1157,
      "step": 5545
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002244831393319486,
      "loss": 2.0735,
      "step": 5550
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00022416173401006932,
      "loss": 2.1343,
      "step": 5555
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00022384026535297963,
      "loss": 2.1022,
      "step": 5560
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00022351873420334615,
      "loss": 2.0955,
      "step": 5565
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0002231971414039991,
      "loss": 2.13,
      "step": 5570
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0002228754877979306,
      "loss": 2.0688,
      "step": 5575
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002225537742282919,
      "loss": 2.1275,
      "step": 5580
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002222320015383916,
      "loss": 2.0992,
      "step": 5585
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00022191017057169318,
      "loss": 2.1206,
      "step": 5590
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0002215882821718129,
      "loss": 2.148,
      "step": 5595
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002212663371825176,
      "loss": 2.1268,
      "step": 5600
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.21875,
      "learning_rate": 0.00022094433644772248,
      "loss": 2.0779,
      "step": 5605
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00022062228081148874,
      "loss": 2.1113,
      "step": 5610
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00022030017111802165,
      "loss": 2.1148,
      "step": 5615
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.203125,
      "learning_rate": 0.00021997800821166807,
      "loss": 2.1307,
      "step": 5620
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00021965579293691442,
      "loss": 2.1319,
      "step": 5625
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00021933352613838435,
      "loss": 2.1281,
      "step": 5630
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00021901120866083651,
      "loss": 2.1421,
      "step": 5635
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00021868884134916265,
      "loss": 2.1655,
      "step": 5640
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00021836642504838473,
      "loss": 2.1082,
      "step": 5645
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00021804396060365355,
      "loss": 2.1059,
      "step": 5650
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00021772144886024583,
      "loss": 2.1674,
      "step": 5655
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00021739889066356232,
      "loss": 2.1147,
      "step": 5660
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00021707628685912572,
      "loss": 2.12,
      "step": 5665
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00021675363829257803,
      "loss": 2.0997,
      "step": 5670
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00021643094580967874,
      "loss": 2.0945,
      "step": 5675
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00021610821025630243,
      "loss": 2.1583,
      "step": 5680
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00021578543247843647,
      "loss": 2.1116,
      "step": 5685
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00021546261332217918,
      "loss": 2.0938,
      "step": 5690
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00021513975363373703,
      "loss": 2.086,
      "step": 5695
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00021481685425942302,
      "loss": 2.1388,
      "step": 5700
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00021449391604565392,
      "loss": 2.0998,
      "step": 5705
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00021417093983894844,
      "loss": 2.1114,
      "step": 5710
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002138479264859249,
      "loss": 2.1571,
      "step": 5715
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.000213524876833299,
      "loss": 2.1065,
      "step": 5720
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00021320179172788155,
      "loss": 2.1273,
      "step": 5725
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0002128786720165763,
      "loss": 2.1065,
      "step": 5730
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00021255551854637762,
      "loss": 2.1065,
      "step": 5735
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00021223233216436857,
      "loss": 2.0759,
      "step": 5740
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.21875,
      "learning_rate": 0.0002119091137177183,
      "loss": 2.0995,
      "step": 5745
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00021158586405368017,
      "loss": 2.1513,
      "step": 5750
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002112625840195893,
      "loss": 2.1142,
      "step": 5755
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002109392744628603,
      "loss": 2.1145,
      "step": 5760
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00021061593623098533,
      "loss": 2.1038,
      "step": 5765
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00021029257017153162,
      "loss": 2.1134,
      "step": 5770
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00020996917713213945,
      "loss": 2.0712,
      "step": 5775
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00020964575796051974,
      "loss": 2.1444,
      "step": 5780
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00020932231350445188,
      "loss": 2.1248,
      "step": 5785
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0002089988446117817,
      "loss": 2.1342,
      "step": 5790
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00020867535213041883,
      "loss": 2.1064,
      "step": 5795
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00020835183690833496,
      "loss": 2.1467,
      "step": 5800
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00020802829979356134,
      "loss": 2.093,
      "step": 5805
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.203125,
      "learning_rate": 0.0002077047416341864,
      "loss": 2.0891,
      "step": 5810
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00020738116327835413,
      "loss": 2.1186,
      "step": 5815
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00020705756557426108,
      "loss": 2.1244,
      "step": 5820
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00020673394937015477,
      "loss": 2.1047,
      "step": 5825
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0002064103155143311,
      "loss": 2.0977,
      "step": 5830
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00020608666485513215,
      "loss": 2.1302,
      "step": 5835
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00020576299824094432,
      "loss": 2.1351,
      "step": 5840
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00020543931652019555,
      "loss": 2.0831,
      "step": 5845
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00020511562054135354,
      "loss": 2.1272,
      "step": 5850
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002047919111529234,
      "loss": 2.1544,
      "step": 5855
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002044681892034452,
      "loss": 2.135,
      "step": 5860
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00020414445554149208,
      "loss": 2.1514,
      "step": 5865
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00020382071101566788,
      "loss": 2.067,
      "step": 5870
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00020349695647460485,
      "loss": 2.1237,
      "step": 5875
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00020317319276696161,
      "loss": 2.1594,
      "step": 5880
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00020284942074142066,
      "loss": 2.1181,
      "step": 5885
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002025256412466864,
      "loss": 2.1467,
      "step": 5890
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00020220185513148277,
      "loss": 2.1358,
      "step": 5895
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00020187806324455104,
      "loss": 2.1006,
      "step": 5900
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00020155426643464773,
      "loss": 2.1606,
      "step": 5905
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00020123046555054215,
      "loss": 2.1072,
      "step": 5910
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00020090666144101436,
      "loss": 2.0934,
      "step": 5915
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00020058285495485275,
      "loss": 2.086,
      "step": 5920
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00020025904694085202,
      "loss": 2.1406,
      "step": 5925
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00019993523824781104,
      "loss": 2.1658,
      "step": 5930
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001996114297245301,
      "loss": 2.0964,
      "step": 5935
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001992876222198094,
      "loss": 2.1084,
      "step": 5940
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00019896381658244622,
      "loss": 2.0951,
      "step": 5945
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00019864001366123307,
      "loss": 2.1141,
      "step": 5950
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00019831621430495532,
      "loss": 2.0948,
      "step": 5955
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00019799241936238908,
      "loss": 2.1407,
      "step": 5960
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00019766862968229865,
      "loss": 2.1298,
      "step": 5965
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00019734484611343467,
      "loss": 2.1166,
      "step": 5970
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00019702106950453193,
      "loss": 2.0888,
      "step": 5975
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00019669730070430663,
      "loss": 2.12,
      "step": 5980
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0001963735405614549,
      "loss": 2.1181,
      "step": 5985
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00019604978992464976,
      "loss": 2.0892,
      "step": 5990
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00019572604964253972,
      "loss": 2.1449,
      "step": 5995
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00019540232056374578,
      "loss": 2.0995,
      "step": 6000
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001950786035368598,
      "loss": 2.138,
      "step": 6005
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00019475489941044204,
      "loss": 2.1263,
      "step": 6010
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00019443120903301871,
      "loss": 2.1197,
      "step": 6015
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00019410753325308042,
      "loss": 2.0856,
      "step": 6020
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00019378387291907909,
      "loss": 2.1015,
      "step": 6025
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0001934602288794263,
      "loss": 2.0782,
      "step": 6030
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.21875,
      "learning_rate": 0.00019313660198249107,
      "loss": 2.1392,
      "step": 6035
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00019281299307659713,
      "loss": 2.0854,
      "step": 6040
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00019248940301002155,
      "loss": 2.1192,
      "step": 6045
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00019216583263099147,
      "loss": 2.1065,
      "step": 6050
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001918422827876829,
      "loss": 2.0768,
      "step": 6055
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00019151875432821773,
      "loss": 2.0851,
      "step": 6060
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00019119524810066175,
      "loss": 2.1027,
      "step": 6065
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001908717649530228,
      "loss": 2.1008,
      "step": 6070
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0001905483057332479,
      "loss": 2.0943,
      "step": 6075
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00019022487128922148,
      "loss": 2.1203,
      "step": 6080
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001899014624687631,
      "loss": 2.1122,
      "step": 6085
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00018957808011962486,
      "loss": 2.1437,
      "step": 6090
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00018925472508948992,
      "loss": 2.073,
      "step": 6095
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00018893139822596938,
      "loss": 2.1426,
      "step": 6100
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00018860810037660085,
      "loss": 2.0974,
      "step": 6105
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00018828483238884564,
      "loss": 2.1141,
      "step": 6110
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00018796159511008702,
      "loss": 2.1222,
      "step": 6115
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00018763838938762756,
      "loss": 2.1244,
      "step": 6120
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001873152160686871,
      "loss": 2.141,
      "step": 6125
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00018699207600040077,
      "loss": 2.0931,
      "step": 6130
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00018666897002981626,
      "loss": 2.104,
      "step": 6135
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00018634589900389217,
      "loss": 2.0929,
      "step": 6140
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00018602286376949515,
      "loss": 2.08,
      "step": 6145
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00018569986517339844,
      "loss": 2.0757,
      "step": 6150
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00018537690406227888,
      "loss": 2.1296,
      "step": 6155
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00018505398128271515,
      "loss": 2.113,
      "step": 6160
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001847310976811856,
      "loss": 2.1245,
      "step": 6165
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00018440825410406575,
      "loss": 2.1231,
      "step": 6170
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00018408545139762627,
      "loss": 2.1119,
      "step": 6175
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00018376269040803057,
      "loss": 2.1047,
      "step": 6180
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001834399719813328,
      "loss": 2.0886,
      "step": 6185
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00018311729696347562,
      "loss": 2.1106,
      "step": 6190
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001827946662002877,
      "loss": 2.0957,
      "step": 6195
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.212890625,
      "learning_rate": 0.000182472080537482,
      "loss": 2.1069,
      "step": 6200
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00018214954082065282,
      "loss": 2.1129,
      "step": 6205
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00018182704789527452,
      "loss": 2.147,
      "step": 6210
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00018150460260669846,
      "loss": 2.1113,
      "step": 6215
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001811822058001512,
      "loss": 2.0983,
      "step": 6220
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00018085985832073237,
      "loss": 2.1165,
      "step": 6225
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00018053756101341206,
      "loss": 2.0929,
      "step": 6230
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.000180215314723029,
      "loss": 2.1439,
      "step": 6235
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001798931202942882,
      "loss": 2.1208,
      "step": 6240
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0001795709785717586,
      "loss": 2.1192,
      "step": 6245
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00017924889039987117,
      "loss": 2.1322,
      "step": 6250
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00017892685662291622,
      "loss": 2.1014,
      "step": 6255
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0001786048780850418,
      "loss": 2.1299,
      "step": 6260
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00017828295563025091,
      "loss": 2.0805,
      "step": 6265
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00017796109010239977,
      "loss": 2.1155,
      "step": 6270
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00017763928234519518,
      "loss": 2.1166,
      "step": 6275
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001773175332021925,
      "loss": 2.1122,
      "step": 6280
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00017699584351679363,
      "loss": 2.1151,
      "step": 6285
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001766742141322444,
      "loss": 2.1076,
      "step": 6290
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00017635264589163275,
      "loss": 2.1442,
      "step": 6295
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001760311396378863,
      "loss": 2.1386,
      "step": 6300
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00017570969621377003,
      "loss": 2.1099,
      "step": 6305
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00017538831646188443,
      "loss": 2.1427,
      "step": 6310
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00017506700122466297,
      "loss": 2.1288,
      "step": 6315
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00017474575134437007,
      "loss": 2.1197,
      "step": 6320
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00017442456766309877,
      "loss": 2.0971,
      "step": 6325
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00017410345102276872,
      "loss": 2.1008,
      "step": 6330
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0001737824022651236,
      "loss": 2.102,
      "step": 6335
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00017346142223172926,
      "loss": 2.1308,
      "step": 6340
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0001731405117639715,
      "loss": 2.1104,
      "step": 6345
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0001728196717030536,
      "loss": 2.0971,
      "step": 6350
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.21875,
      "learning_rate": 0.00017249890288999453,
      "loss": 2.0961,
      "step": 6355
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00017217820616562615,
      "loss": 2.0973,
      "step": 6360
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00017185758237059172,
      "loss": 2.1169,
      "step": 6365
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.000171537032345343,
      "loss": 2.1253,
      "step": 6370
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00017121655693013856,
      "loss": 2.1127,
      "step": 6375
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0001708961569650414,
      "loss": 2.0914,
      "step": 6380
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00017057583328991668,
      "loss": 2.0782,
      "step": 6385
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00017025558674442972,
      "loss": 2.1294,
      "step": 6390
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00016993541816804334,
      "loss": 2.1139,
      "step": 6395
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001696153284000163,
      "loss": 2.0751,
      "step": 6400
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00016929531827940066,
      "loss": 2.11,
      "step": 6405
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.21875,
      "learning_rate": 0.00016897538864503968,
      "loss": 2.1733,
      "step": 6410
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00016865554033556574,
      "loss": 2.1227,
      "step": 6415
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00016833577418939785,
      "loss": 2.1076,
      "step": 6420
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00016801609104473986,
      "loss": 2.113,
      "step": 6425
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0001676964917395779,
      "loss": 2.1067,
      "step": 6430
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00016737697711167836,
      "loss": 2.094,
      "step": 6435
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00016705754799858585,
      "loss": 2.0662,
      "step": 6440
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0001667382052376204,
      "loss": 2.1005,
      "step": 6445
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00016641894966587618,
      "loss": 2.1408,
      "step": 6450
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00016609978212021843,
      "loss": 2.104,
      "step": 6455
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00016578070343728181,
      "loss": 2.1289,
      "step": 6460
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00016546171445346811,
      "loss": 2.0986,
      "step": 6465
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00016514281600494378,
      "loss": 2.1056,
      "step": 6470
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001648240089276382,
      "loss": 2.1461,
      "step": 6475
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00016450529405724097,
      "loss": 2.0915,
      "step": 6480
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00016418667222920029,
      "loss": 2.086,
      "step": 6485
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00016386814427872025,
      "loss": 2.1414,
      "step": 6490
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00016354971104075888,
      "loss": 2.128,
      "step": 6495
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.000163231373350026,
      "loss": 2.1093,
      "step": 6500
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00016291313204098092,
      "loss": 2.1172,
      "step": 6505
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00016259498794783043,
      "loss": 2.1398,
      "step": 6510
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00016227694190452626,
      "loss": 2.1342,
      "step": 6515
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00016195899474476345,
      "loss": 2.1023,
      "step": 6520
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00016164114730197744,
      "loss": 2.0859,
      "step": 6525
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00016132340040934254,
      "loss": 2.1153,
      "step": 6530
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.21875,
      "learning_rate": 0.00016100575489976947,
      "loss": 2.1159,
      "step": 6535
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00016068821160590308,
      "loss": 2.08,
      "step": 6540
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00016037077136012054,
      "loss": 2.1129,
      "step": 6545
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001600534349945285,
      "loss": 2.1458,
      "step": 6550
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00015973620334096159,
      "loss": 2.1361,
      "step": 6555
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00015941907723097994,
      "loss": 2.1387,
      "step": 6560
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0001591020574958669,
      "loss": 2.1254,
      "step": 6565
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.21875,
      "learning_rate": 0.00015878514496662715,
      "loss": 2.1272,
      "step": 6570
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001584683404739841,
      "loss": 2.1448,
      "step": 6575
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00015815164484837832,
      "loss": 2.1431,
      "step": 6580
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00015783505891996466,
      "loss": 2.1139,
      "step": 6585
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00015751858351861054,
      "loss": 2.1558,
      "step": 6590
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001572022194738938,
      "loss": 2.1249,
      "step": 6595
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00015688596761510005,
      "loss": 2.1071,
      "step": 6600
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.21875,
      "learning_rate": 0.00015656982877122134,
      "loss": 2.0894,
      "step": 6605
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0001562538037709529,
      "loss": 2.1351,
      "step": 6610
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00015593789344269188,
      "loss": 2.1364,
      "step": 6615
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00015562209861453487,
      "loss": 2.078,
      "step": 6620
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00015530642011427542,
      "loss": 2.1231,
      "step": 6625
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00015499085876940255,
      "loss": 2.1232,
      "step": 6630
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.21875,
      "learning_rate": 0.00015467541540709772,
      "loss": 2.1206,
      "step": 6635
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00015436009085423354,
      "loss": 2.0971,
      "step": 6640
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.220703125,
      "learning_rate": 0.000154044885937371,
      "loss": 2.0972,
      "step": 6645
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001537298014827573,
      "loss": 2.1167,
      "step": 6650
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00015341483831632434,
      "loss": 2.0823,
      "step": 6655
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00015309999726368555,
      "loss": 2.138,
      "step": 6660
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001527852791501347,
      "loss": 2.1009,
      "step": 6665
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00015247068480064307,
      "loss": 2.0907,
      "step": 6670
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.21875,
      "learning_rate": 0.00015215621503985758,
      "loss": 2.0969,
      "step": 6675
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00015184187069209858,
      "loss": 2.1056,
      "step": 6680
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00015152765258135754,
      "loss": 2.0957,
      "step": 6685
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00015121356153129526,
      "loss": 2.1561,
      "step": 6690
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00015089959836523927,
      "loss": 2.1067,
      "step": 6695
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00015058576390618205,
      "loss": 2.1244,
      "step": 6700
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00015027205897677855,
      "loss": 2.1364,
      "step": 6705
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00014995848439934418,
      "loss": 2.0978,
      "step": 6710
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00014964504099585283,
      "loss": 2.1216,
      "step": 6715
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00014933172958793436,
      "loss": 2.0968,
      "step": 6720
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00014901855099687275,
      "loss": 2.0871,
      "step": 6725
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00014870550604360373,
      "loss": 2.1265,
      "step": 6730
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0001483925955487129,
      "loss": 2.0967,
      "step": 6735
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00014807982033243313,
      "loss": 2.1202,
      "step": 6740
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00014776718121464283,
      "loss": 2.0886,
      "step": 6745
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00014745467901486377,
      "loss": 2.112,
      "step": 6750
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00014714231455225862,
      "loss": 2.1163,
      "step": 6755
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00014683008864562917,
      "loss": 2.1304,
      "step": 6760
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00014651800211341385,
      "loss": 2.0945,
      "step": 6765
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001462060557736858,
      "loss": 2.1094,
      "step": 6770
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.21875,
      "learning_rate": 0.00014589425044415075,
      "loss": 2.0669,
      "step": 6775
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001455825869421447,
      "loss": 2.1211,
      "step": 6780
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00014527106608463206,
      "loss": 2.1007,
      "step": 6785
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.22265625,
      "learning_rate": 0.000144959688688203,
      "loss": 2.1065,
      "step": 6790
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00014464845556907196,
      "loss": 2.089,
      "step": 6795
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.000144337367543075,
      "loss": 2.14,
      "step": 6800
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00014402642542566782,
      "loss": 2.1439,
      "step": 6805
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00014371563003192392,
      "loss": 2.0581,
      "step": 6810
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001434049821765318,
      "loss": 2.1186,
      "step": 6815
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00014309448267379353,
      "loss": 2.1526,
      "step": 6820
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00014278413233762214,
      "loss": 2.071,
      "step": 6825
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00014247393198153974,
      "loss": 2.1107,
      "step": 6830
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001421638824186753,
      "loss": 2.1339,
      "step": 6835
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0001418539844617623,
      "loss": 2.1236,
      "step": 6840
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00014154423892313712,
      "loss": 2.1086,
      "step": 6845
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00014123464661473646,
      "loss": 2.0773,
      "step": 6850
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00014092520834809534,
      "loss": 2.1404,
      "step": 6855
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001406159249343451,
      "loss": 2.0654,
      "step": 6860
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001403067971842109,
      "loss": 2.0919,
      "step": 6865
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.228515625,
      "learning_rate": 0.00013999782590801022,
      "loss": 2.0763,
      "step": 6870
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0001396890119156501,
      "loss": 2.109,
      "step": 6875
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00013938035601662545,
      "loss": 2.117,
      "step": 6880
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00013907185902001663,
      "loss": 2.1378,
      "step": 6885
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00013876352173448764,
      "loss": 2.1208,
      "step": 6890
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00013845534496828368,
      "loss": 2.1245,
      "step": 6895
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00013814732952922918,
      "loss": 2.1393,
      "step": 6900
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00013783947622472587,
      "loss": 2.1493,
      "step": 6905
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001375317858617502,
      "loss": 2.1436,
      "step": 6910
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001372242592468518,
      "loss": 2.1092,
      "step": 6915
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.21875,
      "learning_rate": 0.00013691689718615079,
      "loss": 2.1213,
      "step": 6920
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00013660970048533607,
      "loss": 2.0764,
      "step": 6925
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00013630266994966314,
      "loss": 2.124,
      "step": 6930
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001359958063839518,
      "loss": 2.1211,
      "step": 6935
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00013568911059258436,
      "loss": 2.0699,
      "step": 6940
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00013538258337950302,
      "loss": 2.0943,
      "step": 6945
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001350762255482085,
      "loss": 2.1063,
      "step": 6950
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001347700379017572,
      "loss": 2.1212,
      "step": 6955
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00013446402124275947,
      "loss": 2.0913,
      "step": 6960
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00013415817637337768,
      "loss": 2.1277,
      "step": 6965
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00013385250409532343,
      "loss": 2.1266,
      "step": 6970
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0001335470052098565,
      "loss": 2.0951,
      "step": 6975
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00013324168051778162,
      "loss": 2.1073,
      "step": 6980
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00013293653081944716,
      "loss": 2.1286,
      "step": 6985
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00013263155691474286,
      "loss": 2.0843,
      "step": 6990
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001323267596030973,
      "loss": 2.1228,
      "step": 6995
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0001320221396834767,
      "loss": 2.126,
      "step": 7000
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001317176979543817,
      "loss": 2.0865,
      "step": 7005
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.21875,
      "learning_rate": 0.00013141343521384634,
      "loss": 2.1284,
      "step": 7010
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.228515625,
      "learning_rate": 0.00013110935225943516,
      "loss": 2.0905,
      "step": 7015
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001308054498882414,
      "loss": 2.0852,
      "step": 7020
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00013050172889688536,
      "loss": 2.1067,
      "step": 7025
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00013019819008151128,
      "loss": 2.0757,
      "step": 7030
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0001298948342377864,
      "loss": 2.1301,
      "step": 7035
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00012959166216089797,
      "loss": 2.0895,
      "step": 7040
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00012928867464555177,
      "loss": 2.0816,
      "step": 7045
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001289858724859697,
      "loss": 2.1101,
      "step": 7050
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001286832564758876,
      "loss": 2.1058,
      "step": 7055
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00012838082740855371,
      "loss": 2.1263,
      "step": 7060
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00012807858607672597,
      "loss": 2.0942,
      "step": 7065
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00012777653327267037,
      "loss": 2.1127,
      "step": 7070
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00012747466978815862,
      "loss": 2.0978,
      "step": 7075
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00012717299641446612,
      "loss": 2.1021,
      "step": 7080
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00012687151394237005,
      "loss": 2.1147,
      "step": 7085
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00012657022316214705,
      "loss": 2.0918,
      "step": 7090
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00012626912486357147,
      "loss": 2.1089,
      "step": 7095
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00012596821983591289,
      "loss": 2.1268,
      "step": 7100
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001256675088679345,
      "loss": 2.1026,
      "step": 7105
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00012536699274789059,
      "loss": 2.0627,
      "step": 7110
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0001250666722635247,
      "loss": 2.0675,
      "step": 7115
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00012476654820206773,
      "loss": 2.1035,
      "step": 7120
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001244666213502355,
      "loss": 2.1104,
      "step": 7125
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.21875,
      "learning_rate": 0.00012416689249422714,
      "loss": 2.0844,
      "step": 7130
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00012386736241972246,
      "loss": 2.1405,
      "step": 7135
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00012356803191188034,
      "loss": 2.1525,
      "step": 7140
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001232689017553366,
      "loss": 2.1443,
      "step": 7145
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00012296997273420184,
      "loss": 2.1317,
      "step": 7150
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00012267124563205946,
      "loss": 2.0839,
      "step": 7155
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0001223727212319633,
      "loss": 2.0955,
      "step": 7160
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00012207440031643637,
      "loss": 2.1137,
      "step": 7165
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0001217762836674678,
      "loss": 2.0972,
      "step": 7170
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.21875,
      "learning_rate": 0.00012147837206651148,
      "loss": 2.1198,
      "step": 7175
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00012118066629448388,
      "loss": 2.0673,
      "step": 7180
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00012088316713176166,
      "loss": 2.1093,
      "step": 7185
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00012058587535818036,
      "loss": 2.0892,
      "step": 7190
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.21875,
      "learning_rate": 0.00012028879175303137,
      "loss": 2.1102,
      "step": 7195
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00011999191709506072,
      "loss": 2.1176,
      "step": 7200
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.21875,
      "learning_rate": 0.00011969525216246673,
      "loss": 2.0999,
      "step": 7205
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00011939879773289768,
      "loss": 2.0787,
      "step": 7210
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00011910255458345055,
      "loss": 2.1507,
      "step": 7215
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00011880652349066798,
      "loss": 2.1248,
      "step": 7220
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00011851070523053707,
      "loss": 2.0911,
      "step": 7225
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00011821510057848695,
      "loss": 2.1235,
      "step": 7230
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00011791971030938662,
      "loss": 2.1035,
      "step": 7235
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.21875,
      "learning_rate": 0.00011762453519754357,
      "loss": 2.1228,
      "step": 7240
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00011732957601670076,
      "loss": 2.0994,
      "step": 7245
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00011703483354003553,
      "loss": 2.1336,
      "step": 7250
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.21875,
      "learning_rate": 0.00011674030854015696,
      "loss": 2.108,
      "step": 7255
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00011644600178910421,
      "loss": 2.1178,
      "step": 7260
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001161519140583442,
      "loss": 2.1162,
      "step": 7265
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0001158580461187698,
      "loss": 2.0816,
      "step": 7270
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00011556439874069773,
      "loss": 2.1076,
      "step": 7275
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00011527097269386655,
      "loss": 2.1465,
      "step": 7280
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00011497776874743471,
      "loss": 2.1064,
      "step": 7285
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00011468478766997832,
      "loss": 2.0849,
      "step": 7290
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00011439203022948935,
      "loss": 2.1352,
      "step": 7295
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00011409949719337376,
      "loss": 2.1146,
      "step": 7300
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00011380718932844882,
      "loss": 2.1264,
      "step": 7305
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00011351510740094205,
      "loss": 2.1164,
      "step": 7310
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00011322325217648839,
      "loss": 2.104,
      "step": 7315
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00011293162442012866,
      "loss": 2.0908,
      "step": 7320
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00011264022489630737,
      "loss": 2.1208,
      "step": 7325
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00011234905436887078,
      "loss": 2.1089,
      "step": 7330
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0001120581136010649,
      "loss": 2.1016,
      "step": 7335
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.21875,
      "learning_rate": 0.00011176740335553333,
      "loss": 2.1079,
      "step": 7340
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00011147692439431572,
      "loss": 2.1365,
      "step": 7345
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.00011118667747884517,
      "loss": 2.0991,
      "step": 7350
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0001108966633699466,
      "loss": 2.1581,
      "step": 7355
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00011060688282783469,
      "loss": 2.1253,
      "step": 7360
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00011031733661211184,
      "loss": 2.0904,
      "step": 7365
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00011002802548176623,
      "loss": 2.0793,
      "step": 7370
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00010973895019516974,
      "loss": 2.1235,
      "step": 7375
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.21875,
      "learning_rate": 0.00010945011151007634,
      "loss": 2.1639,
      "step": 7380
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00010916151018361929,
      "loss": 2.106,
      "step": 7385
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00010887314697230997,
      "loss": 2.1463,
      "step": 7390
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001085850226320357,
      "loss": 2.1063,
      "step": 7395
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00010829713791805738,
      "loss": 2.1395,
      "step": 7400
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00010800949358500794,
      "loss": 2.1373,
      "step": 7405
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00010772209038689003,
      "loss": 2.1083,
      "step": 7410
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00010743492907707455,
      "loss": 2.1363,
      "step": 7415
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00010714801040829796,
      "loss": 2.1342,
      "step": 7420
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00010686133513266079,
      "loss": 2.1208,
      "step": 7425
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00010657490400162584,
      "loss": 2.118,
      "step": 7430
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00010628871776601542,
      "loss": 2.1088,
      "step": 7435
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00010600277717601042,
      "loss": 2.1007,
      "step": 7440
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00010571708298114751,
      "loss": 2.1492,
      "step": 7445
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00010543163593031753,
      "loss": 2.1433,
      "step": 7450
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00010514643677176354,
      "loss": 2.1069,
      "step": 7455
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00010486148625307868,
      "loss": 2.1463,
      "step": 7460
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00010457678512120463,
      "loss": 2.0955,
      "step": 7465
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00010429233412242887,
      "loss": 2.0892,
      "step": 7470
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00010400813400238368,
      "loss": 2.0786,
      "step": 7475
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00010372418550604351,
      "loss": 2.1339,
      "step": 7480
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.21875,
      "learning_rate": 0.00010344048937772297,
      "loss": 2.1183,
      "step": 7485
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00010315704636107564,
      "loss": 2.1244,
      "step": 7490
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001028738571990913,
      "loss": 2.0664,
      "step": 7495
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001025909226340944,
      "loss": 2.1083,
      "step": 7500
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001023082434077419,
      "loss": 2.1205,
      "step": 7505
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00010202582026102153,
      "loss": 2.1116,
      "step": 7510
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00010174365393424992,
      "loss": 2.1348,
      "step": 7515
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00010146174516707005,
      "loss": 2.1105,
      "step": 7520
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.21875,
      "learning_rate": 0.00010118009469845027,
      "loss": 2.1106,
      "step": 7525
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00010089870326668154,
      "loss": 2.1576,
      "step": 7530
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.21875,
      "learning_rate": 0.00010061757160937586,
      "loss": 2.1146,
      "step": 7535
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00010033670046346433,
      "loss": 2.0823,
      "step": 7540
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00010005609056519513,
      "loss": 2.0989,
      "step": 7545
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2255859375,
      "learning_rate": 9.97757426501317e-05,
      "loss": 2.1095,
      "step": 7550
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.949565745315055e-05,
      "loss": 2.1397,
      "step": 7555
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.921583570843986e-05,
      "loss": 2.0911,
      "step": 7560
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.220703125,
      "learning_rate": 9.893627814949693e-05,
      "loss": 2.1112,
      "step": 7565
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.865698550912667e-05,
      "loss": 2.0844,
      "step": 7570
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.216796875,
      "learning_rate": 9.837795851943954e-05,
      "loss": 2.1082,
      "step": 7575
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.2138671875,
      "learning_rate": 9.809919791184963e-05,
      "loss": 2.1108,
      "step": 7580
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.216796875,
      "learning_rate": 9.782070441707276e-05,
      "loss": 2.1201,
      "step": 7585
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.212890625,
      "learning_rate": 9.754247876512457e-05,
      "loss": 2.098,
      "step": 7590
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.2236328125,
      "learning_rate": 9.726452168531879e-05,
      "loss": 2.1319,
      "step": 7595
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.2265625,
      "learning_rate": 9.698683390626476e-05,
      "loss": 2.145,
      "step": 7600
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.67094161558661e-05,
      "loss": 2.1141,
      "step": 7605
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.216796875,
      "learning_rate": 9.64322691613188e-05,
      "loss": 2.0464,
      "step": 7610
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.21875,
      "learning_rate": 9.615539364910881e-05,
      "loss": 2.1155,
      "step": 7615
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.22265625,
      "learning_rate": 9.587879034501062e-05,
      "loss": 2.1043,
      "step": 7620
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.228515625,
      "learning_rate": 9.5602459974085e-05,
      "loss": 2.1154,
      "step": 7625
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2138671875,
      "learning_rate": 9.532640326067763e-05,
      "loss": 2.0965,
      "step": 7630
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.505062092841644e-05,
      "loss": 2.0938,
      "step": 7635
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2236328125,
      "learning_rate": 9.477511370021026e-05,
      "loss": 2.0867,
      "step": 7640
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.44998822982471e-05,
      "loss": 2.0947,
      "step": 7645
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.216796875,
      "learning_rate": 9.422492744399137e-05,
      "loss": 2.1152,
      "step": 7650
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2255859375,
      "learning_rate": 9.395024985818309e-05,
      "loss": 2.0936,
      "step": 7655
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.212890625,
      "learning_rate": 9.367585026083518e-05,
      "loss": 2.1191,
      "step": 7660
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2197265625,
      "learning_rate": 9.340172937123201e-05,
      "loss": 2.1362,
      "step": 7665
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.2197265625,
      "learning_rate": 9.312788790792728e-05,
      "loss": 2.1423,
      "step": 7670
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.21875,
      "learning_rate": 9.285432658874216e-05,
      "loss": 2.1184,
      "step": 7675
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.21484375,
      "learning_rate": 9.258104613076385e-05,
      "loss": 2.1006,
      "step": 7680
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.230804725034274e-05,
      "loss": 2.0857,
      "step": 7685
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.203533066309168e-05,
      "loss": 2.1165,
      "step": 7690
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.228515625,
      "learning_rate": 9.176289708388329e-05,
      "loss": 2.1702,
      "step": 7695
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.149074722684814e-05,
      "loss": 2.1102,
      "step": 7700
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.121888180537348e-05,
      "loss": 2.084,
      "step": 7705
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2138671875,
      "learning_rate": 9.094730153210076e-05,
      "loss": 2.1463,
      "step": 7710
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.067600711892396e-05,
      "loss": 2.1166,
      "step": 7715
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.220703125,
      "learning_rate": 9.04049992769877e-05,
      "loss": 2.1383,
      "step": 7720
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.216796875,
      "learning_rate": 9.013427871668562e-05,
      "loss": 2.11,
      "step": 7725
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.986384614765817e-05,
      "loss": 2.1083,
      "step": 7730
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.21875,
      "learning_rate": 8.959370227879067e-05,
      "loss": 2.1375,
      "step": 7735
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.932384781821208e-05,
      "loss": 2.111,
      "step": 7740
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.905428347329245e-05,
      "loss": 2.1026,
      "step": 7745
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.21875,
      "learning_rate": 8.878500995064148e-05,
      "loss": 2.1506,
      "step": 7750
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.851602795610646e-05,
      "loss": 2.1432,
      "step": 7755
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.824733819477051e-05,
      "loss": 2.0726,
      "step": 7760
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2109375,
      "learning_rate": 8.797894137095077e-05,
      "loss": 2.0853,
      "step": 7765
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2236328125,
      "learning_rate": 8.77108381881964e-05,
      "loss": 2.1539,
      "step": 7770
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.220703125,
      "learning_rate": 8.744302934928701e-05,
      "loss": 2.0791,
      "step": 7775
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.717551555623051e-05,
      "loss": 2.0752,
      "step": 7780
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.220703125,
      "learning_rate": 8.690829751026141e-05,
      "loss": 2.1238,
      "step": 7785
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2333984375,
      "learning_rate": 8.664137591183901e-05,
      "loss": 2.1237,
      "step": 7790
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.637475146064554e-05,
      "loss": 2.1183,
      "step": 7795
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.220703125,
      "learning_rate": 8.610842485558428e-05,
      "loss": 2.1416,
      "step": 7800
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2255859375,
      "learning_rate": 8.584239679477775e-05,
      "loss": 2.1213,
      "step": 7805
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.557666797556612e-05,
      "loss": 2.1557,
      "step": 7810
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.228515625,
      "learning_rate": 8.531123909450476e-05,
      "loss": 2.0979,
      "step": 7815
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.21875,
      "learning_rate": 8.504611084736305e-05,
      "loss": 2.1238,
      "step": 7820
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2138671875,
      "learning_rate": 8.478128392912239e-05,
      "loss": 2.1331,
      "step": 7825
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.451675903397416e-05,
      "loss": 2.114,
      "step": 7830
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.21875,
      "learning_rate": 8.425253685531809e-05,
      "loss": 2.152,
      "step": 7835
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.2109375,
      "learning_rate": 8.39886180857604e-05,
      "loss": 2.1003,
      "step": 7840
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.220703125,
      "learning_rate": 8.372500341711199e-05,
      "loss": 2.1093,
      "step": 7845
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.212890625,
      "learning_rate": 8.346169354038657e-05,
      "loss": 2.1132,
      "step": 7850
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.2099609375,
      "learning_rate": 8.319868914579894e-05,
      "loss": 2.0899,
      "step": 7855
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.2236328125,
      "learning_rate": 8.293599092276332e-05,
      "loss": 2.1106,
      "step": 7860
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.224609375,
      "learning_rate": 8.267359955989087e-05,
      "loss": 2.1119,
      "step": 7865
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.2197265625,
      "learning_rate": 8.241151574498896e-05,
      "loss": 2.1188,
      "step": 7870
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.220703125,
      "learning_rate": 8.214974016505838e-05,
      "loss": 2.0889,
      "step": 7875
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.216796875,
      "learning_rate": 8.188827350629213e-05,
      "loss": 2.1212,
      "step": 7880
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.2236328125,
      "learning_rate": 8.162711645407335e-05,
      "loss": 2.111,
      "step": 7885
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.136626969297365e-05,
      "loss": 2.1206,
      "step": 7890
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.2138671875,
      "learning_rate": 8.110573390675125e-05,
      "loss": 2.0869,
      "step": 7895
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.2265625,
      "learning_rate": 8.084550977834915e-05,
      "loss": 2.1158,
      "step": 7900
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.220703125,
      "learning_rate": 8.058559798989362e-05,
      "loss": 2.1113,
      "step": 7905
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.220703125,
      "learning_rate": 8.032599922269206e-05,
      "loss": 2.1123,
      "step": 7910
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.216796875,
      "learning_rate": 8.006671415723108e-05,
      "loss": 2.1223,
      "step": 7915
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.980774347317548e-05,
      "loss": 2.0928,
      "step": 7920
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.21875,
      "learning_rate": 7.954908784936566e-05,
      "loss": 2.0741,
      "step": 7925
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.21875,
      "learning_rate": 7.929074796381618e-05,
      "loss": 2.1044,
      "step": 7930
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.21484375,
      "learning_rate": 7.903272449371395e-05,
      "loss": 2.119,
      "step": 7935
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.2236328125,
      "learning_rate": 7.87750181154167e-05,
      "loss": 2.1355,
      "step": 7940
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.851762950445057e-05,
      "loss": 2.133,
      "step": 7945
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.2216796875,
      "learning_rate": 7.826055933550891e-05,
      "loss": 2.1551,
      "step": 7950
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.800380828245051e-05,
      "loss": 2.1216,
      "step": 7955
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.220703125,
      "learning_rate": 7.774737701829747e-05,
      "loss": 2.0914,
      "step": 7960
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2255859375,
      "learning_rate": 7.749126621523363e-05,
      "loss": 2.1388,
      "step": 7965
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2236328125,
      "learning_rate": 7.723547654460285e-05,
      "loss": 2.0935,
      "step": 7970
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.21484375,
      "learning_rate": 7.698000867690724e-05,
      "loss": 2.091,
      "step": 7975
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.216796875,
      "learning_rate": 7.672486328180529e-05,
      "loss": 2.1153,
      "step": 7980
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2265625,
      "learning_rate": 7.647004102811013e-05,
      "loss": 2.1339,
      "step": 7985
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.216796875,
      "learning_rate": 7.621554258378818e-05,
      "loss": 2.1211,
      "step": 7990
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2216796875,
      "learning_rate": 7.596136861595651e-05,
      "loss": 2.1512,
      "step": 7995
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.2216796875,
      "learning_rate": 7.57075197908822e-05,
      "loss": 2.1514,
      "step": 8000
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.212890625,
      "learning_rate": 7.545399677397964e-05,
      "loss": 2.1236,
      "step": 8005
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.216796875,
      "learning_rate": 7.520080022980935e-05,
      "loss": 2.103,
      "step": 8010
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.2138671875,
      "learning_rate": 7.494793082207605e-05,
      "loss": 2.1457,
      "step": 8015
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.2197265625,
      "learning_rate": 7.46953892136268e-05,
      "loss": 2.1162,
      "step": 8020
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.216796875,
      "learning_rate": 7.444317606644973e-05,
      "loss": 2.1181,
      "step": 8025
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.220703125,
      "learning_rate": 7.419129204167151e-05,
      "loss": 2.1427,
      "step": 8030
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.2236328125,
      "learning_rate": 7.39397377995565e-05,
      "loss": 2.1352,
      "step": 8035
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.21875,
      "learning_rate": 7.368851399950447e-05,
      "loss": 2.0876,
      "step": 8040
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.2138671875,
      "learning_rate": 7.343762130004872e-05,
      "loss": 2.1071,
      "step": 8045
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.2216796875,
      "learning_rate": 7.318706035885507e-05,
      "loss": 2.0977,
      "step": 8050
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.216796875,
      "learning_rate": 7.29368318327195e-05,
      "loss": 2.1158,
      "step": 8055
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.21875,
      "learning_rate": 7.268693637756658e-05,
      "loss": 2.0564,
      "step": 8060
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.21875,
      "learning_rate": 7.243737464844787e-05,
      "loss": 2.1404,
      "step": 8065
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2197265625,
      "learning_rate": 7.218814729954005e-05,
      "loss": 2.1203,
      "step": 8070
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.220703125,
      "learning_rate": 7.193925498414357e-05,
      "loss": 2.1112,
      "step": 8075
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2265625,
      "learning_rate": 7.169069835468017e-05,
      "loss": 2.1359,
      "step": 8080
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2265625,
      "learning_rate": 7.144247806269213e-05,
      "loss": 2.0982,
      "step": 8085
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.220703125,
      "learning_rate": 7.119459475883983e-05,
      "loss": 2.0748,
      "step": 8090
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.094704909290036e-05,
      "loss": 2.1213,
      "step": 8095
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.220703125,
      "learning_rate": 7.069984171376571e-05,
      "loss": 2.1405,
      "step": 8100
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.2197265625,
      "learning_rate": 7.045297326944125e-05,
      "loss": 2.1144,
      "step": 8105
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.21875,
      "learning_rate": 7.020644440704376e-05,
      "loss": 2.0758,
      "step": 8110
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2392578125,
      "learning_rate": 6.996025577279986e-05,
      "loss": 2.1157,
      "step": 8115
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.971440801204454e-05,
      "loss": 2.1075,
      "step": 8120
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2255859375,
      "learning_rate": 6.946890176921915e-05,
      "loss": 2.1304,
      "step": 8125
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.220703125,
      "learning_rate": 6.922373768786954e-05,
      "loss": 2.1381,
      "step": 8130
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2138671875,
      "learning_rate": 6.897891641064511e-05,
      "loss": 2.078,
      "step": 8135
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2216796875,
      "learning_rate": 6.873443857929638e-05,
      "loss": 2.1653,
      "step": 8140
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2265625,
      "learning_rate": 6.849030483467367e-05,
      "loss": 2.1208,
      "step": 8145
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.21484375,
      "learning_rate": 6.82465158167253e-05,
      "loss": 2.0856,
      "step": 8150
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.220703125,
      "learning_rate": 6.800307216449615e-05,
      "loss": 2.1454,
      "step": 8155
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2197265625,
      "learning_rate": 6.775997451612548e-05,
      "loss": 2.1249,
      "step": 8160
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.2197265625,
      "learning_rate": 6.75172235088457e-05,
      "loss": 2.1168,
      "step": 8165
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2138671875,
      "learning_rate": 6.727481977898076e-05,
      "loss": 2.1398,
      "step": 8170
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2138671875,
      "learning_rate": 6.703276396194404e-05,
      "loss": 2.0943,
      "step": 8175
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2119140625,
      "learning_rate": 6.679105669223704e-05,
      "loss": 2.1336,
      "step": 8180
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.224609375,
      "learning_rate": 6.654969860344757e-05,
      "loss": 2.0909,
      "step": 8185
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2109375,
      "learning_rate": 6.630869032824821e-05,
      "loss": 2.1254,
      "step": 8190
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.606803249839448e-05,
      "loss": 2.1044,
      "step": 8195
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2216796875,
      "learning_rate": 6.582772574472325e-05,
      "loss": 2.1604,
      "step": 8200
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2373046875,
      "learning_rate": 6.558777069715138e-05,
      "loss": 2.1012,
      "step": 8205
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.21875,
      "learning_rate": 6.534816798467338e-05,
      "loss": 2.0951,
      "step": 8210
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.2177734375,
      "learning_rate": 6.510891823536054e-05,
      "loss": 2.0685,
      "step": 8215
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.21484375,
      "learning_rate": 6.487002207635877e-05,
      "loss": 2.1161,
      "step": 8220
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.2236328125,
      "learning_rate": 6.463148013388713e-05,
      "loss": 2.1549,
      "step": 8225
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.21875,
      "learning_rate": 6.439329303323616e-05,
      "loss": 2.1182,
      "step": 8230
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.415546139876629e-05,
      "loss": 2.0894,
      "step": 8235
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.216796875,
      "learning_rate": 6.391798585390614e-05,
      "loss": 2.1289,
      "step": 8240
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.2119140625,
      "learning_rate": 6.368086702115081e-05,
      "loss": 2.0734,
      "step": 8245
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.216796875,
      "learning_rate": 6.344410552206066e-05,
      "loss": 2.1034,
      "step": 8250
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.224609375,
      "learning_rate": 6.320770197725911e-05,
      "loss": 2.1449,
      "step": 8255
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.2177734375,
      "learning_rate": 6.297165700643117e-05,
      "loss": 2.0802,
      "step": 8260
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.273597122832224e-05,
      "loss": 2.1051,
      "step": 8265
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.212890625,
      "learning_rate": 6.250064526073598e-05,
      "loss": 2.1103,
      "step": 8270
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.224609375,
      "learning_rate": 6.226567972053286e-05,
      "loss": 2.0815,
      "step": 8275
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.21875,
      "learning_rate": 6.203107522362863e-05,
      "loss": 2.1086,
      "step": 8280
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2177734375,
      "learning_rate": 6.179683238499263e-05,
      "loss": 2.0913,
      "step": 8285
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2197265625,
      "learning_rate": 6.156295181864613e-05,
      "loss": 2.085,
      "step": 8290
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.220703125,
      "learning_rate": 6.132943413766077e-05,
      "loss": 2.0987,
      "step": 8295
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.21875,
      "learning_rate": 6.109627995415712e-05,
      "loss": 2.1141,
      "step": 8300
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.086348987930273e-05,
      "loss": 2.1072,
      "step": 8305
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.220703125,
      "learning_rate": 6.0631064523310756e-05,
      "loss": 2.1292,
      "step": 8310
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2216796875,
      "learning_rate": 6.039900449543836e-05,
      "loss": 2.1378,
      "step": 8315
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.21875,
      "learning_rate": 6.016731040398502e-05,
      "loss": 2.128,
      "step": 8320
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.2158203125,
      "learning_rate": 5.9935982856291005e-05,
      "loss": 2.0829,
      "step": 8325
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.970502245873573e-05,
      "loss": 2.0905,
      "step": 8330
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.947442981673637e-05,
      "loss": 2.1405,
      "step": 8335
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.22265625,
      "learning_rate": 5.924420553474581e-05,
      "loss": 2.122,
      "step": 8340
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.90143502162515e-05,
      "loss": 2.1289,
      "step": 8345
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.8784864463773824e-05,
      "loss": 2.1152,
      "step": 8350
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.22265625,
      "learning_rate": 5.85557488788643e-05,
      "loss": 2.1293,
      "step": 8355
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.21875,
      "learning_rate": 5.832700406210414e-05,
      "loss": 2.1264,
      "step": 8360
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.21484375,
      "learning_rate": 5.809863061310261e-05,
      "loss": 2.0865,
      "step": 8365
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.7870629130495746e-05,
      "loss": 2.0962,
      "step": 8370
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.21875,
      "learning_rate": 5.76430002119442e-05,
      "loss": 2.123,
      "step": 8375
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.2158203125,
      "learning_rate": 5.741574445413218e-05,
      "loss": 2.105,
      "step": 8380
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.718886245276589e-05,
      "loss": 2.0846,
      "step": 8385
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2197265625,
      "learning_rate": 5.6962354802571574e-05,
      "loss": 2.0829,
      "step": 8390
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2158203125,
      "learning_rate": 5.673622209729426e-05,
      "loss": 2.1085,
      "step": 8395
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.651046492969616e-05,
      "loss": 2.1053,
      "step": 8400
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.21484375,
      "learning_rate": 5.628508389155507e-05,
      "loss": 2.1162,
      "step": 8405
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.224609375,
      "learning_rate": 5.606007957366284e-05,
      "loss": 2.1035,
      "step": 8410
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.583545256582374e-05,
      "loss": 2.1368,
      "step": 8415
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2294921875,
      "learning_rate": 5.5611203456853267e-05,
      "loss": 2.1249,
      "step": 8420
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.224609375,
      "learning_rate": 5.538733283457591e-05,
      "loss": 2.145,
      "step": 8425
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.21875,
      "learning_rate": 5.516384128582444e-05,
      "loss": 2.1136,
      "step": 8430
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.2197265625,
      "learning_rate": 5.4940729396437704e-05,
      "loss": 2.117,
      "step": 8435
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.224609375,
      "learning_rate": 5.471799775125943e-05,
      "loss": 2.1447,
      "step": 8440
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.4495646934136625e-05,
      "loss": 2.1128,
      "step": 8445
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.4273677527917966e-05,
      "loss": 2.1114,
      "step": 8450
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.2197265625,
      "learning_rate": 5.405209011445242e-05,
      "loss": 2.1023,
      "step": 8455
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.383088527458753e-05,
      "loss": 2.1129,
      "step": 8460
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.216796875,
      "learning_rate": 5.361006358816818e-05,
      "loss": 2.1446,
      "step": 8465
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.2099609375,
      "learning_rate": 5.338962563403478e-05,
      "loss": 2.0962,
      "step": 8470
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.2236328125,
      "learning_rate": 5.3169571990021684e-05,
      "loss": 2.0977,
      "step": 8475
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.220703125,
      "learning_rate": 5.294990323295621e-05,
      "loss": 2.0961,
      "step": 8480
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.224609375,
      "learning_rate": 5.273061993865651e-05,
      "loss": 2.1561,
      "step": 8485
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.224609375,
      "learning_rate": 5.251172268193041e-05,
      "loss": 2.15,
      "step": 8490
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.2236328125,
      "learning_rate": 5.229321203657382e-05,
      "loss": 2.1232,
      "step": 8495
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.220703125,
      "learning_rate": 5.20750885753692e-05,
      "loss": 2.1288,
      "step": 8500
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.21875,
      "learning_rate": 5.1857352870084086e-05,
      "loss": 2.1265,
      "step": 8505
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.164000549146954e-05,
      "loss": 2.1376,
      "step": 8510
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.21875,
      "learning_rate": 5.1423047009258904e-05,
      "loss": 2.1105,
      "step": 8515
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.120647799216587e-05,
      "loss": 2.1161,
      "step": 8520
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.0990299007883304e-05,
      "loss": 2.1267,
      "step": 8525
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.077451062308174e-05,
      "loss": 2.104,
      "step": 8530
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.216796875,
      "learning_rate": 5.055911340340771e-05,
      "loss": 2.0874,
      "step": 8535
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.0344107913482516e-05,
      "loss": 2.1245,
      "step": 8540
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.012949471690045e-05,
      "loss": 2.0561,
      "step": 8545
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.9915274376227805e-05,
      "loss": 2.1394,
      "step": 8550
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.21875,
      "learning_rate": 4.970144745300063e-05,
      "loss": 2.1378,
      "step": 8555
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.948801450772409e-05,
      "loss": 2.1072,
      "step": 8560
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.9274976099870415e-05,
      "loss": 2.1504,
      "step": 8565
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.220703125,
      "learning_rate": 4.9062332787877705e-05,
      "loss": 2.0927,
      "step": 8570
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.21484375,
      "learning_rate": 4.885008512914837e-05,
      "loss": 2.0769,
      "step": 8575
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.21484375,
      "learning_rate": 4.863823368004763e-05,
      "loss": 2.0962,
      "step": 8580
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.220703125,
      "learning_rate": 4.842677899590238e-05,
      "loss": 2.0833,
      "step": 8585
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.21484375,
      "learning_rate": 4.8215721630999075e-05,
      "loss": 2.0788,
      "step": 8590
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.800506213858293e-05,
      "loss": 2.0668,
      "step": 8595
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.224609375,
      "learning_rate": 4.779480107085632e-05,
      "loss": 2.1316,
      "step": 8600
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.7584938978976845e-05,
      "loss": 2.0993,
      "step": 8605
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.216796875,
      "learning_rate": 4.737547641305668e-05,
      "loss": 2.083,
      "step": 8610
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.224609375,
      "learning_rate": 4.716641392216048e-05,
      "loss": 2.0819,
      "step": 8615
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.695775205430426e-05,
      "loss": 2.0953,
      "step": 8620
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.21484375,
      "learning_rate": 4.674949135645383e-05,
      "loss": 2.1031,
      "step": 8625
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.654163237452345e-05,
      "loss": 2.0836,
      "step": 8630
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.6334175653374476e-05,
      "loss": 2.0672,
      "step": 8635
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.216796875,
      "learning_rate": 4.612712173681353e-05,
      "loss": 2.0927,
      "step": 8640
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2294921875,
      "learning_rate": 4.592047116759164e-05,
      "loss": 2.0775,
      "step": 8645
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2119140625,
      "learning_rate": 4.571422448740246e-05,
      "loss": 2.1157,
      "step": 8650
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.550838223688074e-05,
      "loss": 2.128,
      "step": 8655
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2236328125,
      "learning_rate": 4.530294495560141e-05,
      "loss": 2.1186,
      "step": 8660
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.216796875,
      "learning_rate": 4.5097913182077656e-05,
      "loss": 2.1303,
      "step": 8665
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2265625,
      "learning_rate": 4.4893287453759755e-05,
      "loss": 2.124,
      "step": 8670
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.4689068307033544e-05,
      "loss": 2.1581,
      "step": 8675
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.4485256277219246e-05,
      "loss": 2.1182,
      "step": 8680
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.21875,
      "learning_rate": 4.428185189856986e-05,
      "loss": 2.0936,
      "step": 8685
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.4078855704269575e-05,
      "loss": 2.0769,
      "step": 8690
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.387626822643294e-05,
      "loss": 2.1441,
      "step": 8695
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.21484375,
      "learning_rate": 4.3674089996102966e-05,
      "loss": 2.1038,
      "step": 8700
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.347232154324992e-05,
      "loss": 2.1426,
      "step": 8705
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.32709633967699e-05,
      "loss": 2.0974,
      "step": 8710
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.220703125,
      "learning_rate": 4.307001608448353e-05,
      "loss": 2.0787,
      "step": 8715
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.212890625,
      "learning_rate": 4.2869480133134435e-05,
      "loss": 2.0808,
      "step": 8720
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.224609375,
      "learning_rate": 4.266935606838796e-05,
      "loss": 2.1249,
      "step": 8725
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.246964441482986e-05,
      "loss": 2.1447,
      "step": 8730
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.2177734375,
      "learning_rate": 4.2270345695964734e-05,
      "loss": 2.1202,
      "step": 8735
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.207146043421477e-05,
      "loss": 2.1282,
      "step": 8740
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.21875,
      "learning_rate": 4.1872989150918375e-05,
      "loss": 2.1001,
      "step": 8745
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.220703125,
      "learning_rate": 4.16749323663288e-05,
      "loss": 2.1221,
      "step": 8750
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.21875,
      "learning_rate": 4.147729059961278e-05,
      "loss": 2.1309,
      "step": 8755
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.216796875,
      "learning_rate": 4.128006436884906e-05,
      "loss": 2.152,
      "step": 8760
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.1083254191027384e-05,
      "loss": 2.1294,
      "step": 8765
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.22265625,
      "learning_rate": 4.088686058204656e-05,
      "loss": 2.1232,
      "step": 8770
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.069088405671375e-05,
      "loss": 2.0924,
      "step": 8775
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.21875,
      "learning_rate": 4.049532512874261e-05,
      "loss": 2.1028,
      "step": 8780
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.2265625,
      "learning_rate": 4.0300184310752265e-05,
      "loss": 2.1369,
      "step": 8785
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.2255859375,
      "learning_rate": 4.0105462114265754e-05,
      "loss": 2.0931,
      "step": 8790
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.991115904970888e-05,
      "loss": 2.1279,
      "step": 8795
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.21875,
      "learning_rate": 3.97172756264087e-05,
      "loss": 2.0991,
      "step": 8800
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.220703125,
      "learning_rate": 3.952381235259228e-05,
      "loss": 2.1481,
      "step": 8805
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.933076973538532e-05,
      "loss": 2.0896,
      "step": 8810
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.9138148280811014e-05,
      "loss": 2.1299,
      "step": 8815
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.220703125,
      "learning_rate": 3.894594849378828e-05,
      "loss": 2.1204,
      "step": 8820
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.212890625,
      "learning_rate": 3.8754170878131e-05,
      "loss": 2.0644,
      "step": 8825
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.21875,
      "learning_rate": 3.856281593654623e-05,
      "loss": 2.1133,
      "step": 8830
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.8371884170633134e-05,
      "loss": 2.0851,
      "step": 8835
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.212890625,
      "learning_rate": 3.818137608088161e-05,
      "loss": 2.1087,
      "step": 8840
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.21875,
      "learning_rate": 3.7991292166670966e-05,
      "loss": 2.1384,
      "step": 8845
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.780163292626859e-05,
      "loss": 2.1075,
      "step": 8850
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.76123988568287e-05,
      "loss": 2.0926,
      "step": 8855
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.742359045439105e-05,
      "loss": 2.1276,
      "step": 8860
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.216796875,
      "learning_rate": 3.723520821387958e-05,
      "loss": 2.1149,
      "step": 8865
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.704725262910094e-05,
      "loss": 2.0676,
      "step": 8870
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2275390625,
      "learning_rate": 3.6859724192743704e-05,
      "loss": 2.1569,
      "step": 8875
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.6672623396376584e-05,
      "loss": 2.0912,
      "step": 8880
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.224609375,
      "learning_rate": 3.648595073044729e-05,
      "loss": 2.0974,
      "step": 8885
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.629970668428129e-05,
      "loss": 2.1083,
      "step": 8890
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.212890625,
      "learning_rate": 3.611389174608068e-05,
      "loss": 2.1057,
      "step": 8895
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.592850640292249e-05,
      "loss": 2.0982,
      "step": 8900
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.574355114075773e-05,
      "loss": 2.0863,
      "step": 8905
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.220703125,
      "learning_rate": 3.555902644441016e-05,
      "loss": 2.1439,
      "step": 8910
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.224609375,
      "learning_rate": 3.5374932797574734e-05,
      "loss": 2.1372,
      "step": 8915
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.220703125,
      "learning_rate": 3.5191270682816604e-05,
      "loss": 2.1334,
      "step": 8920
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.5008040581569634e-05,
      "loss": 2.1059,
      "step": 8925
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.220703125,
      "learning_rate": 3.4825242974135474e-05,
      "loss": 2.1059,
      "step": 8930
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.464287833968176e-05,
      "loss": 2.1372,
      "step": 8935
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.4460947156241376e-05,
      "loss": 2.1008,
      "step": 8940
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2216796875,
      "learning_rate": 3.427944990071108e-05,
      "loss": 2.0944,
      "step": 8945
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.409838704884984e-05,
      "loss": 2.1204,
      "step": 8950
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.391775907527834e-05,
      "loss": 2.1195,
      "step": 8955
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2265625,
      "learning_rate": 3.373756645347703e-05,
      "loss": 2.1051,
      "step": 8960
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.355780965578526e-05,
      "loss": 2.0931,
      "step": 8965
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.21484375,
      "learning_rate": 3.337848915339994e-05,
      "loss": 2.1239,
      "step": 8970
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.31996054163743e-05,
      "loss": 2.0869,
      "step": 8975
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.21484375,
      "learning_rate": 3.302115891361683e-05,
      "loss": 2.082,
      "step": 8980
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.2216796875,
      "learning_rate": 3.2843150112889564e-05,
      "loss": 2.085,
      "step": 8985
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.266557948080757e-05,
      "loss": 2.0749,
      "step": 8990
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.21875,
      "learning_rate": 3.2488447482837146e-05,
      "loss": 2.0904,
      "step": 8995
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.212890625,
      "learning_rate": 3.231175458329465e-05,
      "loss": 2.1296,
      "step": 9000
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.216796875,
      "learning_rate": 3.213550124534579e-05,
      "loss": 2.1353,
      "step": 9005
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.220703125,
      "learning_rate": 3.1959687931003765e-05,
      "loss": 2.1098,
      "step": 9010
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.216796875,
      "learning_rate": 3.178431510112845e-05,
      "loss": 2.095,
      "step": 9015
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.21484375,
      "learning_rate": 3.160938321542506e-05,
      "loss": 2.1216,
      "step": 9020
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.143489273244291e-05,
      "loss": 2.1403,
      "step": 9025
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.216796875,
      "learning_rate": 3.126084410957446e-05,
      "loss": 2.0983,
      "step": 9030
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.1087237803053584e-05,
      "loss": 2.0944,
      "step": 9035
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.091407426795503e-05,
      "loss": 2.1514,
      "step": 9040
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.22265625,
      "learning_rate": 3.0741353958192755e-05,
      "loss": 2.0635,
      "step": 9045
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.22265625,
      "learning_rate": 3.0569077326518904e-05,
      "loss": 2.104,
      "step": 9050
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.0397244824522618e-05,
      "loss": 2.0914,
      "step": 9055
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.22265625,
      "learning_rate": 3.0225856902628847e-05,
      "loss": 2.1016,
      "step": 9060
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.0054914010097145e-05,
      "loss": 2.0724,
      "step": 9065
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.9884416595020505e-05,
      "loss": 2.1027,
      "step": 9070
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.216796875,
      "learning_rate": 2.971436510432424e-05,
      "loss": 2.1249,
      "step": 9075
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.9544759983764736e-05,
      "loss": 2.1062,
      "step": 9080
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.9375601677928254e-05,
      "loss": 2.0937,
      "step": 9085
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.9206890630229876e-05,
      "loss": 2.1203,
      "step": 9090
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.9038627282912268e-05,
      "loss": 2.1202,
      "step": 9095
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.212890625,
      "learning_rate": 2.887081207704454e-05,
      "loss": 2.0738,
      "step": 9100
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.220703125,
      "learning_rate": 2.870344545252106e-05,
      "loss": 2.1016,
      "step": 9105
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.8536527848060446e-05,
      "loss": 2.0854,
      "step": 9110
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.21484375,
      "learning_rate": 2.8370059701204122e-05,
      "loss": 2.1124,
      "step": 9115
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.220703125,
      "learning_rate": 2.820404144831541e-05,
      "loss": 2.0718,
      "step": 9120
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.22265625,
      "learning_rate": 2.8038473524578447e-05,
      "loss": 2.0919,
      "step": 9125
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.787335636399675e-05,
      "loss": 2.1262,
      "step": 9130
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.21484375,
      "learning_rate": 2.7708690399392366e-05,
      "loss": 2.1434,
      "step": 9135
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.216796875,
      "learning_rate": 2.7544476062404557e-05,
      "loss": 2.1098,
      "step": 9140
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.738071378348872e-05,
      "loss": 2.1332,
      "step": 9145
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.216796875,
      "learning_rate": 2.7217403991915368e-05,
      "loss": 2.1317,
      "step": 9150
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.7054547115768735e-05,
      "loss": 2.0972,
      "step": 9155
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2265625,
      "learning_rate": 2.6892143581946116e-05,
      "loss": 2.0682,
      "step": 9160
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.220703125,
      "learning_rate": 2.673019381615609e-05,
      "loss": 2.1369,
      "step": 9165
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.6568698242918055e-05,
      "loss": 2.068,
      "step": 9170
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.640765728556074e-05,
      "loss": 2.134,
      "step": 9175
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.6247071366221175e-05,
      "loss": 2.1242,
      "step": 9180
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.224609375,
      "learning_rate": 2.6086940905843606e-05,
      "loss": 2.1113,
      "step": 9185
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.22265625,
      "learning_rate": 2.5927266324178345e-05,
      "loss": 2.151,
      "step": 9190
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2265625,
      "learning_rate": 2.5768048039780858e-05,
      "loss": 2.1334,
      "step": 9195
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.5609286470010262e-05,
      "loss": 2.0881,
      "step": 9200
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.545098203102876e-05,
      "loss": 2.0929,
      "step": 9205
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.21484375,
      "learning_rate": 2.529313513780016e-05,
      "loss": 2.0808,
      "step": 9210
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.513574620408874e-05,
      "loss": 2.1191,
      "step": 9215
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.4978815642458654e-05,
      "loss": 2.0935,
      "step": 9220
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.212890625,
      "learning_rate": 2.482234386427227e-05,
      "loss": 2.106,
      "step": 9225
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.212890625,
      "learning_rate": 2.4666331279689425e-05,
      "loss": 2.1415,
      "step": 9230
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.4510778297666282e-05,
      "loss": 2.1146,
      "step": 9235
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.435568532595427e-05,
      "loss": 2.1097,
      "step": 9240
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.4201052771099008e-05,
      "loss": 2.1454,
      "step": 9245
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.220703125,
      "learning_rate": 2.404688103843902e-05,
      "loss": 2.1059,
      "step": 9250
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.389317053210518e-05,
      "loss": 2.116,
      "step": 9255
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.3739921655019147e-05,
      "loss": 2.0831,
      "step": 9260
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.358713480889254e-05,
      "loss": 2.1126,
      "step": 9265
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.3434810394225927e-05,
      "loss": 2.08,
      "step": 9270
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.21484375,
      "learning_rate": 2.3282948810307637e-05,
      "loss": 2.0732,
      "step": 9275
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2265625,
      "learning_rate": 2.313155045521278e-05,
      "loss": 2.1507,
      "step": 9280
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.2980615725802213e-05,
      "loss": 2.1155,
      "step": 9285
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.216796875,
      "learning_rate": 2.283014501772154e-05,
      "loss": 2.152,
      "step": 9290
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.2138671875,
      "learning_rate": 2.268013872539998e-05,
      "loss": 2.1221,
      "step": 9295
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.21875,
      "learning_rate": 2.2530597242049378e-05,
      "loss": 2.1292,
      "step": 9300
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.224609375,
      "learning_rate": 2.238152095966315e-05,
      "loss": 2.0971,
      "step": 9305
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.2265625,
      "learning_rate": 2.223291026901533e-05,
      "loss": 2.1177,
      "step": 9310
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.220703125,
      "learning_rate": 2.208476555965946e-05,
      "loss": 2.088,
      "step": 9315
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.1937087219927576e-05,
      "loss": 2.1734,
      "step": 9320
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.178987563692938e-05,
      "loss": 2.1248,
      "step": 9325
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.22265625,
      "learning_rate": 2.1643131196550835e-05,
      "loss": 2.1097,
      "step": 9330
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.236328125,
      "learning_rate": 2.1496854283453472e-05,
      "loss": 2.0857,
      "step": 9335
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.212890625,
      "learning_rate": 2.1351045281073412e-05,
      "loss": 2.0775,
      "step": 9340
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.22265625,
      "learning_rate": 2.1205704571620076e-05,
      "loss": 2.1096,
      "step": 9345
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.1060832536075403e-05,
      "loss": 2.1098,
      "step": 9350
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.216796875,
      "learning_rate": 2.0916429554192818e-05,
      "loss": 2.0878,
      "step": 9355
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.216796875,
      "learning_rate": 2.07724960044962e-05,
      "loss": 2.138,
      "step": 9360
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.220703125,
      "learning_rate": 2.0629032264278904e-05,
      "loss": 2.1298,
      "step": 9365
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.21484375,
      "learning_rate": 2.0486038709602706e-05,
      "loss": 2.0918,
      "step": 9370
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.220703125,
      "learning_rate": 2.034351571529709e-05,
      "loss": 2.1409,
      "step": 9375
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.0201463654957766e-05,
      "loss": 2.1113,
      "step": 9380
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.22265625,
      "learning_rate": 2.0059882900946227e-05,
      "loss": 2.1025,
      "step": 9385
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.21875,
      "learning_rate": 1.9918773824388405e-05,
      "loss": 2.0689,
      "step": 9390
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.224609375,
      "learning_rate": 1.977813679517386e-05,
      "loss": 2.1106,
      "step": 9395
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.21875,
      "learning_rate": 1.96379721819548e-05,
      "loss": 2.1136,
      "step": 9400
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.9498280352145004e-05,
      "loss": 2.143,
      "step": 9405
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2275390625,
      "learning_rate": 1.9359061671919032e-05,
      "loss": 2.1268,
      "step": 9410
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.21875,
      "learning_rate": 1.9220316506211077e-05,
      "loss": 2.0934,
      "step": 9415
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.21484375,
      "learning_rate": 1.9082045218714262e-05,
      "loss": 2.1128,
      "step": 9420
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.8944248171879453e-05,
      "loss": 2.1024,
      "step": 9425
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.21875,
      "learning_rate": 1.8806925726914225e-05,
      "loss": 2.122,
      "step": 9430
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.8670078243782353e-05,
      "loss": 2.0871,
      "step": 9435
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.21875,
      "learning_rate": 1.853370608120244e-05,
      "loss": 2.0643,
      "step": 9440
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.839780959664714e-05,
      "loss": 2.1121,
      "step": 9445
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.21875,
      "learning_rate": 1.8262389146342217e-05,
      "loss": 2.0914,
      "step": 9450
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.220703125,
      "learning_rate": 1.8127445085265716e-05,
      "loss": 2.1003,
      "step": 9455
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.79929777671467e-05,
      "loss": 2.0827,
      "step": 9460
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.216796875,
      "learning_rate": 1.785898754446469e-05,
      "loss": 2.1114,
      "step": 9465
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.7725474768448636e-05,
      "loss": 2.0928,
      "step": 9470
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.759243978907583e-05,
      "loss": 2.0907,
      "step": 9475
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.216796875,
      "learning_rate": 1.7459882955071237e-05,
      "loss": 2.1334,
      "step": 9480
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.22265625,
      "learning_rate": 1.732780461390635e-05,
      "loss": 2.096,
      "step": 9485
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.21484375,
      "learning_rate": 1.7196205111798446e-05,
      "loss": 2.1169,
      "step": 9490
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.216796875,
      "learning_rate": 1.7065084793709607e-05,
      "loss": 2.1197,
      "step": 9495
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.693444400334583e-05,
      "loss": 2.0802,
      "step": 9500
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.23046875,
      "learning_rate": 1.680428308315618e-05,
      "loss": 2.079,
      "step": 9505
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.216796875,
      "learning_rate": 1.6674602374331693e-05,
      "loss": 2.0876,
      "step": 9510
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2265625,
      "learning_rate": 1.6545402216804783e-05,
      "loss": 2.1172,
      "step": 9515
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.6416682949248142e-05,
      "loss": 2.0708,
      "step": 9520
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.628844490907384e-05,
      "loss": 2.1358,
      "step": 9525
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.616068843243257e-05,
      "loss": 2.1241,
      "step": 9530
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.6033413854212643e-05,
      "loss": 2.1311,
      "step": 9535
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2119140625,
      "learning_rate": 1.5906621508039342e-05,
      "loss": 2.0832,
      "step": 9540
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.5780311726273634e-05,
      "loss": 2.091,
      "step": 9545
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.21875,
      "learning_rate": 1.5654484840011617e-05,
      "loss": 2.0968,
      "step": 9550
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.220703125,
      "learning_rate": 1.552914117908375e-05,
      "loss": 2.1037,
      "step": 9555
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.216796875,
      "learning_rate": 1.5404281072053517e-05,
      "loss": 2.1281,
      "step": 9560
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.216796875,
      "learning_rate": 1.5279904846217085e-05,
      "loss": 2.1145,
      "step": 9565
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.220703125,
      "learning_rate": 1.515601282760215e-05,
      "loss": 2.1418,
      "step": 9570
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.5032605340967132e-05,
      "loss": 2.0932,
      "step": 9575
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.220703125,
      "learning_rate": 1.4909682709800355e-05,
      "loss": 2.1146,
      "step": 9580
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.4787245256319227e-05,
      "loss": 2.104,
      "step": 9585
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.4665293301469374e-05,
      "loss": 2.0964,
      "step": 9590
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.4543827164923619e-05,
      "loss": 2.1167,
      "step": 9595
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.4422847165081555e-05,
      "loss": 2.0994,
      "step": 9600
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.220703125,
      "learning_rate": 1.4302353619068309e-05,
      "loss": 2.0851,
      "step": 9605
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.216796875,
      "learning_rate": 1.4182346842733873e-05,
      "loss": 2.1048,
      "step": 9610
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.4062827150652302e-05,
      "loss": 2.0988,
      "step": 9615
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.394379485612085e-05,
      "loss": 2.1497,
      "step": 9620
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.224609375,
      "learning_rate": 1.3825250271159173e-05,
      "loss": 2.1561,
      "step": 9625
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2265625,
      "learning_rate": 1.3707193706508391e-05,
      "loss": 2.0875,
      "step": 9630
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.3589625471630562e-05,
      "loss": 2.0693,
      "step": 9635
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.3472545874707565e-05,
      "loss": 2.1013,
      "step": 9640
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.3355955222640326e-05,
      "loss": 2.1421,
      "step": 9645
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.216796875,
      "learning_rate": 1.3239853821048287e-05,
      "loss": 2.0988,
      "step": 9650
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.3124241974268291e-05,
      "loss": 2.1146,
      "step": 9655
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.3009119985353969e-05,
      "loss": 2.1093,
      "step": 9660
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.2894488156074813e-05,
      "loss": 2.1261,
      "step": 9665
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.21484375,
      "learning_rate": 1.2780346786915598e-05,
      "loss": 2.1356,
      "step": 9670
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.2666696177075276e-05,
      "loss": 2.1216,
      "step": 9675
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.224609375,
      "learning_rate": 1.2553536624466456e-05,
      "loss": 2.137,
      "step": 9680
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.224609375,
      "learning_rate": 1.2440868425714613e-05,
      "loss": 2.1244,
      "step": 9685
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.216796875,
      "learning_rate": 1.2328691876157128e-05,
      "loss": 2.1037,
      "step": 9690
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.2217007269842651e-05,
      "loss": 2.0922,
      "step": 9695
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.2105814899530288e-05,
      "loss": 2.1027,
      "step": 9700
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.2294921875,
      "learning_rate": 1.1995115056688889e-05,
      "loss": 2.1215,
      "step": 9705
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.220703125,
      "learning_rate": 1.18849080314962e-05,
      "loss": 2.1065,
      "step": 9710
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.224609375,
      "learning_rate": 1.177519411283814e-05,
      "loss": 2.112,
      "step": 9715
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.1665973588308122e-05,
      "loss": 2.1123,
      "step": 9720
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.22265625,
      "learning_rate": 1.1557246744206084e-05,
      "loss": 2.1194,
      "step": 9725
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2333984375,
      "learning_rate": 1.1449013865538027e-05,
      "loss": 2.0927,
      "step": 9730
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.1341275236015003e-05,
      "loss": 2.1198,
      "step": 9735
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.1234031138052592e-05,
      "loss": 2.0599,
      "step": 9740
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.224609375,
      "learning_rate": 1.1127281852769944e-05,
      "loss": 2.1762,
      "step": 9745
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2255859375,
      "learning_rate": 1.1021027659989225e-05,
      "loss": 2.106,
      "step": 9750
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.0915268838234838e-05,
      "loss": 2.1149,
      "step": 9755
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.22265625,
      "learning_rate": 1.0810005664732558e-05,
      "loss": 2.1219,
      "step": 9760
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.0705238415409068e-05,
      "loss": 2.1004,
      "step": 9765
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.224609375,
      "learning_rate": 1.0600967364891001e-05,
      "loss": 2.0973,
      "step": 9770
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.21875,
      "learning_rate": 1.0497192786504228e-05,
      "loss": 2.1278,
      "step": 9775
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.0393914952273398e-05,
      "loss": 2.1386,
      "step": 9780
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.0291134132920866e-05,
      "loss": 2.1055,
      "step": 9785
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.0188850597866273e-05,
      "loss": 2.1044,
      "step": 9790
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2255859375,
      "learning_rate": 1.0087064615225683e-05,
      "loss": 2.1127,
      "step": 9795
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.985776451810936e-06,
      "loss": 2.1143,
      "step": 9800
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2138671875,
      "learning_rate": 9.884986373128934e-06,
      "loss": 2.1134,
      "step": 9805
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2197265625,
      "learning_rate": 9.78469464338092e-06,
      "loss": 2.1277,
      "step": 9810
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2197265625,
      "learning_rate": 9.684901525461865e-06,
      "loss": 2.1005,
      "step": 9815
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.58560728095974e-06,
      "loss": 2.1352,
      "step": 9820
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.224609375,
      "learning_rate": 9.486812170154724e-06,
      "loss": 2.1596,
      "step": 9825
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.21875,
      "learning_rate": 9.388516452018702e-06,
      "loss": 2.1258,
      "step": 9830
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.290720384214479e-06,
      "loss": 2.159,
      "step": 9835
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.22265625,
      "learning_rate": 9.193424223095103e-06,
      "loss": 2.093,
      "step": 9840
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.096628223703207e-06,
      "loss": 2.0915,
      "step": 9845
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.212890625,
      "learning_rate": 9.00033263977047e-06,
      "loss": 2.1096,
      "step": 9850
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.904537723716621e-06,
      "loss": 2.0945,
      "step": 9855
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.2216796875,
      "learning_rate": 8.809243726649107e-06,
      "loss": 2.111,
      "step": 9860
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.216796875,
      "learning_rate": 8.71445089836238e-06,
      "loss": 2.1174,
      "step": 9865
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.21875,
      "learning_rate": 8.620159487337076e-06,
      "loss": 2.1279,
      "step": 9870
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.526369740739481e-06,
      "loss": 2.1237,
      "step": 9875
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.2314453125,
      "learning_rate": 8.43308190442087e-06,
      "loss": 2.1158,
      "step": 9880
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.228515625,
      "learning_rate": 8.340296222916921e-06,
      "loss": 2.1349,
      "step": 9885
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.2216796875,
      "learning_rate": 8.24801293944688e-06,
      "loss": 2.0741,
      "step": 9890
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2138671875,
      "learning_rate": 8.15623229591318e-06,
      "loss": 2.1205,
      "step": 9895
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.21875,
      "learning_rate": 8.064954532900659e-06,
      "loss": 2.1016,
      "step": 9900
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.97417988967588e-06,
      "loss": 2.1264,
      "step": 9905
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.883908604186685e-06,
      "loss": 2.0931,
      "step": 9910
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.794140913061366e-06,
      "loss": 2.125,
      "step": 9915
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.21875,
      "learning_rate": 7.704877051608206e-06,
      "loss": 2.0809,
      "step": 9920
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2265625,
      "learning_rate": 7.61611725381477e-06,
      "loss": 2.1238,
      "step": 9925
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.21484375,
      "learning_rate": 7.5278617523472985e-06,
      "loss": 2.0693,
      "step": 9930
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2255859375,
      "learning_rate": 7.440110778550224e-06,
      "loss": 2.1027,
      "step": 9935
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.2255859375,
      "learning_rate": 7.352864562445283e-06,
      "loss": 2.1164,
      "step": 9940
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.216796875,
      "learning_rate": 7.266123332731267e-06,
      "loss": 2.1191,
      "step": 9945
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.17988731678314e-06,
      "loss": 2.1598,
      "step": 9950
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.094156740651525e-06,
      "loss": 2.0958,
      "step": 9955
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.220703125,
      "learning_rate": 7.0089318290622375e-06,
      "loss": 2.1188,
      "step": 9960
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2216796875,
      "learning_rate": 6.924212805415553e-06,
      "loss": 2.0864,
      "step": 9965
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2255859375,
      "learning_rate": 6.839999891785609e-06,
      "loss": 2.1835,
      "step": 9970
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.22265625,
      "learning_rate": 6.756293308919892e-06,
      "loss": 2.091,
      "step": 9975
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2177734375,
      "learning_rate": 6.673093276238751e-06,
      "loss": 2.1132,
      "step": 9980
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.224609375,
      "learning_rate": 6.5904000118345745e-06,
      "loss": 2.1281,
      "step": 9985
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.21875,
      "learning_rate": 6.508213732471391e-06,
      "loss": 2.1066,
      "step": 9990
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2236328125,
      "learning_rate": 6.426534653584337e-06,
      "loss": 2.1207,
      "step": 9995
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2236328125,
      "learning_rate": 6.345362989278947e-06,
      "loss": 2.0985,
      "step": 10000
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2236328125,
      "learning_rate": 6.264698952330705e-06,
      "loss": 2.1387,
      "step": 10005
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.21875,
      "learning_rate": 6.18454275418443e-06,
      "loss": 2.1239,
      "step": 10010
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.22265625,
      "learning_rate": 6.104894604953759e-06,
      "loss": 2.0651,
      "step": 10015
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2255859375,
      "learning_rate": 6.0257547134205725e-06,
      "loss": 2.1373,
      "step": 10020
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.216796875,
      "learning_rate": 5.947123287034439e-06,
      "loss": 2.098,
      "step": 10025
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.86900053191215e-06,
      "loss": 2.0951,
      "step": 10030
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.791386652837027e-06,
      "loss": 2.1572,
      "step": 10035
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2294921875,
      "learning_rate": 5.7142818532585515e-06,
      "loss": 2.1361,
      "step": 10040
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.220703125,
      "learning_rate": 5.63768633529167e-06,
      "loss": 2.1041,
      "step": 10045
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.2119140625,
      "learning_rate": 5.5616002997164185e-06,
      "loss": 2.1299,
      "step": 10050
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2294921875,
      "learning_rate": 5.486023945977304e-06,
      "loss": 2.1305,
      "step": 10055
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.4109574721827646e-06,
      "loss": 2.0973,
      "step": 10060
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.336401075104825e-06,
      "loss": 2.0838,
      "step": 10065
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.21875,
      "learning_rate": 5.262354950178217e-06,
      "loss": 2.1095,
      "step": 10070
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.188819291500302e-06,
      "loss": 2.1235,
      "step": 10075
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.220703125,
      "learning_rate": 5.115794291830245e-06,
      "loss": 2.1142,
      "step": 10080
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.21484375,
      "learning_rate": 5.04328014258868e-06,
      "loss": 2.09,
      "step": 10085
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.22265625,
      "learning_rate": 4.971277033857092e-06,
      "loss": 2.1228,
      "step": 10090
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.89978515437739e-06,
      "loss": 2.0957,
      "step": 10095
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.828804691551448e-06,
      "loss": 2.1217,
      "step": 10100
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.758335831440497e-06,
      "loss": 2.133,
      "step": 10105
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.216796875,
      "learning_rate": 4.688378758764689e-06,
      "loss": 2.1422,
      "step": 10110
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.2177734375,
      "learning_rate": 4.618933656902758e-06,
      "loss": 2.1195,
      "step": 10115
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.22265625,
      "learning_rate": 4.5500007078911996e-06,
      "loss": 2.0933,
      "step": 10120
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.2138671875,
      "learning_rate": 4.481580092424187e-06,
      "loss": 2.1086,
      "step": 10125
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.21484375,
      "learning_rate": 4.41367198985283e-06,
      "loss": 2.09,
      "step": 10130
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.2216796875,
      "learning_rate": 4.3462765781848045e-06,
      "loss": 2.107,
      "step": 10135
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.279394034083839e-06,
      "loss": 2.0996,
      "step": 10140
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.213024532869314e-06,
      "loss": 2.1024,
      "step": 10145
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.21875,
      "learning_rate": 4.147168248515798e-06,
      "loss": 2.1083,
      "step": 10150
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.22265625,
      "learning_rate": 4.081825353652424e-06,
      "loss": 2.1142,
      "step": 10155
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.22265625,
      "learning_rate": 4.01699601956278e-06,
      "loss": 2.1095,
      "step": 10160
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2373046875,
      "learning_rate": 3.952680416184151e-06,
      "loss": 2.097,
      "step": 10165
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2275390625,
      "learning_rate": 3.88887871210708e-06,
      "loss": 2.0742,
      "step": 10170
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.220703125,
      "learning_rate": 3.825591074575208e-06,
      "loss": 2.098,
      "step": 10175
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.762817669484564e-06,
      "loss": 2.1232,
      "step": 10180
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2294921875,
      "learning_rate": 3.700558661383191e-06,
      "loss": 2.1437,
      "step": 10185
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.638814213470787e-06,
      "loss": 2.0822,
      "step": 10190
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2333984375,
      "learning_rate": 3.577584487598218e-06,
      "loss": 2.1089,
      "step": 10195
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.51686964426714e-06,
      "loss": 2.1156,
      "step": 10200
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.22265625,
      "learning_rate": 3.4566698426294674e-06,
      "loss": 2.1033,
      "step": 10205
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.396985240487105e-06,
      "loss": 2.0581,
      "step": 10210
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.22265625,
      "learning_rate": 3.3378159942914376e-06,
      "loss": 2.1041,
      "step": 10215
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.21484375,
      "learning_rate": 3.2791622591429536e-06,
      "loss": 2.089,
      "step": 10220
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.220703125,
      "learning_rate": 3.2210241887908444e-06,
      "loss": 2.1456,
      "step": 10225
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.22265625,
      "learning_rate": 3.163401935632537e-06,
      "loss": 2.1347,
      "step": 10230
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.1062956507133867e-06,
      "loss": 2.0823,
      "step": 10235
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.2236328125,
      "learning_rate": 3.0497054837262506e-06,
      "loss": 2.0742,
      "step": 10240
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.9936315830110473e-06,
      "loss": 2.1019,
      "step": 10245
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.21875,
      "learning_rate": 2.938074095554444e-06,
      "loss": 2.0905,
      "step": 10250
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.21875,
      "learning_rate": 2.883033166989413e-06,
      "loss": 2.0628,
      "step": 10255
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.828508941594854e-06,
      "loss": 2.1302,
      "step": 10260
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.774501562295262e-06,
      "loss": 2.0996,
      "step": 10265
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.7210111706603036e-06,
      "loss": 2.113,
      "step": 10270
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.220703125,
      "learning_rate": 2.6680379069044416e-06,
      "loss": 2.1421,
      "step": 10275
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.22265625,
      "learning_rate": 2.6155819098866664e-06,
      "loss": 2.1139,
      "step": 10280
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.22265625,
      "learning_rate": 2.563643317109965e-06,
      "loss": 2.1383,
      "step": 10285
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.22265625,
      "learning_rate": 2.5122222647211424e-06,
      "loss": 2.1012,
      "step": 10290
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.224609375,
      "learning_rate": 2.4613188875102667e-06,
      "loss": 2.1486,
      "step": 10295
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.410933318910513e-06,
      "loss": 2.1413,
      "step": 10300
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.22265625,
      "learning_rate": 2.3610656909976993e-06,
      "loss": 2.1081,
      "step": 10305
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.224609375,
      "learning_rate": 2.3117161344899274e-06,
      "loss": 2.1173,
      "step": 10310
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.220703125,
      "learning_rate": 2.2628847787473427e-06,
      "loss": 2.1125,
      "step": 10315
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.216796875,
      "learning_rate": 2.2145717517716437e-06,
      "loss": 2.1233,
      "step": 10320
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.1667771802059255e-06,
      "loss": 2.1439,
      "step": 10325
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.1195011893341945e-06,
      "loss": 2.1062,
      "step": 10330
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2158203125,
      "learning_rate": 2.07274390308112e-06,
      "loss": 2.1253,
      "step": 10335
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.22265625,
      "learning_rate": 2.026505444011684e-06,
      "loss": 2.1364,
      "step": 10340
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.21875,
      "learning_rate": 1.9807859333308865e-06,
      "loss": 2.1224,
      "step": 10345
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2255859375,
      "learning_rate": 1.9355854908833514e-06,
      "loss": 2.1277,
      "step": 10350
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.220703125,
      "learning_rate": 1.8909042351531459e-06,
      "loss": 2.1251,
      "step": 10355
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.22265625,
      "learning_rate": 1.8467422832633142e-06,
      "loss": 2.1136,
      "step": 10360
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2265625,
      "learning_rate": 1.8030997509757007e-06,
      "loss": 2.1412,
      "step": 10365
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.21484375,
      "learning_rate": 1.7599767526905953e-06,
      "loss": 2.04,
      "step": 10370
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.717373401446376e-06,
      "loss": 2.0922,
      "step": 10375
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.212890625,
      "learning_rate": 1.675289808919378e-06,
      "loss": 2.0762,
      "step": 10380
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.633726085423337e-06,
      "loss": 2.138,
      "step": 10385
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.5926823399093905e-06,
      "loss": 2.1341,
      "step": 10390
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.220703125,
      "learning_rate": 1.5521586799655874e-06,
      "loss": 2.0983,
      "step": 10395
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.224609375,
      "learning_rate": 1.5121552118167125e-06,
      "loss": 2.1025,
      "step": 10400
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.472672040323908e-06,
      "loss": 2.099,
      "step": 10405
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.4337092689845844e-06,
      "loss": 2.1013,
      "step": 10410
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.3952669999318657e-06,
      "loss": 2.1162,
      "step": 10415
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.3573453339345898e-06,
      "loss": 2.1213,
      "step": 10420
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.3199443703969083e-06,
      "loss": 2.1008,
      "step": 10425
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.2830642073580645e-06,
      "loss": 2.1289,
      "step": 10430
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.2294921875,
      "learning_rate": 1.2467049414921273e-06,
      "loss": 2.1208,
      "step": 10435
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.2108666681076796e-06,
      "loss": 2.1018,
      "step": 10440
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.228515625,
      "learning_rate": 1.175549481147753e-06,
      "loss": 2.1416,
      "step": 10445
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.220703125,
      "learning_rate": 1.1407534731892933e-06,
      "loss": 2.1281,
      "step": 10450
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.106478735443184e-06,
      "loss": 2.0847,
      "step": 10455
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.07272535775389e-06,
      "loss": 2.078,
      "step": 10460
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.216796875,
      "learning_rate": 1.039493428599192e-06,
      "loss": 2.1367,
      "step": 10465
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.21875,
      "learning_rate": 1.0067830350900532e-06,
      "loss": 2.12,
      "step": 10470
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.220703125,
      "learning_rate": 9.745942629703075e-07,
      "loss": 2.0917,
      "step": 10475
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.2197265625,
      "learning_rate": 9.429271966164388e-07,
      "loss": 2.1365,
      "step": 10480
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.117819190374022e-07,
      "loss": 2.1,
      "step": 10485
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2236328125,
      "learning_rate": 8.811585118744026e-07,
      "loss": 2.0986,
      "step": 10490
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2265625,
      "learning_rate": 8.510570554006502e-07,
      "loss": 2.1679,
      "step": 10495
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.212890625,
      "learning_rate": 8.21477628521139e-07,
      "loss": 2.1362,
      "step": 10500
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.220703125,
      "learning_rate": 7.924203087725124e-07,
      "loss": 2.0937,
      "step": 10505
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.21875,
      "learning_rate": 7.63885172322798e-07,
      "loss": 2.1098,
      "step": 10510
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.216796875,
      "learning_rate": 7.35872293971207e-07,
      "loss": 2.1114,
      "step": 10515
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.083817471479349e-07,
      "loss": 2.1369,
      "step": 10520
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.212890625,
      "learning_rate": 6.814136039140717e-07,
      "loss": 2.0751,
      "step": 10525
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2197265625,
      "learning_rate": 6.54967934961248e-07,
      "loss": 2.1256,
      "step": 10530
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2294921875,
      "learning_rate": 6.290448096115453e-07,
      "loss": 2.0886,
      "step": 10535
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.216796875,
      "learning_rate": 6.036442958173183e-07,
      "loss": 2.1488,
      "step": 10540
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.22265625,
      "learning_rate": 5.787664601609954e-07,
      "loss": 2.1258,
      "step": 10545
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.216796875,
      "learning_rate": 5.544113678549235e-07,
      "loss": 2.0979,
      "step": 10550
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.30579082741145e-07,
      "loss": 2.0995,
      "step": 10555
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.220703125,
      "learning_rate": 5.072696672913102e-07,
      "loss": 2.1074,
      "step": 10560
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.8448318260641e-07,
      "loss": 2.0968,
      "step": 10565
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.22265625,
      "learning_rate": 4.622196884167318e-07,
      "loss": 2.1238,
      "step": 10570
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.21875,
      "learning_rate": 4.4047924308161516e-07,
      "loss": 2.0813,
      "step": 10575
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.2236328125,
      "learning_rate": 4.1926190358934083e-07,
      "loss": 2.1372,
      "step": 10580
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.216796875,
      "learning_rate": 3.985677255569753e-07,
      "loss": 2.0857,
      "step": 10585
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.2177734375,
      "learning_rate": 3.7839676323023765e-07,
      "loss": 2.1184,
      "step": 10590
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2197265625,
      "learning_rate": 3.5874906948327737e-07,
      "loss": 2.1057,
      "step": 10595
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2138671875,
      "learning_rate": 3.396246958186744e-07,
      "loss": 2.1105,
      "step": 10600
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.21875,
      "learning_rate": 3.210236923671728e-07,
      "loss": 2.1313,
      "step": 10605
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2255859375,
      "learning_rate": 3.029461078876361e-07,
      "loss": 2.158,
      "step": 10610
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2294921875,
      "learning_rate": 2.8539198976686997e-07,
      "loss": 2.1015,
      "step": 10615
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.21484375,
      "learning_rate": 2.6836138401955534e-07,
      "loss": 2.1259,
      "step": 10620
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.22265625,
      "learning_rate": 2.518543352880265e-07,
      "loss": 2.1453,
      "step": 10625
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.35870886842271e-07,
      "loss": 2.1135,
      "step": 10630
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.204110805797077e-07,
      "loss": 2.0939,
      "step": 10635
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.0547495702518682e-07,
      "loss": 2.1107,
      "step": 10640
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.2265625,
      "learning_rate": 1.9106255533083428e-07,
      "loss": 2.0952,
      "step": 10645
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.21875,
      "learning_rate": 1.7717391327585208e-07,
      "loss": 2.1414,
      "step": 10650
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.220703125,
      "learning_rate": 1.6380906726660705e-07,
      "loss": 2.1247,
      "step": 10655
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.5096805233638654e-07,
      "loss": 2.1159,
      "step": 10660
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.3865090214539856e-07,
      "loss": 2.0838,
      "step": 10665
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.220703125,
      "learning_rate": 1.2685764898059393e-07,
      "loss": 2.0818,
      "step": 10670
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.21875,
      "learning_rate": 1.1558832375566653e-07,
      "loss": 2.0904,
      "step": 10675
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.2197265625,
      "learning_rate": 1.0484295601089767e-07,
      "loss": 2.137,
      "step": 10680
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.2294921875,
      "learning_rate": 9.462157391317838e-08,
      "loss": 2.1282,
      "step": 10685
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.2255859375,
      "learning_rate": 8.492420425583181e-08,
      "loss": 2.1308,
      "step": 10690
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.21484375,
      "learning_rate": 7.575087245861313e-08,
      "loss": 2.0737,
      "step": 10695
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.21484375,
      "learning_rate": 6.710160256755416e-08,
      "loss": 2.102,
      "step": 10700
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2236328125,
      "learning_rate": 5.897641725505221e-08,
      "loss": 2.0975,
      "step": 10705
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2177734375,
      "learning_rate": 5.137533781964798e-08,
      "loss": 2.0821,
      "step": 10710
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.220703125,
      "learning_rate": 4.429838418607002e-08,
      "loss": 2.1072,
      "step": 10715
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2158203125,
      "learning_rate": 3.774557490516806e-08,
      "loss": 2.1058,
      "step": 10720
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.224609375,
      "learning_rate": 3.171692715382424e-08,
      "loss": 2.1674,
      "step": 10725
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2197265625,
      "learning_rate": 2.6212456734953094e-08,
      "loss": 2.1061,
      "step": 10730
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.123217807743494e-08,
      "loss": 2.1099,
      "step": 10735
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.6776104236071455e-08,
      "loss": 2.1166,
      "step": 10740
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.220703125,
      "learning_rate": 1.2844246891607904e-08,
      "loss": 2.1206,
      "step": 10745
    },
    {
      "epoch": 1.99,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.436616350622096e-09,
      "loss": 2.1129,
      "step": 10750
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.220703125,
      "learning_rate": 6.5532215455244015e-09,
      "loss": 2.0892,
      "step": 10755
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.228515625,
      "learning_rate": 4.194070034579944e-09,
      "loss": 2.139,
      "step": 10760
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.3591680018419935e-09,
      "loss": 2.1614,
      "step": 10765
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2177734375,
      "learning_rate": 1.0485202571297593e-09,
      "loss": 2.118,
      "step": 10770
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.2177734375,
      "learning_rate": 2.621302360727995e-10,
      "loss": 2.111,
      "step": 10775
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0,
      "loss": 2.0952,
      "step": 10780
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.1485531330108643,
      "eval_runtime": 171.3478,
      "eval_samples_per_second": 28.299,
      "eval_steps_per_second": 3.543,
      "step": 10780
    },
    {
      "epoch": 2.0,
      "step": 10780,
      "total_flos": 6.122136059783414e+17,
      "train_loss": 2.151964877841645,
      "train_runtime": 21635.5911,
      "train_samples_per_second": 7.971,
      "train_steps_per_second": 0.498
    }
  ],
  "logging_steps": 5,
  "max_steps": 10780,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "total_flos": 6.122136059783414e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}