{
  "best_metric": 0.10886295884847641,
  "best_model_checkpoint": "./vit-base-brain-tumor-detection/checkpoint-3700",
  "epoch": 20.0,
  "eval_steps": 100,
  "global_step": 6400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03125,
      "grad_norm": 1.5038443803787231,
      "learning_rate": 0.0001996875,
      "loss": 1.1434,
      "step": 10
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.8518216013908386,
      "learning_rate": 0.000199375,
      "loss": 0.9712,
      "step": 20
    },
    {
      "epoch": 0.09375,
      "grad_norm": 2.0169894695281982,
      "learning_rate": 0.00019906250000000002,
      "loss": 0.9002,
      "step": 30
    },
    {
      "epoch": 0.125,
      "grad_norm": 1.4780454635620117,
      "learning_rate": 0.00019875,
      "loss": 0.9304,
      "step": 40
    },
    {
      "epoch": 0.15625,
      "grad_norm": 0.8681000471115112,
      "learning_rate": 0.00019843750000000002,
      "loss": 0.9191,
      "step": 50
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.9179439544677734,
      "learning_rate": 0.000198125,
      "loss": 0.929,
      "step": 60
    },
    {
      "epoch": 0.21875,
      "grad_norm": 1.7340352535247803,
      "learning_rate": 0.0001978125,
      "loss": 0.878,
      "step": 70
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0093904733657837,
      "learning_rate": 0.00019750000000000003,
      "loss": 0.9233,
      "step": 80
    },
    {
      "epoch": 0.28125,
      "grad_norm": 2.5266435146331787,
      "learning_rate": 0.00019718750000000002,
      "loss": 0.9763,
      "step": 90
    },
    {
      "epoch": 0.3125,
      "grad_norm": 3.176175594329834,
      "learning_rate": 0.000196875,
      "loss": 0.8826,
      "step": 100
    },
    {
      "epoch": 0.3125,
      "eval_accuracy": 0.575,
      "eval_loss": 0.9027458429336548,
      "eval_runtime": 6.7552,
      "eval_samples_per_second": 189.483,
      "eval_steps_per_second": 23.685,
      "step": 100
    },
    {
      "epoch": 0.34375,
      "grad_norm": 2.2321112155914307,
      "learning_rate": 0.00019656250000000001,
      "loss": 0.8653,
      "step": 110
    },
    {
      "epoch": 0.375,
      "grad_norm": 2.9958410263061523,
      "learning_rate": 0.00019625,
      "loss": 0.8645,
      "step": 120
    },
    {
      "epoch": 0.40625,
      "grad_norm": 1.3575609922409058,
      "learning_rate": 0.0001959375,
      "loss": 0.8028,
      "step": 130
    },
    {
      "epoch": 0.4375,
      "grad_norm": 2.6939704418182373,
      "learning_rate": 0.00019562500000000003,
      "loss": 0.7853,
      "step": 140
    },
    {
      "epoch": 0.46875,
      "grad_norm": 3.4280309677124023,
      "learning_rate": 0.0001953125,
      "loss": 0.8574,
      "step": 150
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.5754916667938232,
      "learning_rate": 0.000195,
      "loss": 0.9127,
      "step": 160
    },
    {
      "epoch": 0.53125,
      "grad_norm": 1.0515691041946411,
      "learning_rate": 0.0001946875,
      "loss": 0.6715,
      "step": 170
    },
    {
      "epoch": 0.5625,
      "grad_norm": 4.097135066986084,
      "learning_rate": 0.00019437500000000002,
      "loss": 0.9268,
      "step": 180
    },
    {
      "epoch": 0.59375,
      "grad_norm": 1.1513463258743286,
      "learning_rate": 0.0001940625,
      "loss": 1.1014,
      "step": 190
    },
    {
      "epoch": 0.625,
      "grad_norm": 2.027341604232788,
      "learning_rate": 0.00019375000000000002,
      "loss": 0.8908,
      "step": 200
    },
    {
      "epoch": 0.625,
      "eval_accuracy": 0.5984375,
      "eval_loss": 0.8483627438545227,
      "eval_runtime": 6.1939,
      "eval_samples_per_second": 206.656,
      "eval_steps_per_second": 25.832,
      "step": 200
    },
    {
      "epoch": 0.65625,
      "grad_norm": 3.054396629333496,
      "learning_rate": 0.0001934375,
      "loss": 1.0798,
      "step": 210
    },
    {
      "epoch": 0.6875,
      "grad_norm": 1.4708349704742432,
      "learning_rate": 0.000193125,
      "loss": 0.9002,
      "step": 220
    },
    {
      "epoch": 0.71875,
      "grad_norm": 3.5276074409484863,
      "learning_rate": 0.00019281250000000003,
      "loss": 0.7594,
      "step": 230
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7735655307769775,
      "learning_rate": 0.00019250000000000002,
      "loss": 0.8413,
      "step": 240
    },
    {
      "epoch": 0.78125,
      "grad_norm": 2.533834457397461,
      "learning_rate": 0.0001921875,
      "loss": 0.6925,
      "step": 250
    },
    {
      "epoch": 0.8125,
      "grad_norm": 1.8991681337356567,
      "learning_rate": 0.00019187500000000002,
      "loss": 0.7569,
      "step": 260
    },
    {
      "epoch": 0.84375,
      "grad_norm": 1.8567323684692383,
      "learning_rate": 0.0001915625,
      "loss": 0.8479,
      "step": 270
    },
    {
      "epoch": 0.875,
      "grad_norm": 2.904021739959717,
      "learning_rate": 0.00019125000000000001,
      "loss": 0.7737,
      "step": 280
    },
    {
      "epoch": 0.90625,
      "grad_norm": 2.1224987506866455,
      "learning_rate": 0.00019093750000000003,
      "loss": 0.7647,
      "step": 290
    },
    {
      "epoch": 0.9375,
      "grad_norm": 2.221484661102295,
      "learning_rate": 0.000190625,
      "loss": 0.8229,
      "step": 300
    },
    {
      "epoch": 0.9375,
      "eval_accuracy": 0.66953125,
      "eval_loss": 0.7514046430587769,
      "eval_runtime": 6.6282,
      "eval_samples_per_second": 193.115,
      "eval_steps_per_second": 24.139,
      "step": 300
    },
    {
      "epoch": 0.96875,
      "grad_norm": 2.0415520668029785,
      "learning_rate": 0.0001903125,
      "loss": 0.8343,
      "step": 310
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7857444286346436,
      "learning_rate": 0.00019,
      "loss": 0.7851,
      "step": 320
    },
    {
      "epoch": 1.03125,
      "grad_norm": 3.682004451751709,
      "learning_rate": 0.00018968750000000002,
      "loss": 0.659,
      "step": 330
    },
    {
      "epoch": 1.0625,
      "grad_norm": 3.909653902053833,
      "learning_rate": 0.000189375,
      "loss": 0.5986,
      "step": 340
    },
    {
      "epoch": 1.09375,
      "grad_norm": 2.665677785873413,
      "learning_rate": 0.00018906250000000002,
      "loss": 0.599,
      "step": 350
    },
    {
      "epoch": 1.125,
      "grad_norm": 3.378941059112549,
      "learning_rate": 0.00018875,
      "loss": 0.6162,
      "step": 360
    },
    {
      "epoch": 1.15625,
      "grad_norm": 2.2860324382781982,
      "learning_rate": 0.0001884375,
      "loss": 0.5133,
      "step": 370
    },
    {
      "epoch": 1.1875,
      "grad_norm": 4.41148567199707,
      "learning_rate": 0.000188125,
      "loss": 0.4787,
      "step": 380
    },
    {
      "epoch": 1.21875,
      "grad_norm": 2.802339792251587,
      "learning_rate": 0.00018781250000000002,
      "loss": 0.8346,
      "step": 390
    },
    {
      "epoch": 1.25,
      "grad_norm": 3.0579328536987305,
      "learning_rate": 0.0001875,
      "loss": 0.5299,
      "step": 400
    },
    {
      "epoch": 1.25,
      "eval_accuracy": 0.71640625,
      "eval_loss": 0.6797709465026855,
      "eval_runtime": 6.1605,
      "eval_samples_per_second": 207.776,
      "eval_steps_per_second": 25.972,
      "step": 400
    },
    {
      "epoch": 1.28125,
      "grad_norm": 4.516961574554443,
      "learning_rate": 0.00018718750000000002,
      "loss": 0.5706,
      "step": 410
    },
    {
      "epoch": 1.3125,
      "grad_norm": 4.825928688049316,
      "learning_rate": 0.000186875,
      "loss": 0.4748,
      "step": 420
    },
    {
      "epoch": 1.34375,
      "grad_norm": 4.58062219619751,
      "learning_rate": 0.00018656250000000001,
      "loss": 0.589,
      "step": 430
    },
    {
      "epoch": 1.375,
      "grad_norm": 3.5321483612060547,
      "learning_rate": 0.00018625,
      "loss": 0.4854,
      "step": 440
    },
    {
      "epoch": 1.40625,
      "grad_norm": 2.8487465381622314,
      "learning_rate": 0.0001859375,
      "loss": 0.4736,
      "step": 450
    },
    {
      "epoch": 1.4375,
      "grad_norm": 3.3944597244262695,
      "learning_rate": 0.000185625,
      "loss": 0.5127,
      "step": 460
    },
    {
      "epoch": 1.46875,
      "grad_norm": 3.1018853187561035,
      "learning_rate": 0.0001853125,
      "loss": 0.5043,
      "step": 470
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7340843677520752,
      "learning_rate": 0.00018500000000000002,
      "loss": 0.4293,
      "step": 480
    },
    {
      "epoch": 1.53125,
      "grad_norm": 5.125267505645752,
      "learning_rate": 0.0001846875,
      "loss": 0.4616,
      "step": 490
    },
    {
      "epoch": 1.5625,
      "grad_norm": 4.208590030670166,
      "learning_rate": 0.000184375,
      "loss": 0.5207,
      "step": 500
    },
    {
      "epoch": 1.5625,
      "eval_accuracy": 0.7375,
      "eval_loss": 0.6465662717819214,
      "eval_runtime": 6.4735,
      "eval_samples_per_second": 197.729,
      "eval_steps_per_second": 24.716,
      "step": 500
    },
    {
      "epoch": 1.59375,
      "grad_norm": 3.470857620239258,
      "learning_rate": 0.0001840625,
      "loss": 0.5758,
      "step": 510
    },
    {
      "epoch": 1.625,
      "grad_norm": 2.353254556655884,
      "learning_rate": 0.00018375,
      "loss": 0.4954,
      "step": 520
    },
    {
      "epoch": 1.65625,
      "grad_norm": 5.26539421081543,
      "learning_rate": 0.0001834375,
      "loss": 0.4687,
      "step": 530
    },
    {
      "epoch": 1.6875,
      "grad_norm": 4.430329322814941,
      "learning_rate": 0.00018312500000000002,
      "loss": 0.3183,
      "step": 540
    },
    {
      "epoch": 1.71875,
      "grad_norm": 3.066425323486328,
      "learning_rate": 0.0001828125,
      "loss": 0.4647,
      "step": 550
    },
    {
      "epoch": 1.75,
      "grad_norm": 3.959084987640381,
      "learning_rate": 0.0001825,
      "loss": 0.3383,
      "step": 560
    },
    {
      "epoch": 1.78125,
      "grad_norm": 7.477952480316162,
      "learning_rate": 0.0001821875,
      "loss": 0.4317,
      "step": 570
    },
    {
      "epoch": 1.8125,
      "grad_norm": 3.7508246898651123,
      "learning_rate": 0.00018187500000000002,
      "loss": 0.621,
      "step": 580
    },
    {
      "epoch": 1.84375,
      "grad_norm": 6.189945220947266,
      "learning_rate": 0.0001815625,
      "loss": 0.5086,
      "step": 590
    },
    {
      "epoch": 1.875,
      "grad_norm": 6.064743995666504,
      "learning_rate": 0.00018125000000000001,
      "loss": 0.4967,
      "step": 600
    },
    {
      "epoch": 1.875,
      "eval_accuracy": 0.74609375,
      "eval_loss": 0.6302900314331055,
      "eval_runtime": 6.8591,
      "eval_samples_per_second": 186.614,
      "eval_steps_per_second": 23.327,
      "step": 600
    },
    {
      "epoch": 1.90625,
      "grad_norm": 2.6553187370300293,
      "learning_rate": 0.0001809375,
      "loss": 0.4273,
      "step": 610
    },
    {
      "epoch": 1.9375,
      "grad_norm": 8.86166000366211,
      "learning_rate": 0.000180625,
      "loss": 0.4264,
      "step": 620
    },
    {
      "epoch": 1.96875,
      "grad_norm": 5.616076946258545,
      "learning_rate": 0.00018031250000000003,
      "loss": 0.4907,
      "step": 630
    },
    {
      "epoch": 2.0,
      "grad_norm": 4.036799907684326,
      "learning_rate": 0.00018,
      "loss": 0.4537,
      "step": 640
    },
    {
      "epoch": 2.03125,
      "grad_norm": 4.771281719207764,
      "learning_rate": 0.0001796875,
      "loss": 0.2492,
      "step": 650
    },
    {
      "epoch": 2.0625,
      "grad_norm": 3.654841423034668,
      "learning_rate": 0.000179375,
      "loss": 0.2568,
      "step": 660
    },
    {
      "epoch": 2.09375,
      "grad_norm": 3.043989896774292,
      "learning_rate": 0.0001790625,
      "loss": 0.2354,
      "step": 670
    },
    {
      "epoch": 2.125,
      "grad_norm": 6.0935516357421875,
      "learning_rate": 0.00017875,
      "loss": 0.2106,
      "step": 680
    },
    {
      "epoch": 2.15625,
      "grad_norm": 2.231640577316284,
      "learning_rate": 0.00017843750000000002,
      "loss": 0.2804,
      "step": 690
    },
    {
      "epoch": 2.1875,
      "grad_norm": 8.500765800476074,
      "learning_rate": 0.000178125,
      "loss": 0.3977,
      "step": 700
    },
    {
      "epoch": 2.1875,
      "eval_accuracy": 0.771875,
      "eval_loss": 0.7239754796028137,
      "eval_runtime": 6.8507,
      "eval_samples_per_second": 186.843,
      "eval_steps_per_second": 23.355,
      "step": 700
    },
    {
      "epoch": 2.21875,
      "grad_norm": 5.513547420501709,
      "learning_rate": 0.0001778125,
      "loss": 0.3942,
      "step": 710
    },
    {
      "epoch": 2.25,
      "grad_norm": 6.819497585296631,
      "learning_rate": 0.0001775,
      "loss": 0.2489,
      "step": 720
    },
    {
      "epoch": 2.28125,
      "grad_norm": 2.9887595176696777,
      "learning_rate": 0.00017718750000000002,
      "loss": 0.2054,
      "step": 730
    },
    {
      "epoch": 2.3125,
      "grad_norm": 2.461519479751587,
      "learning_rate": 0.000176875,
      "loss": 0.2252,
      "step": 740
    },
    {
      "epoch": 2.34375,
      "grad_norm": 1.7205028533935547,
      "learning_rate": 0.00017656250000000002,
      "loss": 0.1975,
      "step": 750
    },
    {
      "epoch": 2.375,
      "grad_norm": 0.9731912612915039,
      "learning_rate": 0.00017625,
      "loss": 0.2049,
      "step": 760
    },
    {
      "epoch": 2.40625,
      "grad_norm": 4.950948715209961,
      "learning_rate": 0.0001759375,
      "loss": 0.283,
      "step": 770
    },
    {
      "epoch": 2.4375,
      "grad_norm": 3.5972580909729004,
      "learning_rate": 0.00017562500000000003,
      "loss": 0.2172,
      "step": 780
    },
    {
      "epoch": 2.46875,
      "grad_norm": 8.976056098937988,
      "learning_rate": 0.0001753125,
      "loss": 0.2733,
      "step": 790
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2202197313308716,
      "learning_rate": 0.000175,
      "loss": 0.2744,
      "step": 800
    },
    {
      "epoch": 2.5,
      "eval_accuracy": 0.8734375,
      "eval_loss": 0.35440793633461,
      "eval_runtime": 6.2483,
      "eval_samples_per_second": 204.854,
      "eval_steps_per_second": 25.607,
      "step": 800
    },
    {
      "epoch": 2.53125,
      "grad_norm": 7.248142719268799,
      "learning_rate": 0.0001746875,
      "loss": 0.2173,
      "step": 810
    },
    {
      "epoch": 2.5625,
      "grad_norm": 6.068428039550781,
      "learning_rate": 0.000174375,
      "loss": 0.2172,
      "step": 820
    },
    {
      "epoch": 2.59375,
      "grad_norm": 5.297908306121826,
      "learning_rate": 0.0001740625,
      "loss": 0.3827,
      "step": 830
    },
    {
      "epoch": 2.625,
      "grad_norm": 8.936563491821289,
      "learning_rate": 0.00017375000000000002,
      "loss": 0.2638,
      "step": 840
    },
    {
      "epoch": 2.65625,
      "grad_norm": 5.005488395690918,
      "learning_rate": 0.0001734375,
      "loss": 0.281,
      "step": 850
    },
    {
      "epoch": 2.6875,
      "grad_norm": 0.3621249794960022,
      "learning_rate": 0.000173125,
      "loss": 0.1424,
      "step": 860
    },
    {
      "epoch": 2.71875,
      "grad_norm": 2.321965217590332,
      "learning_rate": 0.0001728125,
      "loss": 0.2586,
      "step": 870
    },
    {
      "epoch": 2.75,
      "grad_norm": 9.775487899780273,
      "learning_rate": 0.00017250000000000002,
      "loss": 0.185,
      "step": 880
    },
    {
      "epoch": 2.78125,
      "grad_norm": 0.6502866744995117,
      "learning_rate": 0.0001721875,
      "loss": 0.1635,
      "step": 890
    },
    {
      "epoch": 2.8125,
      "grad_norm": 3.4629428386688232,
      "learning_rate": 0.00017187500000000002,
      "loss": 0.4271,
      "step": 900
    },
    {
      "epoch": 2.8125,
      "eval_accuracy": 0.89375,
      "eval_loss": 0.30369627475738525,
      "eval_runtime": 6.7624,
      "eval_samples_per_second": 189.281,
      "eval_steps_per_second": 23.66,
      "step": 900
    },
    {
      "epoch": 2.84375,
      "grad_norm": 1.4681894779205322,
      "learning_rate": 0.0001715625,
      "loss": 0.193,
      "step": 910
    },
    {
      "epoch": 2.875,
      "grad_norm": 2.8025732040405273,
      "learning_rate": 0.00017125,
      "loss": 0.2532,
      "step": 920
    },
    {
      "epoch": 2.90625,
      "grad_norm": 5.300601482391357,
      "learning_rate": 0.00017093750000000003,
      "loss": 0.178,
      "step": 930
    },
    {
      "epoch": 2.9375,
      "grad_norm": 0.9630404710769653,
      "learning_rate": 0.00017062500000000001,
      "loss": 0.2453,
      "step": 940
    },
    {
      "epoch": 2.96875,
      "grad_norm": 3.942497968673706,
      "learning_rate": 0.0001703125,
      "loss": 0.2302,
      "step": 950
    },
    {
      "epoch": 3.0,
      "grad_norm": 5.0087809562683105,
      "learning_rate": 0.00017,
      "loss": 0.2989,
      "step": 960
    },
    {
      "epoch": 3.03125,
      "grad_norm": 6.783499717712402,
      "learning_rate": 0.0001696875,
      "loss": 0.3932,
      "step": 970
    },
    {
      "epoch": 3.0625,
      "grad_norm": 4.156474590301514,
      "learning_rate": 0.000169375,
      "loss": 0.1148,
      "step": 980
    },
    {
      "epoch": 3.09375,
      "grad_norm": 0.22207698225975037,
      "learning_rate": 0.00016906250000000002,
      "loss": 0.0925,
      "step": 990
    },
    {
      "epoch": 3.125,
      "grad_norm": 7.46227502822876,
      "learning_rate": 0.00016875,
      "loss": 0.2484,
      "step": 1000
    },
    {
      "epoch": 3.125,
      "eval_accuracy": 0.86015625,
      "eval_loss": 0.4111490249633789,
      "eval_runtime": 6.2118,
      "eval_samples_per_second": 206.058,
      "eval_steps_per_second": 25.757,
      "step": 1000
    },
    {
      "epoch": 3.15625,
      "grad_norm": 1.8679814338684082,
      "learning_rate": 0.0001684375,
      "loss": 0.1024,
      "step": 1010
    },
    {
      "epoch": 3.1875,
      "grad_norm": 1.4344357252120972,
      "learning_rate": 0.000168125,
      "loss": 0.1532,
      "step": 1020
    },
    {
      "epoch": 3.21875,
      "grad_norm": 5.7756218910217285,
      "learning_rate": 0.00016781250000000002,
      "loss": 0.1433,
      "step": 1030
    },
    {
      "epoch": 3.25,
      "grad_norm": 6.2325358390808105,
      "learning_rate": 0.0001675,
      "loss": 0.2056,
      "step": 1040
    },
    {
      "epoch": 3.28125,
      "grad_norm": 0.7062513828277588,
      "learning_rate": 0.00016718750000000002,
      "loss": 0.1883,
      "step": 1050
    },
    {
      "epoch": 3.3125,
      "grad_norm": 0.5934551954269409,
      "learning_rate": 0.000166875,
      "loss": 0.1311,
      "step": 1060
    },
    {
      "epoch": 3.34375,
      "grad_norm": 2.90659236907959,
      "learning_rate": 0.0001665625,
      "loss": 0.0926,
      "step": 1070
    },
    {
      "epoch": 3.375,
      "grad_norm": 0.12364790588617325,
      "learning_rate": 0.00016625000000000003,
      "loss": 0.0998,
      "step": 1080
    },
    {
      "epoch": 3.40625,
      "grad_norm": 8.838610649108887,
      "learning_rate": 0.00016593750000000002,
      "loss": 0.1322,
      "step": 1090
    },
    {
      "epoch": 3.4375,
      "grad_norm": 0.9064333438873291,
      "learning_rate": 0.000165625,
      "loss": 0.0797,
      "step": 1100
    },
    {
      "epoch": 3.4375,
      "eval_accuracy": 0.8953125,
      "eval_loss": 0.3781999349594116,
      "eval_runtime": 6.8762,
      "eval_samples_per_second": 186.149,
      "eval_steps_per_second": 23.269,
      "step": 1100
    },
    {
      "epoch": 3.46875,
      "grad_norm": 0.37031790614128113,
      "learning_rate": 0.0001653125,
      "loss": 0.1157,
      "step": 1110
    },
    {
      "epoch": 3.5,
      "grad_norm": 10.903687477111816,
      "learning_rate": 0.000165,
      "loss": 0.1616,
      "step": 1120
    },
    {
      "epoch": 3.53125,
      "grad_norm": 7.575557231903076,
      "learning_rate": 0.0001646875,
      "loss": 0.0864,
      "step": 1130
    },
    {
      "epoch": 3.5625,
      "grad_norm": 0.7798458337783813,
      "learning_rate": 0.00016437500000000002,
      "loss": 0.1479,
      "step": 1140
    },
    {
      "epoch": 3.59375,
      "grad_norm": 0.30808526277542114,
      "learning_rate": 0.0001640625,
      "loss": 0.091,
      "step": 1150
    },
    {
      "epoch": 3.625,
      "grad_norm": 1.3779064416885376,
      "learning_rate": 0.00016375,
      "loss": 0.0918,
      "step": 1160
    },
    {
      "epoch": 3.65625,
      "grad_norm": 10.742758750915527,
      "learning_rate": 0.0001634375,
      "loss": 0.213,
      "step": 1170
    },
    {
      "epoch": 3.6875,
      "grad_norm": 4.325877666473389,
      "learning_rate": 0.00016312500000000002,
      "loss": 0.1672,
      "step": 1180
    },
    {
      "epoch": 3.71875,
      "grad_norm": 0.06019480153918266,
      "learning_rate": 0.0001628125,
      "loss": 0.1229,
      "step": 1190
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.060391902923583984,
      "learning_rate": 0.00016250000000000002,
      "loss": 0.0662,
      "step": 1200
    },
    {
      "epoch": 3.75,
      "eval_accuracy": 0.9171875,
      "eval_loss": 0.309553325176239,
      "eval_runtime": 6.3504,
      "eval_samples_per_second": 201.561,
      "eval_steps_per_second": 25.195,
      "step": 1200
    },
    {
      "epoch": 3.78125,
      "grad_norm": 7.1998066902160645,
      "learning_rate": 0.0001621875,
      "loss": 0.2368,
      "step": 1210
    },
    {
      "epoch": 3.8125,
      "grad_norm": 2.0582878589630127,
      "learning_rate": 0.000161875,
      "loss": 0.1161,
      "step": 1220
    },
    {
      "epoch": 3.84375,
      "grad_norm": 0.4987049698829651,
      "learning_rate": 0.0001615625,
      "loss": 0.1798,
      "step": 1230
    },
    {
      "epoch": 3.875,
      "grad_norm": 6.303996562957764,
      "learning_rate": 0.00016125000000000002,
      "loss": 0.2193,
      "step": 1240
    },
    {
      "epoch": 3.90625,
      "grad_norm": 0.1131618395447731,
      "learning_rate": 0.0001609375,
      "loss": 0.1626,
      "step": 1250
    },
    {
      "epoch": 3.9375,
      "grad_norm": 8.112652778625488,
      "learning_rate": 0.00016062500000000001,
      "loss": 0.1033,
      "step": 1260
    },
    {
      "epoch": 3.96875,
      "grad_norm": 0.14582502841949463,
      "learning_rate": 0.0001603125,
      "loss": 0.0548,
      "step": 1270
    },
    {
      "epoch": 4.0,
      "grad_norm": 7.206060409545898,
      "learning_rate": 0.00016,
      "loss": 0.1758,
      "step": 1280
    },
    {
      "epoch": 4.03125,
      "grad_norm": 0.12302320450544357,
      "learning_rate": 0.0001596875,
      "loss": 0.0118,
      "step": 1290
    },
    {
      "epoch": 4.0625,
      "grad_norm": 1.9936612844467163,
      "learning_rate": 0.000159375,
      "loss": 0.0894,
      "step": 1300
    },
    {
      "epoch": 4.0625,
      "eval_accuracy": 0.92890625,
      "eval_loss": 0.281791090965271,
      "eval_runtime": 6.6912,
      "eval_samples_per_second": 191.295,
      "eval_steps_per_second": 23.912,
      "step": 1300
    },
    {
      "epoch": 4.09375,
      "grad_norm": 15.060647010803223,
      "learning_rate": 0.0001590625,
      "loss": 0.0328,
      "step": 1310
    },
    {
      "epoch": 4.125,
      "grad_norm": 0.06893154978752136,
      "learning_rate": 0.00015875,
      "loss": 0.0508,
      "step": 1320
    },
    {
      "epoch": 4.15625,
      "grad_norm": 0.38513150811195374,
      "learning_rate": 0.00015843750000000002,
      "loss": 0.054,
      "step": 1330
    },
    {
      "epoch": 4.1875,
      "grad_norm": 9.78172492980957,
      "learning_rate": 0.000158125,
      "loss": 0.0225,
      "step": 1340
    },
    {
      "epoch": 4.21875,
      "grad_norm": 0.9426405429840088,
      "learning_rate": 0.00015781250000000002,
      "loss": 0.0897,
      "step": 1350
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.03851567581295967,
      "learning_rate": 0.0001575,
      "loss": 0.0723,
      "step": 1360
    },
    {
      "epoch": 4.28125,
      "grad_norm": 3.465240001678467,
      "learning_rate": 0.0001571875,
      "loss": 0.0281,
      "step": 1370
    },
    {
      "epoch": 4.3125,
      "grad_norm": 4.168702125549316,
      "learning_rate": 0.000156875,
      "loss": 0.0334,
      "step": 1380
    },
    {
      "epoch": 4.34375,
      "grad_norm": 0.022729417309165,
      "learning_rate": 0.00015656250000000002,
      "loss": 0.0786,
      "step": 1390
    },
    {
      "epoch": 4.375,
      "grad_norm": 4.370222568511963,
      "learning_rate": 0.00015625,
      "loss": 0.1005,
      "step": 1400
    },
    {
      "epoch": 4.375,
      "eval_accuracy": 0.946875,
      "eval_loss": 0.21635571122169495,
      "eval_runtime": 6.8886,
      "eval_samples_per_second": 185.815,
      "eval_steps_per_second": 23.227,
      "step": 1400
    },
    {
      "epoch": 4.40625,
      "grad_norm": 2.6576523780822754,
      "learning_rate": 0.00015593750000000002,
      "loss": 0.0235,
      "step": 1410
    },
    {
      "epoch": 4.4375,
      "grad_norm": 9.146688461303711,
      "learning_rate": 0.000155625,
      "loss": 0.0842,
      "step": 1420
    },
    {
      "epoch": 4.46875,
      "grad_norm": 0.019813504070043564,
      "learning_rate": 0.00015531250000000001,
      "loss": 0.092,
      "step": 1430
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.2571779191493988,
      "learning_rate": 0.000155,
      "loss": 0.0473,
      "step": 1440
    },
    {
      "epoch": 4.53125,
      "grad_norm": 5.699305057525635,
      "learning_rate": 0.0001546875,
      "loss": 0.1053,
      "step": 1450
    },
    {
      "epoch": 4.5625,
      "grad_norm": 0.04671861603856087,
      "learning_rate": 0.000154375,
      "loss": 0.0068,
      "step": 1460
    },
    {
      "epoch": 4.59375,
      "grad_norm": 0.1444374918937683,
      "learning_rate": 0.0001540625,
      "loss": 0.0946,
      "step": 1470
    },
    {
      "epoch": 4.625,
      "grad_norm": 0.06924466788768768,
      "learning_rate": 0.00015375000000000002,
      "loss": 0.0213,
      "step": 1480
    },
    {
      "epoch": 4.65625,
      "grad_norm": 0.03169933706521988,
      "learning_rate": 0.0001534375,
      "loss": 0.24,
      "step": 1490
    },
    {
      "epoch": 4.6875,
      "grad_norm": 4.107417106628418,
      "learning_rate": 0.000153125,
      "loss": 0.0997,
      "step": 1500
    },
    {
      "epoch": 4.6875,
      "eval_accuracy": 0.9109375,
      "eval_loss": 0.3378385901451111,
      "eval_runtime": 6.6178,
      "eval_samples_per_second": 193.418,
      "eval_steps_per_second": 24.177,
      "step": 1500
    },
    {
      "epoch": 4.71875,
      "grad_norm": 0.2504006028175354,
      "learning_rate": 0.0001528125,
      "loss": 0.0903,
      "step": 1510
    },
    {
      "epoch": 4.75,
      "grad_norm": 3.6016788482666016,
      "learning_rate": 0.0001525,
      "loss": 0.1676,
      "step": 1520
    },
    {
      "epoch": 4.78125,
      "grad_norm": 0.02008502557873726,
      "learning_rate": 0.0001521875,
      "loss": 0.0731,
      "step": 1530
    },
    {
      "epoch": 4.8125,
      "grad_norm": 0.019816860556602478,
      "learning_rate": 0.00015187500000000002,
      "loss": 0.0559,
      "step": 1540
    },
    {
      "epoch": 4.84375,
      "grad_norm": 1.7278194427490234,
      "learning_rate": 0.0001515625,
      "loss": 0.0306,
      "step": 1550
    },
    {
      "epoch": 4.875,
      "grad_norm": 0.19783662259578705,
      "learning_rate": 0.00015125,
      "loss": 0.1335,
      "step": 1560
    },
    {
      "epoch": 4.90625,
      "grad_norm": 11.390993118286133,
      "learning_rate": 0.0001509375,
      "loss": 0.1606,
      "step": 1570
    },
    {
      "epoch": 4.9375,
      "grad_norm": 4.290327072143555,
      "learning_rate": 0.00015062500000000002,
      "loss": 0.0807,
      "step": 1580
    },
    {
      "epoch": 4.96875,
      "grad_norm": 0.28723642230033875,
      "learning_rate": 0.0001503125,
      "loss": 0.1673,
      "step": 1590
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.05478620529174805,
      "learning_rate": 0.00015000000000000001,
      "loss": 0.0715,
      "step": 1600
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.91328125,
      "eval_loss": 0.3626965880393982,
      "eval_runtime": 6.9646,
      "eval_samples_per_second": 183.788,
      "eval_steps_per_second": 22.973,
      "step": 1600
    },
    {
      "epoch": 5.03125,
      "grad_norm": 3.0136938095092773,
      "learning_rate": 0.0001496875,
      "loss": 0.1965,
      "step": 1610
    },
    {
      "epoch": 5.0625,
      "grad_norm": 0.11938930302858353,
      "learning_rate": 0.00014937499999999999,
      "loss": 0.0065,
      "step": 1620
    },
    {
      "epoch": 5.09375,
      "grad_norm": 0.04192354157567024,
      "learning_rate": 0.00014906250000000003,
      "loss": 0.0813,
      "step": 1630
    },
    {
      "epoch": 5.125,
      "grad_norm": 0.018585534766316414,
      "learning_rate": 0.00014875,
      "loss": 0.0973,
      "step": 1640
    },
    {
      "epoch": 5.15625,
      "grad_norm": 3.9032442569732666,
      "learning_rate": 0.0001484375,
      "loss": 0.0587,
      "step": 1650
    },
    {
      "epoch": 5.1875,
      "grad_norm": 0.04510480910539627,
      "learning_rate": 0.000148125,
      "loss": 0.1506,
      "step": 1660
    },
    {
      "epoch": 5.21875,
      "grad_norm": 0.2749982178211212,
      "learning_rate": 0.0001478125,
      "loss": 0.1367,
      "step": 1670
    },
    {
      "epoch": 5.25,
      "grad_norm": 6.217327117919922,
      "learning_rate": 0.0001475,
      "loss": 0.0734,
      "step": 1680
    },
    {
      "epoch": 5.28125,
      "grad_norm": 0.11703751981258392,
      "learning_rate": 0.00014718750000000002,
      "loss": 0.0961,
      "step": 1690
    },
    {
      "epoch": 5.3125,
      "grad_norm": 1.1948308944702148,
      "learning_rate": 0.000146875,
      "loss": 0.0567,
      "step": 1700
    },
    {
      "epoch": 5.3125,
      "eval_accuracy": 0.9234375,
      "eval_loss": 0.30611464381217957,
      "eval_runtime": 7.2211,
      "eval_samples_per_second": 177.258,
      "eval_steps_per_second": 22.157,
      "step": 1700
    },
    {
      "epoch": 5.34375,
      "grad_norm": 0.025335168465971947,
      "learning_rate": 0.0001465625,
      "loss": 0.0511,
      "step": 1710
    },
    {
      "epoch": 5.375,
      "grad_norm": 0.033230509608983994,
      "learning_rate": 0.00014625,
      "loss": 0.0857,
      "step": 1720
    },
    {
      "epoch": 5.40625,
      "grad_norm": 0.027418823912739754,
      "learning_rate": 0.00014593750000000002,
      "loss": 0.0054,
      "step": 1730
    },
    {
      "epoch": 5.4375,
      "grad_norm": 7.358951568603516,
      "learning_rate": 0.000145625,
      "loss": 0.0582,
      "step": 1740
    },
    {
      "epoch": 5.46875,
      "grad_norm": 0.06502091139554977,
      "learning_rate": 0.00014531250000000002,
      "loss": 0.11,
      "step": 1750
    },
    {
      "epoch": 5.5,
      "grad_norm": 3.4256505966186523,
      "learning_rate": 0.000145,
      "loss": 0.1004,
      "step": 1760
    },
    {
      "epoch": 5.53125,
      "grad_norm": 0.023563764989376068,
      "learning_rate": 0.0001446875,
      "loss": 0.0066,
      "step": 1770
    },
    {
      "epoch": 5.5625,
      "grad_norm": 0.049036819487810135,
      "learning_rate": 0.00014437500000000003,
      "loss": 0.0271,
      "step": 1780
    },
    {
      "epoch": 5.59375,
      "grad_norm": 0.02042466588318348,
      "learning_rate": 0.0001440625,
      "loss": 0.0132,
      "step": 1790
    },
    {
      "epoch": 5.625,
      "grad_norm": 0.05669878423213959,
      "learning_rate": 0.00014375,
      "loss": 0.0558,
      "step": 1800
    },
    {
      "epoch": 5.625,
      "eval_accuracy": 0.94609375,
      "eval_loss": 0.23927736282348633,
      "eval_runtime": 6.5481,
      "eval_samples_per_second": 195.476,
      "eval_steps_per_second": 24.434,
      "step": 1800
    },
    {
      "epoch": 5.65625,
      "grad_norm": 0.9565289616584778,
      "learning_rate": 0.0001434375,
      "loss": 0.0366,
      "step": 1810
    },
    {
      "epoch": 5.6875,
      "grad_norm": 0.024000531062483788,
      "learning_rate": 0.000143125,
      "loss": 0.14,
      "step": 1820
    },
    {
      "epoch": 5.71875,
      "grad_norm": 6.16231107711792,
      "learning_rate": 0.0001428125,
      "loss": 0.1586,
      "step": 1830
    },
    {
      "epoch": 5.75,
      "grad_norm": 10.281587600708008,
      "learning_rate": 0.00014250000000000002,
      "loss": 0.1454,
      "step": 1840
    },
    {
      "epoch": 5.78125,
      "grad_norm": 8.064495086669922,
      "learning_rate": 0.0001421875,
      "loss": 0.1354,
      "step": 1850
    },
    {
      "epoch": 5.8125,
      "grad_norm": 0.27762091159820557,
      "learning_rate": 0.000141875,
      "loss": 0.0356,
      "step": 1860
    },
    {
      "epoch": 5.84375,
      "grad_norm": 7.199831962585449,
      "learning_rate": 0.0001415625,
      "loss": 0.0485,
      "step": 1870
    },
    {
      "epoch": 5.875,
      "grad_norm": 0.013454968109726906,
      "learning_rate": 0.00014125000000000002,
      "loss": 0.0317,
      "step": 1880
    },
    {
      "epoch": 5.90625,
      "grad_norm": 0.8754172921180725,
      "learning_rate": 0.0001409375,
      "loss": 0.0198,
      "step": 1890
    },
    {
      "epoch": 5.9375,
      "grad_norm": 0.06731715798377991,
      "learning_rate": 0.00014062500000000002,
      "loss": 0.0061,
      "step": 1900
    },
    {
      "epoch": 5.9375,
      "eval_accuracy": 0.95859375,
      "eval_loss": 0.17380020022392273,
      "eval_runtime": 6.5259,
      "eval_samples_per_second": 196.142,
      "eval_steps_per_second": 24.518,
      "step": 1900
    },
    {
      "epoch": 5.96875,
      "grad_norm": 1.3012233972549438,
      "learning_rate": 0.0001403125,
      "loss": 0.1023,
      "step": 1910
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.058877017349004745,
      "learning_rate": 0.00014,
      "loss": 0.0555,
      "step": 1920
    },
    {
      "epoch": 6.03125,
      "grad_norm": 1.3218984603881836,
      "learning_rate": 0.00013968750000000003,
      "loss": 0.1671,
      "step": 1930
    },
    {
      "epoch": 6.0625,
      "grad_norm": 0.7441987991333008,
      "learning_rate": 0.000139375,
      "loss": 0.046,
      "step": 1940
    },
    {
      "epoch": 6.09375,
      "grad_norm": 0.3219761550426483,
      "learning_rate": 0.0001390625,
      "loss": 0.0741,
      "step": 1950
    },
    {
      "epoch": 6.125,
      "grad_norm": 0.3803882598876953,
      "learning_rate": 0.00013875,
      "loss": 0.1102,
      "step": 1960
    },
    {
      "epoch": 6.15625,
      "grad_norm": 0.034619808197021484,
      "learning_rate": 0.0001384375,
      "loss": 0.0844,
      "step": 1970
    },
    {
      "epoch": 6.1875,
      "grad_norm": 0.03565617650747299,
      "learning_rate": 0.000138125,
      "loss": 0.0162,
      "step": 1980
    },
    {
      "epoch": 6.21875,
      "grad_norm": 0.05813159421086311,
      "learning_rate": 0.00013781250000000002,
      "loss": 0.0303,
      "step": 1990
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.16888560354709625,
      "learning_rate": 0.0001375,
      "loss": 0.0449,
      "step": 2000
    },
    {
      "epoch": 6.25,
      "eval_accuracy": 0.94921875,
      "eval_loss": 0.20937061309814453,
      "eval_runtime": 6.6824,
      "eval_samples_per_second": 191.549,
      "eval_steps_per_second": 23.944,
      "step": 2000
    },
    {
      "epoch": 6.28125,
      "grad_norm": 0.13594718277454376,
      "learning_rate": 0.0001371875,
      "loss": 0.0127,
      "step": 2010
    },
    {
      "epoch": 6.3125,
      "grad_norm": 5.2705183029174805,
      "learning_rate": 0.000136875,
      "loss": 0.032,
      "step": 2020
    },
    {
      "epoch": 6.34375,
      "grad_norm": 11.641499519348145,
      "learning_rate": 0.00013656250000000002,
      "loss": 0.063,
      "step": 2030
    },
    {
      "epoch": 6.375,
      "grad_norm": 0.017323823645710945,
      "learning_rate": 0.00013625,
      "loss": 0.0106,
      "step": 2040
    },
    {
      "epoch": 6.40625,
      "grad_norm": 0.029373859986662865,
      "learning_rate": 0.00013593750000000002,
      "loss": 0.0032,
      "step": 2050
    },
    {
      "epoch": 6.4375,
      "grad_norm": 0.3746764063835144,
      "learning_rate": 0.000135625,
      "loss": 0.0606,
      "step": 2060
    },
    {
      "epoch": 6.46875,
      "grad_norm": 0.1948755830526352,
      "learning_rate": 0.0001353125,
      "loss": 0.0763,
      "step": 2070
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.017781907692551613,
      "learning_rate": 0.00013500000000000003,
      "loss": 0.1066,
      "step": 2080
    },
    {
      "epoch": 6.53125,
      "grad_norm": 7.0899577140808105,
      "learning_rate": 0.00013468750000000001,
      "loss": 0.0805,
      "step": 2090
    },
    {
      "epoch": 6.5625,
      "grad_norm": 0.2712390720844269,
      "learning_rate": 0.000134375,
      "loss": 0.0073,
      "step": 2100
    },
    {
      "epoch": 6.5625,
      "eval_accuracy": 0.95390625,
      "eval_loss": 0.18335095047950745,
      "eval_runtime": 7.1604,
      "eval_samples_per_second": 178.761,
      "eval_steps_per_second": 22.345,
      "step": 2100
    },
    {
      "epoch": 6.59375,
      "grad_norm": 18.94267463684082,
      "learning_rate": 0.0001340625,
      "loss": 0.0978,
      "step": 2110
    },
    {
      "epoch": 6.625,
      "grad_norm": 12.507984161376953,
      "learning_rate": 0.00013375,
      "loss": 0.1564,
      "step": 2120
    },
    {
      "epoch": 6.65625,
      "grad_norm": 0.11404982954263687,
      "learning_rate": 0.0001334375,
      "loss": 0.2087,
      "step": 2130
    },
    {
      "epoch": 6.6875,
      "grad_norm": 5.735910415649414,
      "learning_rate": 0.00013312500000000002,
      "loss": 0.1009,
      "step": 2140
    },
    {
      "epoch": 6.71875,
      "grad_norm": 0.24378864467144012,
      "learning_rate": 0.0001328125,
      "loss": 0.117,
      "step": 2150
    },
    {
      "epoch": 6.75,
      "grad_norm": 1.378218412399292,
      "learning_rate": 0.0001325,
      "loss": 0.0559,
      "step": 2160
    },
    {
      "epoch": 6.78125,
      "grad_norm": 0.11509452760219574,
      "learning_rate": 0.0001321875,
      "loss": 0.1116,
      "step": 2170
    },
    {
      "epoch": 6.8125,
      "grad_norm": 0.024781059473752975,
      "learning_rate": 0.00013187500000000002,
      "loss": 0.0162,
      "step": 2180
    },
    {
      "epoch": 6.84375,
      "grad_norm": 0.013253854587674141,
      "learning_rate": 0.0001315625,
      "loss": 0.0124,
      "step": 2190
    },
    {
      "epoch": 6.875,
      "grad_norm": 0.009435000829398632,
      "learning_rate": 0.00013125000000000002,
      "loss": 0.0425,
      "step": 2200
    },
    {
      "epoch": 6.875,
      "eval_accuracy": 0.9265625,
      "eval_loss": 0.2847265601158142,
      "eval_runtime": 6.4944,
      "eval_samples_per_second": 197.092,
      "eval_steps_per_second": 24.637,
      "step": 2200
    },
    {
      "epoch": 6.90625,
      "grad_norm": 0.009621995501220226,
      "learning_rate": 0.0001309375,
      "loss": 0.0092,
      "step": 2210
    },
    {
      "epoch": 6.9375,
      "grad_norm": 13.595647811889648,
      "learning_rate": 0.000130625,
      "loss": 0.033,
      "step": 2220
    },
    {
      "epoch": 6.96875,
      "grad_norm": 0.046547386795282364,
      "learning_rate": 0.0001303125,
      "loss": 0.0697,
      "step": 2230
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.018974941223859787,
      "learning_rate": 0.00013000000000000002,
      "loss": 0.0325,
      "step": 2240
    },
    {
      "epoch": 7.03125,
      "grad_norm": 0.011768043972551823,
      "learning_rate": 0.0001296875,
      "loss": 0.1392,
      "step": 2250
    },
    {
      "epoch": 7.0625,
      "grad_norm": 0.10145868360996246,
      "learning_rate": 0.00012937500000000001,
      "loss": 0.0034,
      "step": 2260
    },
    {
      "epoch": 7.09375,
      "grad_norm": 2.6255838871002197,
      "learning_rate": 0.0001290625,
      "loss": 0.0205,
      "step": 2270
    },
    {
      "epoch": 7.125,
      "grad_norm": 7.713711261749268,
      "learning_rate": 0.00012875,
      "loss": 0.085,
      "step": 2280
    },
    {
      "epoch": 7.15625,
      "grad_norm": 0.015717368572950363,
      "learning_rate": 0.0001284375,
      "loss": 0.073,
      "step": 2290
    },
    {
      "epoch": 7.1875,
      "grad_norm": 0.023385796695947647,
      "learning_rate": 0.000128125,
      "loss": 0.0397,
      "step": 2300
    },
    {
      "epoch": 7.1875,
      "eval_accuracy": 0.9125,
      "eval_loss": 0.40313416719436646,
      "eval_runtime": 7.0181,
      "eval_samples_per_second": 182.385,
      "eval_steps_per_second": 22.798,
      "step": 2300
    },
    {
      "epoch": 7.21875,
      "grad_norm": 0.05634606257081032,
      "learning_rate": 0.0001278125,
      "loss": 0.0661,
      "step": 2310
    },
    {
      "epoch": 7.25,
      "grad_norm": 0.0715768113732338,
      "learning_rate": 0.0001275,
      "loss": 0.0194,
      "step": 2320
    },
    {
      "epoch": 7.28125,
      "grad_norm": 0.03949156031012535,
      "learning_rate": 0.00012718750000000002,
      "loss": 0.0039,
      "step": 2330
    },
    {
      "epoch": 7.3125,
      "grad_norm": 0.011410553939640522,
      "learning_rate": 0.000126875,
      "loss": 0.0305,
      "step": 2340
    },
    {
      "epoch": 7.34375,
      "grad_norm": 0.00789843499660492,
      "learning_rate": 0.0001265625,
      "loss": 0.0022,
      "step": 2350
    },
    {
      "epoch": 7.375,
      "grad_norm": 0.6067203283309937,
      "learning_rate": 0.00012625,
      "loss": 0.0022,
      "step": 2360
    },
    {
      "epoch": 7.40625,
      "grad_norm": 3.3476336002349854,
      "learning_rate": 0.0001259375,
      "loss": 0.0642,
      "step": 2370
    },
    {
      "epoch": 7.4375,
      "grad_norm": 0.017286688089370728,
      "learning_rate": 0.000125625,
      "loss": 0.0081,
      "step": 2380
    },
    {
      "epoch": 7.46875,
      "grad_norm": 0.007009466178715229,
      "learning_rate": 0.00012531250000000002,
      "loss": 0.0308,
      "step": 2390
    },
    {
      "epoch": 7.5,
      "grad_norm": 5.178075313568115,
      "learning_rate": 0.000125,
      "loss": 0.0284,
      "step": 2400
    },
    {
      "epoch": 7.5,
      "eval_accuracy": 0.940625,
      "eval_loss": 0.29945191740989685,
      "eval_runtime": 6.4921,
      "eval_samples_per_second": 197.161,
      "eval_steps_per_second": 24.645,
      "step": 2400
    },
    {
      "epoch": 7.53125,
      "grad_norm": 0.03435547277331352,
      "learning_rate": 0.0001246875,
      "loss": 0.0786,
      "step": 2410
    },
    {
      "epoch": 7.5625,
      "grad_norm": 0.2063717395067215,
      "learning_rate": 0.000124375,
      "loss": 0.0575,
      "step": 2420
    },
    {
      "epoch": 7.59375,
      "grad_norm": 0.021375322714447975,
      "learning_rate": 0.00012406250000000001,
      "loss": 0.0017,
      "step": 2430
    },
    {
      "epoch": 7.625,
      "grad_norm": 0.010769976302981377,
      "learning_rate": 0.00012375,
      "loss": 0.0437,
      "step": 2440
    },
    {
      "epoch": 7.65625,
      "grad_norm": 0.01713966764509678,
      "learning_rate": 0.0001234375,
      "loss": 0.0018,
      "step": 2450
    },
    {
      "epoch": 7.6875,
      "grad_norm": 0.007944900542497635,
      "learning_rate": 0.000123125,
      "loss": 0.011,
      "step": 2460
    },
    {
      "epoch": 7.71875,
      "grad_norm": 0.041198715567588806,
      "learning_rate": 0.0001228125,
      "loss": 0.0019,
      "step": 2470
    },
    {
      "epoch": 7.75,
      "grad_norm": 0.009142044000327587,
      "learning_rate": 0.00012250000000000002,
      "loss": 0.0287,
      "step": 2480
    },
    {
      "epoch": 7.78125,
      "grad_norm": 0.00957415159791708,
      "learning_rate": 0.0001221875,
      "loss": 0.0015,
      "step": 2490
    },
    {
      "epoch": 7.8125,
      "grad_norm": 0.018495427444577217,
      "learning_rate": 0.00012187500000000001,
      "loss": 0.0158,
      "step": 2500
    },
    {
      "epoch": 7.8125,
      "eval_accuracy": 0.96640625,
      "eval_loss": 0.19092732667922974,
      "eval_runtime": 6.9695,
      "eval_samples_per_second": 183.658,
      "eval_steps_per_second": 22.957,
      "step": 2500
    },
    {
      "epoch": 7.84375,
      "grad_norm": 0.005418677814304829,
      "learning_rate": 0.00012156250000000001,
      "loss": 0.0196,
      "step": 2510
    },
    {
      "epoch": 7.875,
      "grad_norm": 0.30978670716285706,
      "learning_rate": 0.00012124999999999999,
      "loss": 0.0327,
      "step": 2520
    },
    {
      "epoch": 7.90625,
      "grad_norm": 0.006560006178915501,
      "learning_rate": 0.00012093750000000002,
      "loss": 0.0079,
      "step": 2530
    },
    {
      "epoch": 7.9375,
      "grad_norm": 0.006091834977269173,
      "learning_rate": 0.000120625,
      "loss": 0.0086,
      "step": 2540
    },
    {
      "epoch": 7.96875,
      "grad_norm": 4.461112022399902,
      "learning_rate": 0.0001203125,
      "loss": 0.0603,
      "step": 2550
    },
    {
      "epoch": 8.0,
      "grad_norm": 18.875268936157227,
      "learning_rate": 0.00012,
      "loss": 0.0695,
      "step": 2560
    },
    {
      "epoch": 8.03125,
      "grad_norm": 0.006029163487255573,
      "learning_rate": 0.0001196875,
      "loss": 0.0015,
      "step": 2570
    },
    {
      "epoch": 8.0625,
      "grad_norm": 0.00949636660516262,
      "learning_rate": 0.00011937500000000001,
      "loss": 0.0029,
      "step": 2580
    },
    {
      "epoch": 8.09375,
      "grad_norm": 0.011131886392831802,
      "learning_rate": 0.00011906250000000001,
      "loss": 0.0037,
      "step": 2590
    },
    {
      "epoch": 8.125,
      "grad_norm": 0.43485090136528015,
      "learning_rate": 0.00011875,
      "loss": 0.006,
      "step": 2600
    },
    {
      "epoch": 8.125,
      "eval_accuracy": 0.9296875,
      "eval_loss": 0.35239508748054504,
      "eval_runtime": 6.7581,
      "eval_samples_per_second": 189.401,
      "eval_steps_per_second": 23.675,
      "step": 2600
    },
    {
      "epoch": 8.15625,
      "grad_norm": 0.004519434180110693,
      "learning_rate": 0.0001184375,
      "loss": 0.0774,
      "step": 2610
    },
    {
      "epoch": 8.1875,
      "grad_norm": 6.793065547943115,
      "learning_rate": 0.000118125,
      "loss": 0.0311,
      "step": 2620
    },
    {
      "epoch": 8.21875,
      "grad_norm": 0.0822053775191307,
      "learning_rate": 0.00011781250000000001,
      "loss": 0.0164,
      "step": 2630
    },
    {
      "epoch": 8.25,
      "grad_norm": 0.0055194334127008915,
      "learning_rate": 0.00011750000000000001,
      "loss": 0.0019,
      "step": 2640
    },
    {
      "epoch": 8.28125,
      "grad_norm": 0.016379250213503838,
      "learning_rate": 0.00011718750000000001,
      "loss": 0.0206,
      "step": 2650
    },
    {
      "epoch": 8.3125,
      "grad_norm": 0.00409812992438674,
      "learning_rate": 0.000116875,
      "loss": 0.0017,
      "step": 2660
    },
    {
      "epoch": 8.34375,
      "grad_norm": 0.004736763890832663,
      "learning_rate": 0.0001165625,
      "loss": 0.0173,
      "step": 2670
    },
    {
      "epoch": 8.375,
      "grad_norm": 0.006192313041538,
      "learning_rate": 0.00011625000000000002,
      "loss": 0.0059,
      "step": 2680
    },
    {
      "epoch": 8.40625,
      "grad_norm": 0.019787801429629326,
      "learning_rate": 0.0001159375,
      "loss": 0.0559,
      "step": 2690
    },
    {
      "epoch": 8.4375,
      "grad_norm": 0.945681095123291,
      "learning_rate": 0.000115625,
      "loss": 0.0017,
      "step": 2700
    },
    {
      "epoch": 8.4375,
      "eval_accuracy": 0.96171875,
      "eval_loss": 0.19076545536518097,
      "eval_runtime": 6.3936,
      "eval_samples_per_second": 200.199,
      "eval_steps_per_second": 25.025,
      "step": 2700
    },
    {
      "epoch": 8.46875,
      "grad_norm": 0.3741857409477234,
      "learning_rate": 0.0001153125,
      "loss": 0.001,
      "step": 2710
    },
    {
      "epoch": 8.5,
      "grad_norm": 0.0038401270285248756,
      "learning_rate": 0.00011499999999999999,
      "loss": 0.0011,
      "step": 2720
    },
    {
      "epoch": 8.53125,
      "grad_norm": 0.005083560012280941,
      "learning_rate": 0.00011468750000000002,
      "loss": 0.0008,
      "step": 2730
    },
    {
      "epoch": 8.5625,
      "grad_norm": 0.0035322927869856358,
      "learning_rate": 0.00011437500000000002,
      "loss": 0.0183,
      "step": 2740
    },
    {
      "epoch": 8.59375,
      "grad_norm": 0.005947966128587723,
      "learning_rate": 0.0001140625,
      "loss": 0.0011,
      "step": 2750
    },
    {
      "epoch": 8.625,
      "grad_norm": 5.917221546173096,
      "learning_rate": 0.00011375,
      "loss": 0.0021,
      "step": 2760
    },
    {
      "epoch": 8.65625,
      "grad_norm": 0.0036874141078442335,
      "learning_rate": 0.0001134375,
      "loss": 0.0186,
      "step": 2770
    },
    {
      "epoch": 8.6875,
      "grad_norm": 0.005372929852455854,
      "learning_rate": 0.00011312500000000001,
      "loss": 0.0241,
      "step": 2780
    },
    {
      "epoch": 8.71875,
      "grad_norm": 0.004914106801152229,
      "learning_rate": 0.00011281250000000001,
      "loss": 0.0421,
      "step": 2790
    },
    {
      "epoch": 8.75,
      "grad_norm": 0.02046949602663517,
      "learning_rate": 0.00011250000000000001,
      "loss": 0.0026,
      "step": 2800
    },
    {
      "epoch": 8.75,
      "eval_accuracy": 0.9625,
      "eval_loss": 0.17868757247924805,
      "eval_runtime": 6.3446,
      "eval_samples_per_second": 201.746,
      "eval_steps_per_second": 25.218,
      "step": 2800
    },
    {
      "epoch": 8.78125,
      "grad_norm": 0.023118741810321808,
      "learning_rate": 0.0001121875,
      "loss": 0.0016,
      "step": 2810
    },
    {
      "epoch": 8.8125,
      "grad_norm": 16.51854133605957,
      "learning_rate": 0.000111875,
      "loss": 0.0685,
      "step": 2820
    },
    {
      "epoch": 8.84375,
      "grad_norm": 0.011586804874241352,
      "learning_rate": 0.00011156250000000001,
      "loss": 0.0323,
      "step": 2830
    },
    {
      "epoch": 8.875,
      "grad_norm": 0.018752580508589745,
      "learning_rate": 0.00011125000000000001,
      "loss": 0.0012,
      "step": 2840
    },
    {
      "epoch": 8.90625,
      "grad_norm": 0.014984137378633022,
      "learning_rate": 0.0001109375,
      "loss": 0.0288,
      "step": 2850
    },
    {
      "epoch": 8.9375,
      "grad_norm": 12.684847831726074,
      "learning_rate": 0.00011065625,
      "loss": 0.1146,
      "step": 2860
    },
    {
      "epoch": 8.96875,
      "grad_norm": 0.09825449436903,
      "learning_rate": 0.00011034375000000001,
      "loss": 0.0147,
      "step": 2870
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.011579761281609535,
      "learning_rate": 0.00011003125000000001,
      "loss": 0.0012,
      "step": 2880
    },
    {
      "epoch": 9.03125,
      "grad_norm": 0.009486192837357521,
      "learning_rate": 0.00010971875000000001,
      "loss": 0.0257,
      "step": 2890
    },
    {
      "epoch": 9.0625,
      "grad_norm": 0.005146791692823172,
      "learning_rate": 0.00010940624999999999,
      "loss": 0.001,
      "step": 2900
    },
    {
      "epoch": 9.0625,
      "eval_accuracy": 0.96875,
      "eval_loss": 0.1328631341457367,
      "eval_runtime": 6.918,
      "eval_samples_per_second": 185.024,
      "eval_steps_per_second": 23.128,
      "step": 2900
    },
    {
      "epoch": 9.09375,
      "grad_norm": 0.005191614385694265,
      "learning_rate": 0.00010909374999999999,
      "loss": 0.0013,
      "step": 2910
    },
    {
      "epoch": 9.125,
      "grad_norm": 0.024972369894385338,
      "learning_rate": 0.00010878125000000002,
      "loss": 0.0356,
      "step": 2920
    },
    {
      "epoch": 9.15625,
      "grad_norm": 0.00825554970651865,
      "learning_rate": 0.00010846875,
      "loss": 0.0009,
      "step": 2930
    },
    {
      "epoch": 9.1875,
      "grad_norm": 0.004267066717147827,
      "learning_rate": 0.00010815625,
      "loss": 0.0395,
      "step": 2940
    },
    {
      "epoch": 9.21875,
      "grad_norm": 0.7409490346908569,
      "learning_rate": 0.00010784375,
      "loss": 0.0037,
      "step": 2950
    },
    {
      "epoch": 9.25,
      "grad_norm": 0.006523266900330782,
      "learning_rate": 0.00010753124999999999,
      "loss": 0.0205,
      "step": 2960
    },
    {
      "epoch": 9.28125,
      "grad_norm": 0.00338526233099401,
      "learning_rate": 0.00010721875000000001,
      "loss": 0.0185,
      "step": 2970
    },
    {
      "epoch": 9.3125,
      "grad_norm": 0.00550084188580513,
      "learning_rate": 0.00010690625000000001,
      "loss": 0.0135,
      "step": 2980
    },
    {
      "epoch": 9.34375,
      "grad_norm": 0.0031597930938005447,
      "learning_rate": 0.00010659375,
      "loss": 0.0012,
      "step": 2990
    },
    {
      "epoch": 9.375,
      "grad_norm": 0.004799798596650362,
      "learning_rate": 0.00010628125,
      "loss": 0.0497,
      "step": 3000
    },
    {
      "epoch": 9.375,
      "eval_accuracy": 0.959375,
      "eval_loss": 0.1878364086151123,
      "eval_runtime": 6.9355,
      "eval_samples_per_second": 184.558,
      "eval_steps_per_second": 23.07,
      "step": 3000
    },
    {
      "epoch": 9.40625,
      "grad_norm": 0.03460833802819252,
      "learning_rate": 0.00010596875,
      "loss": 0.002,
      "step": 3010
    },
    {
      "epoch": 9.4375,
      "grad_norm": 2.7818896770477295,
      "learning_rate": 0.00010565625000000001,
      "loss": 0.0333,
      "step": 3020
    },
    {
      "epoch": 9.46875,
      "grad_norm": 0.00796230137348175,
      "learning_rate": 0.00010534375000000001,
      "loss": 0.041,
      "step": 3030
    },
    {
      "epoch": 9.5,
      "grad_norm": 0.0041164797730743885,
      "learning_rate": 0.00010503125000000001,
      "loss": 0.071,
      "step": 3040
    },
    {
      "epoch": 9.53125,
      "grad_norm": 0.004923259373754263,
      "learning_rate": 0.00010471875,
      "loss": 0.0016,
      "step": 3050
    },
    {
      "epoch": 9.5625,
      "grad_norm": 0.03983521834015846,
      "learning_rate": 0.00010440625,
      "loss": 0.0245,
      "step": 3060
    },
    {
      "epoch": 9.59375,
      "grad_norm": 0.005190184339880943,
      "learning_rate": 0.00010409375,
      "loss": 0.0134,
      "step": 3070
    },
    {
      "epoch": 9.625,
      "grad_norm": 0.05828193947672844,
      "learning_rate": 0.00010378125,
      "loss": 0.0599,
      "step": 3080
    },
    {
      "epoch": 9.65625,
      "grad_norm": 0.0039481050334870815,
      "learning_rate": 0.00010346875,
      "loss": 0.0581,
      "step": 3090
    },
    {
      "epoch": 9.6875,
      "grad_norm": 0.007265688385814428,
      "learning_rate": 0.00010315625,
      "loss": 0.09,
      "step": 3100
    },
    {
      "epoch": 9.6875,
      "eval_accuracy": 0.96484375,
      "eval_loss": 0.1753551959991455,
      "eval_runtime": 6.4623,
      "eval_samples_per_second": 198.073,
      "eval_steps_per_second": 24.759,
      "step": 3100
    },
    {
      "epoch": 9.71875,
      "grad_norm": 0.009442336857318878,
      "learning_rate": 0.00010284374999999999,
      "loss": 0.0009,
      "step": 3110
    },
    {
      "epoch": 9.75,
      "grad_norm": 0.0029557254165410995,
      "learning_rate": 0.00010253125000000002,
      "loss": 0.0011,
      "step": 3120
    },
    {
      "epoch": 9.78125,
      "grad_norm": 0.6488510370254517,
      "learning_rate": 0.00010221875,
      "loss": 0.0207,
      "step": 3130
    },
    {
      "epoch": 9.8125,
      "grad_norm": 0.008899745531380177,
      "learning_rate": 0.00010190625,
      "loss": 0.0106,
      "step": 3140
    },
    {
      "epoch": 9.84375,
      "grad_norm": 0.002760515082627535,
      "learning_rate": 0.00010159375,
      "loss": 0.0172,
      "step": 3150
    },
    {
      "epoch": 9.875,
      "grad_norm": 0.9725448489189148,
      "learning_rate": 0.00010128125,
      "loss": 0.0065,
      "step": 3160
    },
    {
      "epoch": 9.90625,
      "grad_norm": 0.002948402427136898,
      "learning_rate": 0.00010096875000000001,
      "loss": 0.0008,
      "step": 3170
    },
    {
      "epoch": 9.9375,
      "grad_norm": 0.0027872335631400347,
      "learning_rate": 0.00010065625000000001,
      "loss": 0.0023,
      "step": 3180
    },
    {
      "epoch": 9.96875,
      "grad_norm": 0.00609046733006835,
      "learning_rate": 0.00010034375000000001,
      "loss": 0.0385,
      "step": 3190
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.025308266282081604,
      "learning_rate": 0.00010003125,
      "loss": 0.0046,
      "step": 3200
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9671875,
      "eval_loss": 0.15844720602035522,
      "eval_runtime": 6.5917,
      "eval_samples_per_second": 194.184,
      "eval_steps_per_second": 24.273,
      "step": 3200
    },
    {
      "epoch": 10.03125,
      "grad_norm": 0.0036033540964126587,
      "learning_rate": 9.971875000000001e-05,
      "loss": 0.0007,
      "step": 3210
    },
    {
      "epoch": 10.0625,
      "grad_norm": 0.005487027112394571,
      "learning_rate": 9.940625000000001e-05,
      "loss": 0.0893,
      "step": 3220
    },
    {
      "epoch": 10.09375,
      "grad_norm": 0.017159543931484222,
      "learning_rate": 9.909375e-05,
      "loss": 0.003,
      "step": 3230
    },
    {
      "epoch": 10.125,
      "grad_norm": 0.006799894850701094,
      "learning_rate": 9.878125e-05,
      "loss": 0.0026,
      "step": 3240
    },
    {
      "epoch": 10.15625,
      "grad_norm": 5.967584609985352,
      "learning_rate": 9.846875e-05,
      "loss": 0.035,
      "step": 3250
    },
    {
      "epoch": 10.1875,
      "grad_norm": 0.027164561673998833,
      "learning_rate": 9.815625e-05,
      "loss": 0.0187,
      "step": 3260
    },
    {
      "epoch": 10.21875,
      "grad_norm": 0.010049635544419289,
      "learning_rate": 9.784375e-05,
      "loss": 0.0012,
      "step": 3270
    },
    {
      "epoch": 10.25,
      "grad_norm": 0.006957762409001589,
      "learning_rate": 9.753125e-05,
      "loss": 0.0009,
      "step": 3280
    },
    {
      "epoch": 10.28125,
      "grad_norm": 0.009408445097506046,
      "learning_rate": 9.721875e-05,
      "loss": 0.0016,
      "step": 3290
    },
    {
      "epoch": 10.3125,
      "grad_norm": 0.004588890355080366,
      "learning_rate": 9.690625000000001e-05,
      "loss": 0.0006,
      "step": 3300
    },
    {
      "epoch": 10.3125,
      "eval_accuracy": 0.96484375,
      "eval_loss": 0.20075881481170654,
      "eval_runtime": 6.9603,
      "eval_samples_per_second": 183.901,
      "eval_steps_per_second": 22.988,
      "step": 3300
    },
    {
      "epoch": 10.34375,
      "grad_norm": 0.0026115712244063616,
      "learning_rate": 9.659375e-05,
      "loss": 0.0008,
      "step": 3310
    },
    {
      "epoch": 10.375,
      "grad_norm": 0.010643471032381058,
      "learning_rate": 9.628125e-05,
      "loss": 0.0424,
      "step": 3320
    },
    {
      "epoch": 10.40625,
      "grad_norm": 23.894906997680664,
      "learning_rate": 9.596875000000001e-05,
      "loss": 0.0637,
      "step": 3330
    },
    {
      "epoch": 10.4375,
      "grad_norm": 0.00423228507861495,
      "learning_rate": 9.565625e-05,
      "loss": 0.0012,
      "step": 3340
    },
    {
      "epoch": 10.46875,
      "grad_norm": 0.0030155859421938658,
      "learning_rate": 9.534375000000001e-05,
      "loss": 0.0186,
      "step": 3350
    },
    {
      "epoch": 10.5,
      "grad_norm": 0.003306414932012558,
      "learning_rate": 9.503125000000001e-05,
      "loss": 0.0363,
      "step": 3360
    },
    {
      "epoch": 10.53125,
      "grad_norm": 0.0035443324595689774,
      "learning_rate": 9.471875e-05,
      "loss": 0.0269,
      "step": 3370
    },
    {
      "epoch": 10.5625,
      "grad_norm": 0.00464298902079463,
      "learning_rate": 9.440625000000001e-05,
      "loss": 0.0006,
      "step": 3380
    },
    {
      "epoch": 10.59375,
      "grad_norm": 0.003652532584965229,
      "learning_rate": 9.409375000000001e-05,
      "loss": 0.0006,
      "step": 3390
    },
    {
      "epoch": 10.625,
      "grad_norm": 0.0041074915789067745,
      "learning_rate": 9.378125e-05,
      "loss": 0.0008,
      "step": 3400
    },
    {
      "epoch": 10.625,
      "eval_accuracy": 0.975,
      "eval_loss": 0.12715043127536774,
      "eval_runtime": 6.4535,
      "eval_samples_per_second": 198.341,
      "eval_steps_per_second": 24.793,
      "step": 3400
    },
    {
      "epoch": 10.65625,
      "grad_norm": 0.003479533363133669,
      "learning_rate": 9.346875e-05,
      "loss": 0.0022,
      "step": 3410
    },
    {
      "epoch": 10.6875,
      "grad_norm": 0.002636971650645137,
      "learning_rate": 9.315625e-05,
      "loss": 0.0264,
      "step": 3420
    },
    {
      "epoch": 10.71875,
      "grad_norm": 0.0028728495817631483,
      "learning_rate": 9.284375e-05,
      "loss": 0.0121,
      "step": 3430
    },
    {
      "epoch": 10.75,
      "grad_norm": 0.0024031987413764,
      "learning_rate": 9.253125e-05,
      "loss": 0.0147,
      "step": 3440
    },
    {
      "epoch": 10.78125,
      "grad_norm": 11.363987922668457,
      "learning_rate": 9.221875000000002e-05,
      "loss": 0.0628,
      "step": 3450
    },
    {
      "epoch": 10.8125,
      "grad_norm": 0.01323883980512619,
      "learning_rate": 9.190625e-05,
      "loss": 0.0016,
      "step": 3460
    },
    {
      "epoch": 10.84375,
      "grad_norm": 0.005075179971754551,
      "learning_rate": 9.159375e-05,
      "loss": 0.0298,
      "step": 3470
    },
    {
      "epoch": 10.875,
      "grad_norm": 0.03815023973584175,
      "learning_rate": 9.128125000000001e-05,
      "loss": 0.0006,
      "step": 3480
    },
    {
      "epoch": 10.90625,
      "grad_norm": 0.028709406033158302,
      "learning_rate": 9.096875e-05,
      "loss": 0.0165,
      "step": 3490
    },
    {
      "epoch": 10.9375,
      "grad_norm": 0.002887293929234147,
      "learning_rate": 9.065625000000001e-05,
      "loss": 0.028,
      "step": 3500
    },
    {
      "epoch": 10.9375,
      "eval_accuracy": 0.9765625,
      "eval_loss": 0.14528806507587433,
      "eval_runtime": 6.3552,
      "eval_samples_per_second": 201.41,
      "eval_steps_per_second": 25.176,
      "step": 3500
    },
    {
      "epoch": 10.96875,
      "grad_norm": 0.0032660234719514847,
      "learning_rate": 9.034375000000001e-05,
      "loss": 0.0008,
      "step": 3510
    },
    {
      "epoch": 11.0,
      "grad_norm": 0.003932199906557798,
      "learning_rate": 9.003125e-05,
      "loss": 0.0327,
      "step": 3520
    },
    {
      "epoch": 11.03125,
      "grad_norm": 0.0021477933041751385,
      "learning_rate": 8.971875000000001e-05,
      "loss": 0.0005,
      "step": 3530
    },
    {
      "epoch": 11.0625,
      "grad_norm": 0.002401479985564947,
      "learning_rate": 8.940625000000001e-05,
      "loss": 0.0306,
      "step": 3540
    },
    {
      "epoch": 11.09375,
      "grad_norm": 8.878641128540039,
      "learning_rate": 8.909375000000001e-05,
      "loss": 0.0523,
      "step": 3550
    },
    {
      "epoch": 11.125,
      "grad_norm": 0.0029539538081735373,
      "learning_rate": 8.878125e-05,
      "loss": 0.0006,
      "step": 3560
    },
    {
      "epoch": 11.15625,
      "grad_norm": 0.004306386224925518,
      "learning_rate": 8.846875e-05,
      "loss": 0.001,
      "step": 3570
    },
    {
      "epoch": 11.1875,
      "grad_norm": 0.013514366932213306,
      "learning_rate": 8.815625e-05,
      "loss": 0.0005,
      "step": 3580
    },
    {
      "epoch": 11.21875,
      "grad_norm": 0.013187670148909092,
      "learning_rate": 8.784375e-05,
      "loss": 0.0005,
      "step": 3590
    },
    {
      "epoch": 11.25,
      "grad_norm": 0.002293806755915284,
      "learning_rate": 8.753125e-05,
      "loss": 0.0005,
      "step": 3600
    },
    {
      "epoch": 11.25,
      "eval_accuracy": 0.975,
      "eval_loss": 0.1256314218044281,
      "eval_runtime": 6.8185,
      "eval_samples_per_second": 187.725,
      "eval_steps_per_second": 23.466,
      "step": 3600
    },
    {
      "epoch": 11.28125,
      "grad_norm": 0.0030747319106012583,
      "learning_rate": 8.721875e-05,
      "loss": 0.0213,
      "step": 3610
    },
    {
      "epoch": 11.3125,
      "grad_norm": 0.002191092586144805,
      "learning_rate": 8.690625e-05,
      "loss": 0.0014,
      "step": 3620
    },
    {
      "epoch": 11.34375,
      "grad_norm": 15.502256393432617,
      "learning_rate": 8.659375e-05,
      "loss": 0.0188,
      "step": 3630
    },
    {
      "epoch": 11.375,
      "grad_norm": 0.002841190667822957,
      "learning_rate": 8.628125e-05,
      "loss": 0.0006,
      "step": 3640
    },
    {
      "epoch": 11.40625,
      "grad_norm": 0.009304801933467388,
      "learning_rate": 8.596875000000001e-05,
      "loss": 0.0005,
      "step": 3650
    },
    {
      "epoch": 11.4375,
      "grad_norm": 0.002438412746414542,
      "learning_rate": 8.565625e-05,
      "loss": 0.0004,
      "step": 3660
    },
    {
      "epoch": 11.46875,
      "grad_norm": 0.0025553186424076557,
      "learning_rate": 8.534375e-05,
      "loss": 0.0005,
      "step": 3670
    },
    {
      "epoch": 11.5,
      "grad_norm": 0.007872486487030983,
      "learning_rate": 8.503125000000001e-05,
      "loss": 0.0004,
      "step": 3680
    },
    {
      "epoch": 11.53125,
      "grad_norm": 0.003905347315594554,
      "learning_rate": 8.471875e-05,
      "loss": 0.0004,
      "step": 3690
    },
    {
      "epoch": 11.5625,
      "grad_norm": 0.0032572925556451082,
      "learning_rate": 8.440625000000001e-05,
      "loss": 0.0005,
      "step": 3700
    },
    {
      "epoch": 11.5625,
      "eval_accuracy": 0.97890625,
      "eval_loss": 0.10886295884847641,
      "eval_runtime": 6.8932,
      "eval_samples_per_second": 185.689,
      "eval_steps_per_second": 23.211,
      "step": 3700
    },
    {
      "epoch": 11.59375,
      "grad_norm": 0.005474507808685303,
      "learning_rate": 8.409375000000001e-05,
      "loss": 0.0004,
      "step": 3710
    },
    {
      "epoch": 11.625,
      "grad_norm": 0.003425732720643282,
      "learning_rate": 8.378125e-05,
      "loss": 0.0005,
      "step": 3720
    },
    {
      "epoch": 11.65625,
      "grad_norm": 0.010524573735892773,
      "learning_rate": 8.346875e-05,
      "loss": 0.0004,
      "step": 3730
    },
    {
      "epoch": 11.6875,
      "grad_norm": 0.002859619678929448,
      "learning_rate": 8.315625e-05,
      "loss": 0.0004,
      "step": 3740
    },
    {
      "epoch": 11.71875,
      "grad_norm": 0.0019814125262200832,
      "learning_rate": 8.284375e-05,
      "loss": 0.0003,
      "step": 3750
    },
    {
      "epoch": 11.75,
      "grad_norm": 0.0019714718218892813,
      "learning_rate": 8.253125e-05,
      "loss": 0.0004,
      "step": 3760
    },
    {
      "epoch": 11.78125,
      "grad_norm": 0.0027500391006469727,
      "learning_rate": 8.221875e-05,
      "loss": 0.0004,
      "step": 3770
    },
    {
      "epoch": 11.8125,
      "grad_norm": 0.001707090763375163,
      "learning_rate": 8.190625e-05,
      "loss": 0.0004,
      "step": 3780
    },
    {
      "epoch": 11.84375,
      "grad_norm": 0.0019228786695748568,
      "learning_rate": 8.159375e-05,
      "loss": 0.0004,
      "step": 3790
    },
    {
      "epoch": 11.875,
      "grad_norm": 0.0019343816675245762,
      "learning_rate": 8.128125000000001e-05,
      "loss": 0.0004,
      "step": 3800
    },
    {
      "epoch": 11.875,
      "eval_accuracy": 0.978125,
      "eval_loss": 0.1097874864935875,
      "eval_runtime": 6.3764,
      "eval_samples_per_second": 200.739,
      "eval_steps_per_second": 25.092,
      "step": 3800
    },
    {
      "epoch": 11.90625,
      "grad_norm": 0.001760054030455649,
      "learning_rate": 8.096875e-05,
      "loss": 0.0003,
      "step": 3810
    },
    {
      "epoch": 11.9375,
      "grad_norm": 0.005902810953557491,
      "learning_rate": 8.065625e-05,
      "loss": 0.0024,
      "step": 3820
    },
    {
      "epoch": 11.96875,
      "grad_norm": 0.0037092138081789017,
      "learning_rate": 8.034375000000001e-05,
      "loss": 0.0464,
      "step": 3830
    },
    {
      "epoch": 12.0,
      "grad_norm": 0.016846604645252228,
      "learning_rate": 8.003125e-05,
      "loss": 0.0023,
      "step": 3840
    },
    {
      "epoch": 12.03125,
      "grad_norm": 0.0020532661583274603,
      "learning_rate": 7.971875000000001e-05,
      "loss": 0.0004,
      "step": 3850
    },
    {
      "epoch": 12.0625,
      "grad_norm": 0.005589602515101433,
      "learning_rate": 7.940625000000001e-05,
      "loss": 0.0003,
      "step": 3860
    },
    {
      "epoch": 12.09375,
      "grad_norm": 0.008196796290576458,
      "learning_rate": 7.909375e-05,
      "loss": 0.0025,
      "step": 3870
    },
    {
      "epoch": 12.125,
      "grad_norm": 0.7199010848999023,
      "learning_rate": 7.878125000000001e-05,
      "loss": 0.0313,
      "step": 3880
    },
    {
      "epoch": 12.15625,
      "grad_norm": 0.0015094159170985222,
      "learning_rate": 7.846875e-05,
      "loss": 0.0007,
      "step": 3890
    },
    {
      "epoch": 12.1875,
      "grad_norm": 0.019164152443408966,
      "learning_rate": 7.815625e-05,
      "loss": 0.0003,
      "step": 3900
    },
    {
      "epoch": 12.1875,
      "eval_accuracy": 0.9625,
      "eval_loss": 0.17790503799915314,
      "eval_runtime": 6.2858,
      "eval_samples_per_second": 203.634,
      "eval_steps_per_second": 25.454,
      "step": 3900
    },
    {
      "epoch": 12.21875,
      "grad_norm": 0.002245397539809346,
      "learning_rate": 7.784375e-05,
      "loss": 0.0009,
      "step": 3910
    },
    {
      "epoch": 12.25,
      "grad_norm": 0.0015030098147690296,
      "learning_rate": 7.753125e-05,
      "loss": 0.0003,
      "step": 3920
    },
    {
      "epoch": 12.28125,
      "grad_norm": 0.11134529858827591,
      "learning_rate": 7.721875e-05,
      "loss": 0.0005,
      "step": 3930
    },
    {
      "epoch": 12.3125,
      "grad_norm": 0.002319012302905321,
      "learning_rate": 7.690625e-05,
      "loss": 0.0332,
      "step": 3940
    },
    {
      "epoch": 12.34375,
      "grad_norm": 0.002300011459738016,
      "learning_rate": 7.659375000000002e-05,
      "loss": 0.0005,
      "step": 3950
    },
    {
      "epoch": 12.375,
      "grad_norm": 0.004005583468824625,
      "learning_rate": 7.628125e-05,
      "loss": 0.0006,
      "step": 3960
    },
    {
      "epoch": 12.40625,
      "grad_norm": 0.0016996270278468728,
      "learning_rate": 7.596875e-05,
      "loss": 0.0052,
      "step": 3970
    },
    {
      "epoch": 12.4375,
      "grad_norm": 0.0016149221919476986,
      "learning_rate": 7.565625000000001e-05,
      "loss": 0.0003,
      "step": 3980
    },
    {
      "epoch": 12.46875,
      "grad_norm": 0.004814577754586935,
      "learning_rate": 7.534375e-05,
      "loss": 0.0041,
      "step": 3990
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.010618665255606174,
      "learning_rate": 7.503125000000001e-05,
      "loss": 0.0163,
      "step": 4000
    },
    {
      "epoch": 12.5,
      "eval_accuracy": 0.95390625,
      "eval_loss": 0.25004318356513977,
      "eval_runtime": 6.3222,
      "eval_samples_per_second": 202.461,
      "eval_steps_per_second": 25.308,
      "step": 4000
    },
    {
      "epoch": 12.53125,
      "grad_norm": 0.010156257078051567,
      "learning_rate": 7.471875000000001e-05,
      "loss": 0.0349,
      "step": 4010
    },
    {
      "epoch": 12.5625,
      "grad_norm": 0.015218588523566723,
      "learning_rate": 7.440625e-05,
      "loss": 0.1117,
      "step": 4020
    },
    {
      "epoch": 12.59375,
      "grad_norm": 0.00266676745377481,
      "learning_rate": 7.409375000000001e-05,
      "loss": 0.0004,
      "step": 4030
    },
    {
      "epoch": 12.625,
      "grad_norm": 0.001848011976107955,
      "learning_rate": 7.378125000000001e-05,
      "loss": 0.001,
      "step": 4040
    },
    {
      "epoch": 12.65625,
      "grad_norm": 0.014798123389482498,
      "learning_rate": 7.346875000000001e-05,
      "loss": 0.0005,
      "step": 4050
    },
    {
      "epoch": 12.6875,
      "grad_norm": 0.0023186809848994017,
      "learning_rate": 7.315625e-05,
      "loss": 0.0371,
      "step": 4060
    },
    {
      "epoch": 12.71875,
      "grad_norm": 0.001521400292403996,
      "learning_rate": 7.284375e-05,
      "loss": 0.0036,
      "step": 4070
    },
    {
      "epoch": 12.75,
      "grad_norm": 0.0016448964597657323,
      "learning_rate": 7.253125e-05,
      "loss": 0.0003,
      "step": 4080
    },
    {
      "epoch": 12.78125,
      "grad_norm": 0.003130651544779539,
      "learning_rate": 7.221875e-05,
      "loss": 0.0004,
      "step": 4090
    },
    {
      "epoch": 12.8125,
      "grad_norm": 0.001429844181984663,
      "learning_rate": 7.190625e-05,
      "loss": 0.0003,
      "step": 4100
    },
    {
      "epoch": 12.8125,
      "eval_accuracy": 0.9734375,
      "eval_loss": 0.1555672138929367,
      "eval_runtime": 6.821,
      "eval_samples_per_second": 187.655,
      "eval_steps_per_second": 23.457,
      "step": 4100
    },
    {
      "epoch": 12.84375,
      "grad_norm": 0.0016585810808464885,
      "learning_rate": 7.159375e-05,
      "loss": 0.0097,
      "step": 4110
    },
    {
      "epoch": 12.875,
      "grad_norm": 0.0025756254326552153,
      "learning_rate": 7.128125e-05,
      "loss": 0.0003,
      "step": 4120
    },
    {
      "epoch": 12.90625,
      "grad_norm": 0.0014616530388593674,
      "learning_rate": 7.096875e-05,
      "loss": 0.0011,
      "step": 4130
    },
    {
      "epoch": 12.9375,
      "grad_norm": 0.001520119491033256,
      "learning_rate": 7.065625e-05,
      "loss": 0.0004,
      "step": 4140
    },
    {
      "epoch": 12.96875,
      "grad_norm": 0.0017162116710096598,
      "learning_rate": 7.034375000000001e-05,
      "loss": 0.0381,
      "step": 4150
    },
    {
      "epoch": 13.0,
      "grad_norm": 0.0027610217221081257,
      "learning_rate": 7.003125e-05,
      "loss": 0.0286,
      "step": 4160
    },
    {
      "epoch": 13.03125,
      "grad_norm": 0.0013741106959059834,
      "learning_rate": 6.971875e-05,
      "loss": 0.0099,
      "step": 4170
    },
    {
      "epoch": 13.0625,
      "grad_norm": 0.08995310217142105,
      "learning_rate": 6.940625000000001e-05,
      "loss": 0.0005,
      "step": 4180
    },
    {
      "epoch": 13.09375,
      "grad_norm": 0.002924903528764844,
      "learning_rate": 6.909375e-05,
      "loss": 0.0003,
      "step": 4190
    },
    {
      "epoch": 13.125,
      "grad_norm": 0.013759560883045197,
      "learning_rate": 6.878125000000001e-05,
      "loss": 0.0003,
      "step": 4200
    },
    {
      "epoch": 13.125,
      "eval_accuracy": 0.97421875,
      "eval_loss": 0.12048967182636261,
      "eval_runtime": 7.0166,
      "eval_samples_per_second": 182.425,
      "eval_steps_per_second": 22.803,
      "step": 4200
    },
    {
      "epoch": 13.15625,
      "grad_norm": 1.506090760231018,
      "learning_rate": 6.846875000000001e-05,
      "loss": 0.001,
      "step": 4210
    },
    {
      "epoch": 13.1875,
      "grad_norm": 0.001895511755719781,
      "learning_rate": 6.815624999999999e-05,
      "loss": 0.0003,
      "step": 4220
    },
    {
      "epoch": 13.21875,
      "grad_norm": 0.006083915941417217,
      "learning_rate": 6.784375e-05,
      "loss": 0.0004,
      "step": 4230
    },
    {
      "epoch": 13.25,
      "grad_norm": 19.973133087158203,
      "learning_rate": 6.753125e-05,
      "loss": 0.0203,
      "step": 4240
    },
    {
      "epoch": 13.28125,
      "grad_norm": 0.0028971272986382246,
      "learning_rate": 6.721875e-05,
      "loss": 0.0003,
      "step": 4250
    },
    {
      "epoch": 13.3125,
      "grad_norm": 0.022618748247623444,
      "learning_rate": 6.690625e-05,
      "loss": 0.0003,
      "step": 4260
    },
    {
      "epoch": 13.34375,
      "grad_norm": 0.001784573425538838,
      "learning_rate": 6.659375e-05,
      "loss": 0.0004,
      "step": 4270
    },
    {
      "epoch": 13.375,
      "grad_norm": 0.004388020373880863,
      "learning_rate": 6.628125e-05,
      "loss": 0.0003,
      "step": 4280
    },
    {
      "epoch": 13.40625,
      "grad_norm": 0.0013094112509861588,
      "learning_rate": 6.596875e-05,
      "loss": 0.0003,
      "step": 4290
    },
    {
      "epoch": 13.4375,
      "grad_norm": 0.001405878458172083,
      "learning_rate": 6.565625000000001e-05,
      "loss": 0.0002,
      "step": 4300
    },
    {
      "epoch": 13.4375,
      "eval_accuracy": 0.971875,
      "eval_loss": 0.15426388382911682,
      "eval_runtime": 6.7768,
      "eval_samples_per_second": 188.88,
      "eval_steps_per_second": 23.61,
      "step": 4300
    },
    {
      "epoch": 13.46875,
      "grad_norm": 0.001250360975973308,
      "learning_rate": 6.534375e-05,
      "loss": 0.0003,
      "step": 4310
    },
    {
      "epoch": 13.5,
      "grad_norm": 0.0014933838974684477,
      "learning_rate": 6.503125e-05,
      "loss": 0.0003,
      "step": 4320
    },
    {
      "epoch": 13.53125,
      "grad_norm": 0.0013753235107287765,
      "learning_rate": 6.471875000000001e-05,
      "loss": 0.0003,
      "step": 4330
    },
    {
      "epoch": 13.5625,
      "grad_norm": 0.0015354104107245803,
      "learning_rate": 6.440625e-05,
      "loss": 0.0004,
      "step": 4340
    },
    {
      "epoch": 13.59375,
      "grad_norm": 0.0018886132165789604,
      "learning_rate": 6.409375000000001e-05,
      "loss": 0.0145,
      "step": 4350
    },
    {
      "epoch": 13.625,
      "grad_norm": 0.0012300664093345404,
      "learning_rate": 6.378125000000001e-05,
      "loss": 0.0003,
      "step": 4360
    },
    {
      "epoch": 13.65625,
      "grad_norm": 0.0015221609501168132,
      "learning_rate": 6.346875e-05,
      "loss": 0.0248,
      "step": 4370
    },
    {
      "epoch": 13.6875,
      "grad_norm": 0.0017438657814636827,
      "learning_rate": 6.315625000000001e-05,
      "loss": 0.0003,
      "step": 4380
    },
    {
      "epoch": 13.71875,
      "grad_norm": 0.0020069123711436987,
      "learning_rate": 6.284375e-05,
      "loss": 0.0002,
      "step": 4390
    },
    {
      "epoch": 13.75,
      "grad_norm": 0.0013785591581836343,
      "learning_rate": 6.253125e-05,
      "loss": 0.0002,
      "step": 4400
    },
    {
      "epoch": 13.75,
      "eval_accuracy": 0.975,
      "eval_loss": 0.15483084321022034,
      "eval_runtime": 6.2696,
      "eval_samples_per_second": 204.159,
      "eval_steps_per_second": 25.52,
      "step": 4400
    },
    {
      "epoch": 13.78125,
      "grad_norm": 1.8585741519927979,
      "learning_rate": 6.221875e-05,
      "loss": 0.0007,
      "step": 4410
    },
    {
      "epoch": 13.8125,
      "grad_norm": 0.0011989879421889782,
      "learning_rate": 6.190625e-05,
      "loss": 0.0003,
      "step": 4420
    },
    {
      "epoch": 13.84375,
      "grad_norm": 10.47256851196289,
      "learning_rate": 6.159375e-05,
      "loss": 0.0275,
      "step": 4430
    },
    {
      "epoch": 13.875,
      "grad_norm": 1.398241639137268,
      "learning_rate": 6.128125e-05,
      "loss": 0.0011,
      "step": 4440
    },
    {
      "epoch": 13.90625,
      "grad_norm": 0.0019299676641821861,
      "learning_rate": 6.096875000000001e-05,
      "loss": 0.0002,
      "step": 4450
    },
    {
      "epoch": 13.9375,
      "grad_norm": 0.0016532372683286667,
      "learning_rate": 6.065625e-05,
      "loss": 0.0003,
      "step": 4460
    },
    {
      "epoch": 13.96875,
      "grad_norm": 0.002245826181024313,
      "learning_rate": 6.034375e-05,
      "loss": 0.0259,
      "step": 4470
    },
    {
      "epoch": 14.0,
      "grad_norm": 0.002995037008076906,
      "learning_rate": 6.0031250000000006e-05,
      "loss": 0.0003,
      "step": 4480
    },
    {
      "epoch": 14.03125,
      "grad_norm": 0.0015550617827102542,
      "learning_rate": 5.971875e-05,
      "loss": 0.0002,
      "step": 4490
    },
    {
      "epoch": 14.0625,
      "grad_norm": 0.0018337038345634937,
      "learning_rate": 5.940625000000001e-05,
      "loss": 0.0003,
      "step": 4500
    },
    {
      "epoch": 14.0625,
      "eval_accuracy": 0.975,
      "eval_loss": 0.14965741336345673,
      "eval_runtime": 6.9272,
      "eval_samples_per_second": 184.78,
      "eval_steps_per_second": 23.097,
      "step": 4500
    },
    {
      "epoch": 14.09375,
      "grad_norm": 0.0013705631718039513,
      "learning_rate": 5.909375e-05,
      "loss": 0.0002,
      "step": 4510
    },
    {
      "epoch": 14.125,
      "grad_norm": 0.0015554011333733797,
      "learning_rate": 5.8781249999999996e-05,
      "loss": 0.0003,
      "step": 4520
    },
    {
      "epoch": 14.15625,
      "grad_norm": 0.0012009346392005682,
      "learning_rate": 5.846875000000001e-05,
      "loss": 0.0002,
      "step": 4530
    },
    {
      "epoch": 14.1875,
      "grad_norm": 0.0011792039731517434,
      "learning_rate": 5.815625e-05,
      "loss": 0.0002,
      "step": 4540
    },
    {
      "epoch": 14.21875,
      "grad_norm": 9.245038032531738,
      "learning_rate": 5.784375000000001e-05,
      "loss": 0.0038,
      "step": 4550
    },
    {
      "epoch": 14.25,
      "grad_norm": 0.0013690210180357099,
      "learning_rate": 5.7531250000000006e-05,
      "loss": 0.0003,
      "step": 4560
    },
    {
      "epoch": 14.28125,
      "grad_norm": 0.0013083881931379437,
      "learning_rate": 5.721875e-05,
      "loss": 0.0012,
      "step": 4570
    },
    {
      "epoch": 14.3125,
      "grad_norm": 34.58968734741211,
      "learning_rate": 5.6906250000000004e-05,
      "loss": 0.02,
      "step": 4580
    },
    {
      "epoch": 14.34375,
      "grad_norm": 0.0016999391373246908,
      "learning_rate": 5.6593750000000003e-05,
      "loss": 0.0002,
      "step": 4590
    },
    {
      "epoch": 14.375,
      "grad_norm": 0.0012809026520699263,
      "learning_rate": 5.628125000000001e-05,
      "loss": 0.0002,
      "step": 4600
    },
    {
      "epoch": 14.375,
      "eval_accuracy": 0.9640625,
      "eval_loss": 0.23174042999744415,
      "eval_runtime": 6.8846,
      "eval_samples_per_second": 185.923,
      "eval_steps_per_second": 23.24,
      "step": 4600
    },
    {
      "epoch": 14.40625,
      "grad_norm": 0.001300526550039649,
      "learning_rate": 5.596875e-05,
      "loss": 0.0006,
      "step": 4610
    },
    {
      "epoch": 14.4375,
      "grad_norm": 0.0010748986387625337,
      "learning_rate": 5.565625e-05,
      "loss": 0.0002,
      "step": 4620
    },
    {
      "epoch": 14.46875,
      "grad_norm": 0.0012471479130908847,
      "learning_rate": 5.534375000000001e-05,
      "loss": 0.0012,
      "step": 4630
    },
    {
      "epoch": 14.5,
      "grad_norm": 0.003839879296720028,
      "learning_rate": 5.503125e-05,
      "loss": 0.0002,
      "step": 4640
    },
    {
      "epoch": 14.53125,
      "grad_norm": 0.0013317788252606988,
      "learning_rate": 5.4718750000000005e-05,
      "loss": 0.0003,
      "step": 4650
    },
    {
      "epoch": 14.5625,
      "grad_norm": 0.001501008402556181,
      "learning_rate": 5.4406250000000004e-05,
      "loss": 0.0193,
      "step": 4660
    },
    {
      "epoch": 14.59375,
      "grad_norm": 0.0018197267781943083,
      "learning_rate": 5.409375e-05,
      "loss": 0.0035,
      "step": 4670
    },
    {
      "epoch": 14.625,
      "grad_norm": 0.0025837391149252653,
      "learning_rate": 5.378125e-05,
      "loss": 0.0005,
      "step": 4680
    },
    {
      "epoch": 14.65625,
      "grad_norm": 0.0016741983126848936,
      "learning_rate": 5.346875e-05,
      "loss": 0.0004,
      "step": 4690
    },
    {
      "epoch": 14.6875,
      "grad_norm": 0.036828938871622086,
      "learning_rate": 5.315625000000001e-05,
      "loss": 0.0003,
      "step": 4700
    },
    {
      "epoch": 14.6875,
      "eval_accuracy": 0.978125,
      "eval_loss": 0.14183716475963593,
      "eval_runtime": 6.7893,
      "eval_samples_per_second": 188.532,
      "eval_steps_per_second": 23.566,
      "step": 4700
    },
    {
      "epoch": 14.71875,
      "grad_norm": 0.0009561299229972064,
      "learning_rate": 5.284375e-05,
      "loss": 0.0002,
      "step": 4710
    },
    {
      "epoch": 14.75,
      "grad_norm": 0.001365609117783606,
      "learning_rate": 5.253125e-05,
      "loss": 0.0002,
      "step": 4720
    },
    {
      "epoch": 14.78125,
      "grad_norm": 0.0010248490143567324,
      "learning_rate": 5.2218750000000006e-05,
      "loss": 0.0002,
      "step": 4730
    },
    {
      "epoch": 14.8125,
      "grad_norm": 0.001031559775583446,
      "learning_rate": 5.190625e-05,
      "loss": 0.0002,
      "step": 4740
    },
    {
      "epoch": 14.84375,
      "grad_norm": 0.001064546755515039,
      "learning_rate": 5.159375000000001e-05,
      "loss": 0.0002,
      "step": 4750
    },
    {
      "epoch": 14.875,
      "grad_norm": 0.0011686970246955752,
      "learning_rate": 5.128125e-05,
      "loss": 0.0002,
      "step": 4760
    },
    {
      "epoch": 14.90625,
      "grad_norm": 0.0009815421653911471,
      "learning_rate": 5.0968749999999995e-05,
      "loss": 0.0002,
      "step": 4770
    },
    {
      "epoch": 14.9375,
      "grad_norm": 0.05007686838507652,
      "learning_rate": 5.065625000000001e-05,
      "loss": 0.0004,
      "step": 4780
    },
    {
      "epoch": 14.96875,
      "grad_norm": 0.001136138685978949,
      "learning_rate": 5.034375e-05,
      "loss": 0.0308,
      "step": 4790
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.0011941984994336963,
      "learning_rate": 5.0031250000000007e-05,
      "loss": 0.0002,
      "step": 4800
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9734375,
      "eval_loss": 0.15367402136325836,
      "eval_runtime": 6.8694,
      "eval_samples_per_second": 186.333,
      "eval_steps_per_second": 23.292,
      "step": 4800
    },
    {
      "epoch": 15.03125,
      "grad_norm": 0.0012780999531969428,
      "learning_rate": 4.9718750000000006e-05,
      "loss": 0.0003,
      "step": 4810
    },
    {
      "epoch": 15.0625,
      "grad_norm": 0.0019433508859947324,
      "learning_rate": 4.9406250000000005e-05,
      "loss": 0.0002,
      "step": 4820
    },
    {
      "epoch": 15.09375,
      "grad_norm": 0.0011221400927752256,
      "learning_rate": 4.9093750000000004e-05,
      "loss": 0.0002,
      "step": 4830
    },
    {
      "epoch": 15.125,
      "grad_norm": 0.0010336939012631774,
      "learning_rate": 4.878125e-05,
      "loss": 0.0002,
      "step": 4840
    },
    {
      "epoch": 15.15625,
      "grad_norm": 0.11716494709253311,
      "learning_rate": 4.846875e-05,
      "loss": 0.0002,
      "step": 4850
    },
    {
      "epoch": 15.1875,
      "grad_norm": 0.0021654649171978235,
      "learning_rate": 4.815625e-05,
      "loss": 0.0002,
      "step": 4860
    },
    {
      "epoch": 15.21875,
      "grad_norm": 0.0010919362539425492,
      "learning_rate": 4.784375e-05,
      "loss": 0.0002,
      "step": 4870
    },
    {
      "epoch": 15.25,
      "grad_norm": 0.00111663737334311,
      "learning_rate": 4.753125000000001e-05,
      "loss": 0.0002,
      "step": 4880
    },
    {
      "epoch": 15.28125,
      "grad_norm": 0.0014220753218978643,
      "learning_rate": 4.721875e-05,
      "loss": 0.0002,
      "step": 4890
    },
    {
      "epoch": 15.3125,
      "grad_norm": 0.001022492302581668,
      "learning_rate": 4.690625e-05,
      "loss": 0.0002,
      "step": 4900
    },
    {
      "epoch": 15.3125,
      "eval_accuracy": 0.978125,
      "eval_loss": 0.14259929955005646,
      "eval_runtime": 6.2156,
      "eval_samples_per_second": 205.935,
      "eval_steps_per_second": 25.742,
      "step": 4900
    },
    {
      "epoch": 15.34375,
      "grad_norm": 0.001057266490533948,
      "learning_rate": 4.6593750000000004e-05,
      "loss": 0.0002,
      "step": 4910
    },
    {
      "epoch": 15.375,
      "grad_norm": 0.001144145498983562,
      "learning_rate": 4.6281250000000003e-05,
      "loss": 0.0002,
      "step": 4920
    },
    {
      "epoch": 15.40625,
      "grad_norm": 0.0009692934690974653,
      "learning_rate": 4.596875e-05,
      "loss": 0.0002,
      "step": 4930
    },
    {
      "epoch": 15.4375,
      "grad_norm": 0.000945160398259759,
      "learning_rate": 4.565625e-05,
      "loss": 0.0002,
      "step": 4940
    },
    {
      "epoch": 15.46875,
      "grad_norm": 0.0009887829655781388,
      "learning_rate": 4.534375e-05,
      "loss": 0.0002,
      "step": 4950
    },
    {
      "epoch": 15.5,
      "grad_norm": 0.002527805743739009,
      "learning_rate": 4.503125e-05,
      "loss": 0.0002,
      "step": 4960
    },
    {
      "epoch": 15.53125,
      "grad_norm": 0.0008647911017760634,
      "learning_rate": 4.4718750000000006e-05,
      "loss": 0.0002,
      "step": 4970
    },
    {
      "epoch": 15.5625,
      "grad_norm": 0.0011423344258219004,
      "learning_rate": 4.4406250000000005e-05,
      "loss": 0.0002,
      "step": 4980
    },
    {
      "epoch": 15.59375,
      "grad_norm": 0.0022281960118561983,
      "learning_rate": 4.409375e-05,
      "loss": 0.0002,
      "step": 4990
    },
    {
      "epoch": 15.625,
      "grad_norm": 0.0011202392634004354,
      "learning_rate": 4.3781250000000004e-05,
      "loss": 0.0002,
      "step": 5000
    },
    {
      "epoch": 15.625,
      "eval_accuracy": 0.98203125,
      "eval_loss": 0.12530331313610077,
      "eval_runtime": 6.8051,
      "eval_samples_per_second": 188.096,
      "eval_steps_per_second": 23.512,
      "step": 5000
    },
    {
      "epoch": 15.65625,
      "grad_norm": 0.0008281685295514762,
      "learning_rate": 4.346875e-05,
      "loss": 0.0002,
      "step": 5010
    },
    {
      "epoch": 15.6875,
      "grad_norm": 0.001133206533268094,
      "learning_rate": 4.315625e-05,
      "loss": 0.0002,
      "step": 5020
    },
    {
      "epoch": 15.71875,
      "grad_norm": 0.000926899432670325,
      "learning_rate": 4.284375000000001e-05,
      "loss": 0.0002,
      "step": 5030
    },
    {
      "epoch": 15.75,
      "grad_norm": 0.0009977484587579966,
      "learning_rate": 4.253125e-05,
      "loss": 0.0006,
      "step": 5040
    },
    {
      "epoch": 15.78125,
      "grad_norm": 0.0011439846130087972,
      "learning_rate": 4.221875e-05,
      "loss": 0.0428,
      "step": 5050
    },
    {
      "epoch": 15.8125,
      "grad_norm": 0.0012359012616798282,
      "learning_rate": 4.1906250000000006e-05,
      "loss": 0.0002,
      "step": 5060
    },
    {
      "epoch": 15.84375,
      "grad_norm": 0.0009622338111512363,
      "learning_rate": 4.1593750000000005e-05,
      "loss": 0.0002,
      "step": 5070
    },
    {
      "epoch": 15.875,
      "grad_norm": 0.0010638950625434518,
      "learning_rate": 4.1281250000000004e-05,
      "loss": 0.0002,
      "step": 5080
    },
    {
      "epoch": 15.90625,
      "grad_norm": 0.0011404824908822775,
      "learning_rate": 4.096875e-05,
      "loss": 0.0002,
      "step": 5090
    },
    {
      "epoch": 15.9375,
      "grad_norm": 0.0012403588043525815,
      "learning_rate": 4.065625e-05,
      "loss": 0.0002,
      "step": 5100
    },
    {
      "epoch": 15.9375,
      "eval_accuracy": 0.98359375,
      "eval_loss": 0.1128150224685669,
      "eval_runtime": 6.2783,
      "eval_samples_per_second": 203.878,
      "eval_steps_per_second": 25.485,
      "step": 5100
    },
    {
      "epoch": 15.96875,
      "grad_norm": 0.001123543013818562,
      "learning_rate": 4.034375e-05,
      "loss": 0.0002,
      "step": 5110
    },
    {
      "epoch": 16.0,
      "grad_norm": 0.001036747358739376,
      "learning_rate": 4.003125e-05,
      "loss": 0.0002,
      "step": 5120
    },
    {
      "epoch": 16.03125,
      "grad_norm": 0.0013284431770443916,
      "learning_rate": 3.9718750000000007e-05,
      "loss": 0.0002,
      "step": 5130
    },
    {
      "epoch": 16.0625,
      "grad_norm": 0.0008586233016103506,
      "learning_rate": 3.940625e-05,
      "loss": 0.0002,
      "step": 5140
    },
    {
      "epoch": 16.09375,
      "grad_norm": 0.0012248932616785169,
      "learning_rate": 3.909375e-05,
      "loss": 0.0002,
      "step": 5150
    },
    {
      "epoch": 16.125,
      "grad_norm": 0.006662206724286079,
      "learning_rate": 3.8781250000000004e-05,
      "loss": 0.0002,
      "step": 5160
    },
    {
      "epoch": 16.15625,
      "grad_norm": 0.0011745645897462964,
      "learning_rate": 3.846875e-05,
      "loss": 0.0002,
      "step": 5170
    },
    {
      "epoch": 16.1875,
      "grad_norm": 0.0010919731575995684,
      "learning_rate": 3.815625e-05,
      "loss": 0.0002,
      "step": 5180
    },
    {
      "epoch": 16.21875,
      "grad_norm": 0.0010604038834571838,
      "learning_rate": 3.784375e-05,
      "loss": 0.0002,
      "step": 5190
    },
    {
      "epoch": 16.25,
      "grad_norm": 0.0011869663139805198,
      "learning_rate": 3.753125e-05,
      "loss": 0.0002,
      "step": 5200
    },
    {
      "epoch": 16.25,
      "eval_accuracy": 0.98046875,
      "eval_loss": 0.1246190294623375,
      "eval_runtime": 6.8551,
      "eval_samples_per_second": 186.722,
      "eval_steps_per_second": 23.34,
      "step": 5200
    },
    {
      "epoch": 16.28125,
      "grad_norm": 0.0010713781230151653,
      "learning_rate": 3.721875e-05,
      "loss": 0.0002,
      "step": 5210
    },
    {
      "epoch": 16.3125,
      "grad_norm": 0.0007901139324530959,
      "learning_rate": 3.6906250000000006e-05,
      "loss": 0.0002,
      "step": 5220
    },
    {
      "epoch": 16.34375,
      "grad_norm": 0.001095872139558196,
      "learning_rate": 3.6593750000000005e-05,
      "loss": 0.0002,
      "step": 5230
    },
    {
      "epoch": 16.375,
      "grad_norm": 0.0010994484182447195,
      "learning_rate": 3.628125e-05,
      "loss": 0.0002,
      "step": 5240
    },
    {
      "epoch": 16.40625,
      "grad_norm": 0.0017003176035359502,
      "learning_rate": 3.5968750000000004e-05,
      "loss": 0.0002,
      "step": 5250
    },
    {
      "epoch": 16.4375,
      "grad_norm": 0.0007925952086225152,
      "learning_rate": 3.565625e-05,
      "loss": 0.0002,
      "step": 5260
    },
    {
      "epoch": 16.46875,
      "grad_norm": 0.0008510766783729196,
      "learning_rate": 3.534375e-05,
      "loss": 0.0002,
      "step": 5270
    },
    {
      "epoch": 16.5,
      "grad_norm": 0.18106159567832947,
      "learning_rate": 3.503125e-05,
      "loss": 0.0002,
      "step": 5280
    },
    {
      "epoch": 16.53125,
      "grad_norm": 0.0008954983204603195,
      "learning_rate": 3.471875e-05,
      "loss": 0.0002,
      "step": 5290
    },
    {
      "epoch": 16.5625,
      "grad_norm": 0.0011244250927120447,
      "learning_rate": 3.440625e-05,
      "loss": 0.0002,
      "step": 5300
    },
    {
      "epoch": 16.5625,
      "eval_accuracy": 0.9828125,
      "eval_loss": 0.11365531384944916,
      "eval_runtime": 6.8269,
      "eval_samples_per_second": 187.495,
      "eval_steps_per_second": 23.437,
      "step": 5300
    },
    {
      "epoch": 16.59375,
      "grad_norm": 0.0009858476696535945,
      "learning_rate": 3.4093750000000005e-05,
      "loss": 0.0002,
      "step": 5310
    },
    {
      "epoch": 16.625,
      "grad_norm": 0.0012397761456668377,
      "learning_rate": 3.3781250000000005e-05,
      "loss": 0.0002,
      "step": 5320
    },
    {
      "epoch": 16.65625,
      "grad_norm": 0.0008871417958289385,
      "learning_rate": 3.3468750000000004e-05,
      "loss": 0.0002,
      "step": 5330
    },
    {
      "epoch": 16.6875,
      "grad_norm": 0.0007771385135129094,
      "learning_rate": 3.315625e-05,
      "loss": 0.0002,
      "step": 5340
    },
    {
      "epoch": 16.71875,
      "grad_norm": 0.0007905489183031023,
      "learning_rate": 3.284375e-05,
      "loss": 0.0002,
      "step": 5350
    },
    {
      "epoch": 16.75,
      "grad_norm": 0.0009036400006152689,
      "learning_rate": 3.253125e-05,
      "loss": 0.0001,
      "step": 5360
    },
    {
      "epoch": 16.78125,
      "grad_norm": 0.0007690058555454016,
      "learning_rate": 3.221875e-05,
      "loss": 0.0002,
      "step": 5370
    },
    {
      "epoch": 16.8125,
      "grad_norm": 0.0011280475882813334,
      "learning_rate": 3.1906250000000006e-05,
      "loss": 0.0002,
      "step": 5380
    },
    {
      "epoch": 16.84375,
      "grad_norm": 0.000766371435020119,
      "learning_rate": 3.159375e-05,
      "loss": 0.0002,
      "step": 5390
    },
    {
      "epoch": 16.875,
      "grad_norm": 0.005487964954227209,
      "learning_rate": 3.128125e-05,
      "loss": 0.0001,
      "step": 5400
    },
    {
      "epoch": 16.875,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11014194786548615,
      "eval_runtime": 6.2479,
      "eval_samples_per_second": 204.867,
      "eval_steps_per_second": 25.608,
      "step": 5400
    },
    {
      "epoch": 16.90625,
      "grad_norm": 0.0009019103599712253,
      "learning_rate": 3.0968750000000004e-05,
      "loss": 0.0002,
      "step": 5410
    },
    {
      "epoch": 16.9375,
      "grad_norm": 0.0008425983251072466,
      "learning_rate": 3.065625e-05,
      "loss": 0.0001,
      "step": 5420
    },
    {
      "epoch": 16.96875,
      "grad_norm": 0.000717274087946862,
      "learning_rate": 3.0343750000000006e-05,
      "loss": 0.0002,
      "step": 5430
    },
    {
      "epoch": 17.0,
      "grad_norm": 0.0007335466798394918,
      "learning_rate": 3.0031249999999998e-05,
      "loss": 0.0001,
      "step": 5440
    },
    {
      "epoch": 17.03125,
      "grad_norm": 0.0015981667675077915,
      "learning_rate": 2.971875e-05,
      "loss": 0.0001,
      "step": 5450
    },
    {
      "epoch": 17.0625,
      "grad_norm": 0.000765695353038609,
      "learning_rate": 2.9406250000000003e-05,
      "loss": 0.0002,
      "step": 5460
    },
    {
      "epoch": 17.09375,
      "grad_norm": 0.0017417181516066194,
      "learning_rate": 2.9093750000000002e-05,
      "loss": 0.0002,
      "step": 5470
    },
    {
      "epoch": 17.125,
      "grad_norm": 0.0010076581966131926,
      "learning_rate": 2.8781250000000005e-05,
      "loss": 0.0001,
      "step": 5480
    },
    {
      "epoch": 17.15625,
      "grad_norm": 0.0008663997869007289,
      "learning_rate": 2.846875e-05,
      "loss": 0.0002,
      "step": 5490
    },
    {
      "epoch": 17.1875,
      "grad_norm": 0.0007040807977318764,
      "learning_rate": 2.815625e-05,
      "loss": 0.0001,
      "step": 5500
    },
    {
      "epoch": 17.1875,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11123112589120865,
      "eval_runtime": 7.2338,
      "eval_samples_per_second": 176.947,
      "eval_steps_per_second": 22.118,
      "step": 5500
    },
    {
      "epoch": 17.21875,
      "grad_norm": 0.0008027940057218075,
      "learning_rate": 2.7843750000000003e-05,
      "loss": 0.0001,
      "step": 5510
    },
    {
      "epoch": 17.25,
      "grad_norm": 0.0009467356721870601,
      "learning_rate": 2.7531250000000002e-05,
      "loss": 0.0002,
      "step": 5520
    },
    {
      "epoch": 17.28125,
      "grad_norm": 0.0007668856414966285,
      "learning_rate": 2.7218750000000004e-05,
      "loss": 0.0002,
      "step": 5530
    },
    {
      "epoch": 17.3125,
      "grad_norm": 0.0011194231919944286,
      "learning_rate": 2.690625e-05,
      "loss": 0.0002,
      "step": 5540
    },
    {
      "epoch": 17.34375,
      "grad_norm": 0.0010725741740316153,
      "learning_rate": 2.659375e-05,
      "loss": 0.0002,
      "step": 5550
    },
    {
      "epoch": 17.375,
      "grad_norm": 0.0013631158508360386,
      "learning_rate": 2.6281250000000002e-05,
      "loss": 0.0002,
      "step": 5560
    },
    {
      "epoch": 17.40625,
      "grad_norm": 0.0008963380823843181,
      "learning_rate": 2.5968750000000004e-05,
      "loss": 0.0001,
      "step": 5570
    },
    {
      "epoch": 17.4375,
      "grad_norm": 0.0008231993415392935,
      "learning_rate": 2.5656250000000004e-05,
      "loss": 0.0001,
      "step": 5580
    },
    {
      "epoch": 17.46875,
      "grad_norm": 0.0007719449349679053,
      "learning_rate": 2.534375e-05,
      "loss": 0.0002,
      "step": 5590
    },
    {
      "epoch": 17.5,
      "grad_norm": 0.0008315558661706746,
      "learning_rate": 2.5031250000000002e-05,
      "loss": 0.0001,
      "step": 5600
    },
    {
      "epoch": 17.5,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11211228370666504,
      "eval_runtime": 6.9482,
      "eval_samples_per_second": 184.221,
      "eval_steps_per_second": 23.028,
      "step": 5600
    },
    {
      "epoch": 17.53125,
      "grad_norm": 0.0007284819148480892,
      "learning_rate": 2.471875e-05,
      "loss": 0.0001,
      "step": 5610
    },
    {
      "epoch": 17.5625,
      "grad_norm": 0.0006902694585733116,
      "learning_rate": 2.440625e-05,
      "loss": 0.0001,
      "step": 5620
    },
    {
      "epoch": 17.59375,
      "grad_norm": 0.0007927055121399462,
      "learning_rate": 2.409375e-05,
      "loss": 0.0001,
      "step": 5630
    },
    {
      "epoch": 17.625,
      "grad_norm": 0.0008816330228000879,
      "learning_rate": 2.3781250000000002e-05,
      "loss": 0.0001,
      "step": 5640
    },
    {
      "epoch": 17.65625,
      "grad_norm": 0.00072521495167166,
      "learning_rate": 2.346875e-05,
      "loss": 0.0001,
      "step": 5650
    },
    {
      "epoch": 17.6875,
      "grad_norm": 0.0009029438951984048,
      "learning_rate": 2.315625e-05,
      "loss": 0.0001,
      "step": 5660
    },
    {
      "epoch": 17.71875,
      "grad_norm": 0.0007223158027045429,
      "learning_rate": 2.284375e-05,
      "loss": 0.0001,
      "step": 5670
    },
    {
      "epoch": 17.75,
      "grad_norm": 0.0007876714807935059,
      "learning_rate": 2.2531250000000002e-05,
      "loss": 0.0001,
      "step": 5680
    },
    {
      "epoch": 17.78125,
      "grad_norm": 0.0010673877550289035,
      "learning_rate": 2.221875e-05,
      "loss": 0.0001,
      "step": 5690
    },
    {
      "epoch": 17.8125,
      "grad_norm": 0.001492728479206562,
      "learning_rate": 2.190625e-05,
      "loss": 0.0001,
      "step": 5700
    },
    {
      "epoch": 17.8125,
      "eval_accuracy": 0.98359375,
      "eval_loss": 0.11293692886829376,
      "eval_runtime": 6.3413,
      "eval_samples_per_second": 201.851,
      "eval_steps_per_second": 25.231,
      "step": 5700
    },
    {
      "epoch": 17.84375,
      "grad_norm": 0.0008830283186398447,
      "learning_rate": 2.1593750000000003e-05,
      "loss": 0.0001,
      "step": 5710
    },
    {
      "epoch": 17.875,
      "grad_norm": 0.0006816980894654989,
      "learning_rate": 2.128125e-05,
      "loss": 0.0001,
      "step": 5720
    },
    {
      "epoch": 17.90625,
      "grad_norm": 0.0008512111380696297,
      "learning_rate": 2.096875e-05,
      "loss": 0.0001,
      "step": 5730
    },
    {
      "epoch": 17.9375,
      "grad_norm": 0.0008296071318909526,
      "learning_rate": 2.065625e-05,
      "loss": 0.0001,
      "step": 5740
    },
    {
      "epoch": 17.96875,
      "grad_norm": 0.0007265993626788259,
      "learning_rate": 2.034375e-05,
      "loss": 0.0001,
      "step": 5750
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.001100354827940464,
      "learning_rate": 2.0031250000000002e-05,
      "loss": 0.0001,
      "step": 5760
    },
    {
      "epoch": 18.03125,
      "grad_norm": 0.0008238213486038148,
      "learning_rate": 1.9718749999999998e-05,
      "loss": 0.0001,
      "step": 5770
    },
    {
      "epoch": 18.0625,
      "grad_norm": 0.0010971089359372854,
      "learning_rate": 1.940625e-05,
      "loss": 0.0002,
      "step": 5780
    },
    {
      "epoch": 18.09375,
      "grad_norm": 0.0007329233339987695,
      "learning_rate": 1.9093750000000003e-05,
      "loss": 0.0002,
      "step": 5790
    },
    {
      "epoch": 18.125,
      "grad_norm": 0.0008615129627287388,
      "learning_rate": 1.878125e-05,
      "loss": 0.0001,
      "step": 5800
    },
    {
      "epoch": 18.125,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11349210888147354,
      "eval_runtime": 6.4474,
      "eval_samples_per_second": 198.53,
      "eval_steps_per_second": 24.816,
      "step": 5800
    },
    {
      "epoch": 18.15625,
      "grad_norm": 0.0007490671123377979,
      "learning_rate": 1.846875e-05,
      "loss": 0.0001,
      "step": 5810
    },
    {
      "epoch": 18.1875,
      "grad_norm": 0.0011704419739544392,
      "learning_rate": 1.815625e-05,
      "loss": 0.0001,
      "step": 5820
    },
    {
      "epoch": 18.21875,
      "grad_norm": 0.0007434096769429743,
      "learning_rate": 1.784375e-05,
      "loss": 0.0002,
      "step": 5830
    },
    {
      "epoch": 18.25,
      "grad_norm": 0.0007305287872441113,
      "learning_rate": 1.7531250000000003e-05,
      "loss": 0.0001,
      "step": 5840
    },
    {
      "epoch": 18.28125,
      "grad_norm": 0.0011461430694907904,
      "learning_rate": 1.7218750000000002e-05,
      "loss": 0.0001,
      "step": 5850
    },
    {
      "epoch": 18.3125,
      "grad_norm": 0.001021283445879817,
      "learning_rate": 1.690625e-05,
      "loss": 0.0001,
      "step": 5860
    },
    {
      "epoch": 18.34375,
      "grad_norm": 0.0006852642400190234,
      "learning_rate": 1.659375e-05,
      "loss": 0.0001,
      "step": 5870
    },
    {
      "epoch": 18.375,
      "grad_norm": 0.0007609901949763298,
      "learning_rate": 1.628125e-05,
      "loss": 0.0001,
      "step": 5880
    },
    {
      "epoch": 18.40625,
      "grad_norm": 0.0038062427192926407,
      "learning_rate": 1.5968750000000002e-05,
      "loss": 0.0001,
      "step": 5890
    },
    {
      "epoch": 18.4375,
      "grad_norm": 0.0007008612737990916,
      "learning_rate": 1.565625e-05,
      "loss": 0.0001,
      "step": 5900
    },
    {
      "epoch": 18.4375,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11404496431350708,
      "eval_runtime": 6.2975,
      "eval_samples_per_second": 203.256,
      "eval_steps_per_second": 25.407,
      "step": 5900
    },
    {
      "epoch": 18.46875,
      "grad_norm": 0.001011393149383366,
      "learning_rate": 1.534375e-05,
      "loss": 0.0001,
      "step": 5910
    },
    {
      "epoch": 18.5,
      "grad_norm": 0.0007079701754264534,
      "learning_rate": 1.503125e-05,
      "loss": 0.0001,
      "step": 5920
    },
    {
      "epoch": 18.53125,
      "grad_norm": 0.001830734545364976,
      "learning_rate": 1.471875e-05,
      "loss": 0.0001,
      "step": 5930
    },
    {
      "epoch": 18.5625,
      "grad_norm": 0.0007542877574451268,
      "learning_rate": 1.4406250000000001e-05,
      "loss": 0.0001,
      "step": 5940
    },
    {
      "epoch": 18.59375,
      "grad_norm": 0.0006455339025706053,
      "learning_rate": 1.409375e-05,
      "loss": 0.0001,
      "step": 5950
    },
    {
      "epoch": 18.625,
      "grad_norm": 0.0009898885618895292,
      "learning_rate": 1.3781250000000001e-05,
      "loss": 0.0001,
      "step": 5960
    },
    {
      "epoch": 18.65625,
      "grad_norm": 0.0009699007496237755,
      "learning_rate": 1.3468749999999999e-05,
      "loss": 0.0001,
      "step": 5970
    },
    {
      "epoch": 18.6875,
      "grad_norm": 0.0007313139503821731,
      "learning_rate": 1.3156250000000001e-05,
      "loss": 0.0001,
      "step": 5980
    },
    {
      "epoch": 18.71875,
      "grad_norm": 0.0010632864432409406,
      "learning_rate": 1.2843750000000002e-05,
      "loss": 0.0001,
      "step": 5990
    },
    {
      "epoch": 18.75,
      "grad_norm": 0.000874130695592612,
      "learning_rate": 1.253125e-05,
      "loss": 0.0001,
      "step": 6000
    },
    {
      "epoch": 18.75,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11457158625125885,
      "eval_runtime": 6.7696,
      "eval_samples_per_second": 189.082,
      "eval_steps_per_second": 23.635,
      "step": 6000
    },
    {
      "epoch": 18.78125,
      "grad_norm": 0.0007607897859998047,
      "learning_rate": 1.221875e-05,
      "loss": 0.0001,
      "step": 6010
    },
    {
      "epoch": 18.8125,
      "grad_norm": 0.0006213558372110128,
      "learning_rate": 1.1906250000000001e-05,
      "loss": 0.0001,
      "step": 6020
    },
    {
      "epoch": 18.84375,
      "grad_norm": 0.0009066364727914333,
      "learning_rate": 1.159375e-05,
      "loss": 0.0001,
      "step": 6030
    },
    {
      "epoch": 18.875,
      "grad_norm": 0.0008698303136043251,
      "learning_rate": 1.128125e-05,
      "loss": 0.0001,
      "step": 6040
    },
    {
      "epoch": 18.90625,
      "grad_norm": 0.0008234538836404681,
      "learning_rate": 1.096875e-05,
      "loss": 0.0001,
      "step": 6050
    },
    {
      "epoch": 18.9375,
      "grad_norm": 0.0009270088630728424,
      "learning_rate": 1.0656250000000002e-05,
      "loss": 0.0001,
      "step": 6060
    },
    {
      "epoch": 18.96875,
      "grad_norm": 0.0007937824120745063,
      "learning_rate": 1.034375e-05,
      "loss": 0.0001,
      "step": 6070
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.0009614995797164738,
      "learning_rate": 1.003125e-05,
      "loss": 0.0001,
      "step": 6080
    },
    {
      "epoch": 19.03125,
      "grad_norm": 0.0006743675330653787,
      "learning_rate": 9.71875e-06,
      "loss": 0.0001,
      "step": 6090
    },
    {
      "epoch": 19.0625,
      "grad_norm": 0.0007780918967910111,
      "learning_rate": 9.40625e-06,
      "loss": 0.0001,
      "step": 6100
    },
    {
      "epoch": 19.0625,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11495751142501831,
      "eval_runtime": 6.295,
      "eval_samples_per_second": 203.337,
      "eval_steps_per_second": 25.417,
      "step": 6100
    },
    {
      "epoch": 19.09375,
      "grad_norm": 0.0010844800854101777,
      "learning_rate": 9.09375e-06,
      "loss": 0.0001,
      "step": 6110
    },
    {
      "epoch": 19.125,
      "grad_norm": 0.0007763968897052109,
      "learning_rate": 8.78125e-06,
      "loss": 0.0001,
      "step": 6120
    },
    {
      "epoch": 19.15625,
      "grad_norm": 0.0006313940975815058,
      "learning_rate": 8.468750000000001e-06,
      "loss": 0.0001,
      "step": 6130
    },
    {
      "epoch": 19.1875,
      "grad_norm": 0.0010176225332543254,
      "learning_rate": 8.15625e-06,
      "loss": 0.0001,
      "step": 6140
    },
    {
      "epoch": 19.21875,
      "grad_norm": 0.0006113633280619979,
      "learning_rate": 7.84375e-06,
      "loss": 0.0001,
      "step": 6150
    },
    {
      "epoch": 19.25,
      "grad_norm": 0.0009647855767980218,
      "learning_rate": 7.531250000000001e-06,
      "loss": 0.0001,
      "step": 6160
    },
    {
      "epoch": 19.28125,
      "grad_norm": 0.0008165627950802445,
      "learning_rate": 7.21875e-06,
      "loss": 0.0001,
      "step": 6170
    },
    {
      "epoch": 19.3125,
      "grad_norm": 0.000816655985545367,
      "learning_rate": 6.90625e-06,
      "loss": 0.0001,
      "step": 6180
    },
    {
      "epoch": 19.34375,
      "grad_norm": 0.0008548317127861083,
      "learning_rate": 6.59375e-06,
      "loss": 0.0001,
      "step": 6190
    },
    {
      "epoch": 19.375,
      "grad_norm": 0.0006300067761912942,
      "learning_rate": 6.281249999999999e-06,
      "loss": 0.0001,
      "step": 6200
    },
    {
      "epoch": 19.375,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11527726799249649,
      "eval_runtime": 6.2032,
      "eval_samples_per_second": 206.344,
      "eval_steps_per_second": 25.793,
      "step": 6200
    },
    {
      "epoch": 19.40625,
      "grad_norm": 0.0009761661058291793,
      "learning_rate": 5.96875e-06,
      "loss": 0.0001,
      "step": 6210
    },
    {
      "epoch": 19.4375,
      "grad_norm": 0.0007255289820022881,
      "learning_rate": 5.65625e-06,
      "loss": 0.0001,
      "step": 6220
    },
    {
      "epoch": 19.46875,
      "grad_norm": 0.0006587054231204093,
      "learning_rate": 5.34375e-06,
      "loss": 0.0001,
      "step": 6230
    },
    {
      "epoch": 19.5,
      "grad_norm": 0.0009364929865114391,
      "learning_rate": 5.03125e-06,
      "loss": 0.0001,
      "step": 6240
    },
    {
      "epoch": 19.53125,
      "grad_norm": 0.0006768327439203858,
      "learning_rate": 4.71875e-06,
      "loss": 0.0001,
      "step": 6250
    },
    {
      "epoch": 19.5625,
      "grad_norm": 0.0015788966557011008,
      "learning_rate": 4.40625e-06,
      "loss": 0.0001,
      "step": 6260
    },
    {
      "epoch": 19.59375,
      "grad_norm": 0.0006980461766943336,
      "learning_rate": 4.09375e-06,
      "loss": 0.0001,
      "step": 6270
    },
    {
      "epoch": 19.625,
      "grad_norm": 0.0008529416518285871,
      "learning_rate": 3.78125e-06,
      "loss": 0.0001,
      "step": 6280
    },
    {
      "epoch": 19.65625,
      "grad_norm": 0.0008394841570407152,
      "learning_rate": 3.4687500000000005e-06,
      "loss": 0.0001,
      "step": 6290
    },
    {
      "epoch": 19.6875,
      "grad_norm": 0.0008695558062754571,
      "learning_rate": 3.15625e-06,
      "loss": 0.0001,
      "step": 6300
    },
    {
      "epoch": 19.6875,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11545456945896149,
      "eval_runtime": 6.236,
      "eval_samples_per_second": 205.259,
      "eval_steps_per_second": 25.657,
      "step": 6300
    },
    {
      "epoch": 19.71875,
      "grad_norm": 0.0012889541685581207,
      "learning_rate": 2.84375e-06,
      "loss": 0.0001,
      "step": 6310
    },
    {
      "epoch": 19.75,
      "grad_norm": 0.0006377049721777439,
      "learning_rate": 2.53125e-06,
      "loss": 0.0001,
      "step": 6320
    },
    {
      "epoch": 19.78125,
      "grad_norm": 0.0007706707692705095,
      "learning_rate": 2.21875e-06,
      "loss": 0.0001,
      "step": 6330
    },
    {
      "epoch": 19.8125,
      "grad_norm": 0.0007259553531184793,
      "learning_rate": 1.90625e-06,
      "loss": 0.0001,
      "step": 6340
    },
    {
      "epoch": 19.84375,
      "grad_norm": 0.0009117970475926995,
      "learning_rate": 1.5937500000000002e-06,
      "loss": 0.0001,
      "step": 6350
    },
    {
      "epoch": 19.875,
      "grad_norm": 0.0008638539584353566,
      "learning_rate": 1.28125e-06,
      "loss": 0.0001,
      "step": 6360
    },
    {
      "epoch": 19.90625,
      "grad_norm": 0.0006554033607244492,
      "learning_rate": 9.6875e-07,
      "loss": 0.0001,
      "step": 6370
    },
    {
      "epoch": 19.9375,
      "grad_norm": 0.00083553371950984,
      "learning_rate": 6.5625e-07,
      "loss": 0.0001,
      "step": 6380
    },
    {
      "epoch": 19.96875,
      "grad_norm": 0.0006699099321849644,
      "learning_rate": 3.4375000000000004e-07,
      "loss": 0.0001,
      "step": 6390
    },
    {
      "epoch": 20.0,
      "grad_norm": 0.000951381167396903,
      "learning_rate": 3.1250000000000005e-08,
      "loss": 0.0001,
      "step": 6400
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.984375,
      "eval_loss": 0.11548350006341934,
      "eval_runtime": 6.596,
      "eval_samples_per_second": 194.058,
      "eval_steps_per_second": 24.257,
      "step": 6400
    },
    {
      "epoch": 20.0,
      "step": 6400,
      "total_flos": 7.935321977546342e+18,
      "train_loss": 0.10612523018964566,
      "train_runtime": 3045.4016,
      "train_samples_per_second": 33.624,
      "train_steps_per_second": 2.102
    }
  ],
  "logging_steps": 10,
  "max_steps": 6400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.935321977546342e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}