{
  "best_metric": 0.9925,
  "best_model_checkpoint": "./model/checkpoint-4382",
  "epoch": 30.0,
  "eval_steps": 500,
  "global_step": 9390,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 1.9169329073482426e-06,
      "loss": 1.4575,
      "step": 10
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.046858359957402e-06,
      "loss": 1.3457,
      "step": 20
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.17678381256656e-06,
      "loss": 1.2723,
      "step": 30
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.306709265175718e-06,
      "loss": 1.0944,
      "step": 40
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.0223642172523962e-05,
      "loss": 0.8709,
      "step": 50
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.235356762513312e-05,
      "loss": 0.7749,
      "step": 60
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.4270500532481365e-05,
      "loss": 0.7163,
      "step": 70
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.6400425985090524e-05,
      "loss": 0.6857,
      "step": 80
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.8530351437699682e-05,
      "loss": 0.6966,
      "step": 90
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.066027689030884e-05,
      "loss": 0.6462,
      "step": 100
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.2790202342918e-05,
      "loss": 0.5817,
      "step": 110
    },
    {
      "epoch": 0.38,
      "learning_rate": 2.4920127795527157e-05,
      "loss": 0.5795,
      "step": 120
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.7050053248136315e-05,
      "loss": 0.6217,
      "step": 130
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.9179978700745477e-05,
      "loss": 0.6524,
      "step": 140
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.130990415335464e-05,
      "loss": 0.5952,
      "step": 150
    },
    {
      "epoch": 0.51,
      "learning_rate": 3.343982960596379e-05,
      "loss": 0.6165,
      "step": 160
    },
    {
      "epoch": 0.54,
      "learning_rate": 3.556975505857295e-05,
      "loss": 0.5443,
      "step": 170
    },
    {
      "epoch": 0.58,
      "learning_rate": 3.769968051118211e-05,
      "loss": 0.5546,
      "step": 180
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.9829605963791265e-05,
      "loss": 0.6735,
      "step": 190
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.195953141640043e-05,
      "loss": 0.6384,
      "step": 200
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.408945686900959e-05,
      "loss": 0.5343,
      "step": 210
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.6219382321618746e-05,
      "loss": 0.5015,
      "step": 220
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.8349307774227905e-05,
      "loss": 0.5105,
      "step": 230
    },
    {
      "epoch": 0.77,
      "learning_rate": 5.047923322683706e-05,
      "loss": 0.4695,
      "step": 240
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.260915867944622e-05,
      "loss": 0.4371,
      "step": 250
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.4739084132055386e-05,
      "loss": 0.4953,
      "step": 260
    },
    {
      "epoch": 0.86,
      "learning_rate": 5.686900958466454e-05,
      "loss": 0.4942,
      "step": 270
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.8998935037273696e-05,
      "loss": 0.5203,
      "step": 280
    },
    {
      "epoch": 0.93,
      "learning_rate": 6.112886048988285e-05,
      "loss": 0.4275,
      "step": 290
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.30457933972311e-05,
      "loss": 0.5087,
      "step": 300
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.517571884984026e-05,
      "loss": 0.4345,
      "step": 310
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8075,
      "eval_loss": 0.41250982880592346,
      "eval_runtime": 5.1206,
      "eval_samples_per_second": 78.116,
      "eval_steps_per_second": 9.764,
      "step": 313
    },
    {
      "epoch": 1.02,
      "learning_rate": 6.730564430244943e-05,
      "loss": 0.3821,
      "step": 320
    },
    {
      "epoch": 1.05,
      "learning_rate": 6.943556975505858e-05,
      "loss": 0.3927,
      "step": 330
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.156549520766773e-05,
      "loss": 0.3204,
      "step": 340
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.36954206602769e-05,
      "loss": 0.3437,
      "step": 350
    },
    {
      "epoch": 1.15,
      "learning_rate": 7.582534611288606e-05,
      "loss": 0.4701,
      "step": 360
    },
    {
      "epoch": 1.18,
      "learning_rate": 7.795527156549521e-05,
      "loss": 0.3827,
      "step": 370
    },
    {
      "epoch": 1.21,
      "learning_rate": 8.008519701810438e-05,
      "loss": 0.3503,
      "step": 380
    },
    {
      "epoch": 1.25,
      "learning_rate": 8.221512247071353e-05,
      "loss": 0.3412,
      "step": 390
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.434504792332268e-05,
      "loss": 0.3545,
      "step": 400
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.647497337593184e-05,
      "loss": 0.4233,
      "step": 410
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.860489882854101e-05,
      "loss": 0.3038,
      "step": 420
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.073482428115016e-05,
      "loss": 0.2849,
      "step": 430
    },
    {
      "epoch": 1.41,
      "learning_rate": 9.286474973375933e-05,
      "loss": 0.3164,
      "step": 440
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.499467518636849e-05,
      "loss": 0.3366,
      "step": 450
    },
    {
      "epoch": 1.47,
      "learning_rate": 9.712460063897764e-05,
      "loss": 0.3453,
      "step": 460
    },
    {
      "epoch": 1.5,
      "learning_rate": 9.92545260915868e-05,
      "loss": 0.3852,
      "step": 470
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00010138445154419596,
      "loss": 0.3932,
      "step": 480
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00010351437699680511,
      "loss": 0.3085,
      "step": 490
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00010564430244941428,
      "loss": 0.4473,
      "step": 500
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00010777422790202343,
      "loss": 0.381,
      "step": 510
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0001099041533546326,
      "loss": 0.3917,
      "step": 520
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00011203407880724174,
      "loss": 0.3807,
      "step": 530
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00011416400425985092,
      "loss": 0.3381,
      "step": 540
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00011629392971246007,
      "loss": 0.3207,
      "step": 550
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00011842385516506922,
      "loss": 0.2718,
      "step": 560
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00012055378061767839,
      "loss": 0.3206,
      "step": 570
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00012268370607028753,
      "loss": 0.3239,
      "step": 580
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0001248136315228967,
      "loss": 0.2478,
      "step": 590
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00012694355697550586,
      "loss": 0.1582,
      "step": 600
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.000129073482428115,
      "loss": 0.2166,
      "step": 610
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001312034078807242,
      "loss": 0.5444,
      "step": 620
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9075,
      "eval_loss": 0.28795498609542847,
      "eval_runtime": 5.2021,
      "eval_samples_per_second": 76.892,
      "eval_steps_per_second": 9.612,
      "step": 626
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00013333333333333334,
      "loss": 0.2552,
      "step": 630
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0001354632587859425,
      "loss": 0.2285,
      "step": 640
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00013759318423855164,
      "loss": 0.296,
      "step": 650
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00013972310969116082,
      "loss": 0.2718,
      "step": 660
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00014185303514376997,
      "loss": 0.1728,
      "step": 670
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00014398296059637912,
      "loss": 0.1852,
      "step": 680
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0001461128860489883,
      "loss": 0.2437,
      "step": 690
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00014824281150159745,
      "loss": 0.2741,
      "step": 700
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0001503727369542066,
      "loss": 0.1316,
      "step": 710
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00015250266240681576,
      "loss": 0.2716,
      "step": 720
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00015463258785942494,
      "loss": 0.2318,
      "step": 730
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001567625133120341,
      "loss": 0.4942,
      "step": 740
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00015889243876464324,
      "loss": 0.218,
      "step": 750
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00016102236421725242,
      "loss": 0.2507,
      "step": 760
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00016315228966986157,
      "loss": 0.2782,
      "step": 770
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00016528221512247072,
      "loss": 0.1941,
      "step": 780
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00016741214057507987,
      "loss": 0.4018,
      "step": 790
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00016954206602768902,
      "loss": 0.3268,
      "step": 800
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001716719914802982,
      "loss": 0.2585,
      "step": 810
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017380191693290735,
      "loss": 0.3031,
      "step": 820
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017593184238551653,
      "loss": 0.313,
      "step": 830
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017806176783812566,
      "loss": 0.3424,
      "step": 840
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00018019169329073484,
      "loss": 0.2635,
      "step": 850
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.000182321618743344,
      "loss": 0.2479,
      "step": 860
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00018445154419595314,
      "loss": 0.2651,
      "step": 870
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00018658146964856232,
      "loss": 0.228,
      "step": 880
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00018871139510117147,
      "loss": 0.288,
      "step": 890
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00019084132055378062,
      "loss": 0.2814,
      "step": 900
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00019297124600638977,
      "loss": 0.2692,
      "step": 910
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00019510117145899895,
      "loss": 0.235,
      "step": 920
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0001972310969116081,
      "loss": 0.2855,
      "step": 930
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.88,
      "eval_loss": 0.3687535524368286,
      "eval_runtime": 6.15,
      "eval_samples_per_second": 65.041,
      "eval_steps_per_second": 8.13,
      "step": 939
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00019936102236421725,
      "loss": 0.2939,
      "step": 940
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.00019983433913146374,
      "loss": 0.3828,
      "step": 950
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001995976807478405,
      "loss": 0.1991,
      "step": 960
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.00019936102236421725,
      "loss": 0.2212,
      "step": 970
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.00019912436398059402,
      "loss": 0.2374,
      "step": 980
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.0001988877055969708,
      "loss": 0.1184,
      "step": 990
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.00019865104721334756,
      "loss": 0.1316,
      "step": 1000
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.0001984143888297243,
      "loss": 0.2247,
      "step": 1010
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.00019817773044610105,
      "loss": 0.2707,
      "step": 1020
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.00019794107206247782,
      "loss": 0.1726,
      "step": 1030
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00019770441367885456,
      "loss": 0.2527,
      "step": 1040
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00019746775529523133,
      "loss": 0.2112,
      "step": 1050
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.0001972310969116081,
      "loss": 0.1687,
      "step": 1060
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.00019699443852798487,
      "loss": 0.3774,
      "step": 1070
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00019675778014436161,
      "loss": 0.2342,
      "step": 1080
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00019652112176073838,
      "loss": 0.195,
      "step": 1090
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00019628446337711515,
      "loss": 0.1589,
      "step": 1100
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.00019604780499349192,
      "loss": 0.2088,
      "step": 1110
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.00019581114660986867,
      "loss": 0.2503,
      "step": 1120
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.0001955744882262454,
      "loss": 0.1818,
      "step": 1130
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.00019533782984262218,
      "loss": 0.2153,
      "step": 1140
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.00019510117145899895,
      "loss": 0.3874,
      "step": 1150
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.0001948645130753757,
      "loss": 0.179,
      "step": 1160
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.00019462785469175246,
      "loss": 0.1135,
      "step": 1170
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.00019439119630812923,
      "loss": 0.3062,
      "step": 1180
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.00019415453792450598,
      "loss": 0.163,
      "step": 1190
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.00019391787954088275,
      "loss": 0.2267,
      "step": 1200
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.00019368122115725952,
      "loss": 0.1419,
      "step": 1210
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.00019344456277363629,
      "loss": 0.1842,
      "step": 1220
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.00019320790439001303,
      "loss": 0.1742,
      "step": 1230
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.00019297124600638977,
      "loss": 0.1806,
      "step": 1240
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.00019273458762276654,
      "loss": 0.1533,
      "step": 1250
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.94,
      "eval_loss": 0.2066069096326828,
      "eval_runtime": 5.147,
      "eval_samples_per_second": 77.715,
      "eval_steps_per_second": 9.714,
      "step": 1252
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.0001924979292391433,
      "loss": 0.153,
      "step": 1260
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.00019226127085552005,
      "loss": 0.1504,
      "step": 1270
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.00019202461247189682,
      "loss": 0.1796,
      "step": 1280
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.0001917879540882736,
      "loss": 0.1309,
      "step": 1290
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00019155129570465034,
      "loss": 0.1167,
      "step": 1300
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.0001913146373210271,
      "loss": 0.0874,
      "step": 1310
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.00019107797893740388,
      "loss": 0.137,
      "step": 1320
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.00019084132055378062,
      "loss": 0.0889,
      "step": 1330
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.0001906046621701574,
      "loss": 0.1578,
      "step": 1340
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.00019036800378653413,
      "loss": 0.1551,
      "step": 1350
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.0001901313454029109,
      "loss": 0.087,
      "step": 1360
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.00018989468701928767,
      "loss": 0.0657,
      "step": 1370
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.00018965802863566442,
      "loss": 0.1458,
      "step": 1380
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.00018942137025204119,
      "loss": 0.2047,
      "step": 1390
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.00018918471186841796,
      "loss": 0.1534,
      "step": 1400
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.00018894805348479473,
      "loss": 0.11,
      "step": 1410
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00018871139510117147,
      "loss": 0.0986,
      "step": 1420
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00018847473671754824,
      "loss": 0.1409,
      "step": 1430
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.00018823807833392498,
      "loss": 0.0848,
      "step": 1440
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.00018800141995030175,
      "loss": 0.1584,
      "step": 1450
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001877647615666785,
      "loss": 0.1823,
      "step": 1460
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00018752810318305526,
      "loss": 0.1486,
      "step": 1470
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.00018729144479943203,
      "loss": 0.0709,
      "step": 1480
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.00018705478641580878,
      "loss": 0.1484,
      "step": 1490
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.00018681812803218555,
      "loss": 0.1866,
      "step": 1500
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.00018658146964856232,
      "loss": 0.1004,
      "step": 1510
    },
    {
      "epoch": 4.86,
      "learning_rate": 0.0001863448112649391,
      "loss": 0.095,
      "step": 1520
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.00018610815288131583,
      "loss": 0.1464,
      "step": 1530
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.0001858714944976926,
      "loss": 0.14,
      "step": 1540
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.00018563483611406934,
      "loss": 0.1327,
      "step": 1550
    },
    {
      "epoch": 4.98,
      "learning_rate": 0.0001853981777304461,
      "loss": 0.0551,
      "step": 1560
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9725,
      "eval_loss": 0.12336118519306183,
      "eval_runtime": 5.8571,
      "eval_samples_per_second": 68.293,
      "eval_steps_per_second": 8.537,
      "step": 1565
    },
    {
      "epoch": 5.02,
      "learning_rate": 0.00018516151934682286,
      "loss": 0.1012,
      "step": 1570
    },
    {
      "epoch": 5.05,
      "learning_rate": 0.00018492486096319963,
      "loss": 0.0478,
      "step": 1580
    },
    {
      "epoch": 5.08,
      "learning_rate": 0.0001846882025795764,
      "loss": 0.1106,
      "step": 1590
    },
    {
      "epoch": 5.11,
      "learning_rate": 0.00018445154419595314,
      "loss": 0.1352,
      "step": 1600
    },
    {
      "epoch": 5.14,
      "learning_rate": 0.0001842148858123299,
      "loss": 0.2177,
      "step": 1610
    },
    {
      "epoch": 5.18,
      "learning_rate": 0.00018397822742870668,
      "loss": 0.1155,
      "step": 1620
    },
    {
      "epoch": 5.21,
      "learning_rate": 0.00018374156904508345,
      "loss": 0.067,
      "step": 1630
    },
    {
      "epoch": 5.24,
      "learning_rate": 0.0001835049106614602,
      "loss": 0.0788,
      "step": 1640
    },
    {
      "epoch": 5.27,
      "learning_rate": 0.00018326825227783696,
      "loss": 0.1365,
      "step": 1650
    },
    {
      "epoch": 5.3,
      "learning_rate": 0.0001830315938942137,
      "loss": 0.096,
      "step": 1660
    },
    {
      "epoch": 5.34,
      "learning_rate": 0.00018279493551059047,
      "loss": 0.1184,
      "step": 1670
    },
    {
      "epoch": 5.37,
      "learning_rate": 0.00018255827712696722,
      "loss": 0.0786,
      "step": 1680
    },
    {
      "epoch": 5.4,
      "learning_rate": 0.000182321618743344,
      "loss": 0.1599,
      "step": 1690
    },
    {
      "epoch": 5.43,
      "learning_rate": 0.00018208496035972076,
      "loss": 0.0736,
      "step": 1700
    },
    {
      "epoch": 5.46,
      "learning_rate": 0.0001818483019760975,
      "loss": 0.1095,
      "step": 1710
    },
    {
      "epoch": 5.5,
      "learning_rate": 0.00018161164359247427,
      "loss": 0.1107,
      "step": 1720
    },
    {
      "epoch": 5.53,
      "learning_rate": 0.00018137498520885104,
      "loss": 0.0465,
      "step": 1730
    },
    {
      "epoch": 5.56,
      "learning_rate": 0.0001811383268252278,
      "loss": 0.1244,
      "step": 1740
    },
    {
      "epoch": 5.59,
      "learning_rate": 0.00018090166844160455,
      "loss": 0.1155,
      "step": 1750
    },
    {
      "epoch": 5.62,
      "learning_rate": 0.00018066501005798132,
      "loss": 0.1181,
      "step": 1760
    },
    {
      "epoch": 5.65,
      "learning_rate": 0.00018042835167435807,
      "loss": 0.0315,
      "step": 1770
    },
    {
      "epoch": 5.69,
      "learning_rate": 0.00018019169329073484,
      "loss": 0.0446,
      "step": 1780
    },
    {
      "epoch": 5.72,
      "learning_rate": 0.00017995503490711158,
      "loss": 0.1342,
      "step": 1790
    },
    {
      "epoch": 5.75,
      "learning_rate": 0.00017971837652348835,
      "loss": 0.1308,
      "step": 1800
    },
    {
      "epoch": 5.78,
      "learning_rate": 0.00017948171813986512,
      "loss": 0.0927,
      "step": 1810
    },
    {
      "epoch": 5.81,
      "learning_rate": 0.0001792450597562419,
      "loss": 0.1147,
      "step": 1820
    },
    {
      "epoch": 5.85,
      "learning_rate": 0.00017900840137261863,
      "loss": 0.0814,
      "step": 1830
    },
    {
      "epoch": 5.88,
      "learning_rate": 0.0001787717429889954,
      "loss": 0.1629,
      "step": 1840
    },
    {
      "epoch": 5.91,
      "learning_rate": 0.00017853508460537217,
      "loss": 0.1288,
      "step": 1850
    },
    {
      "epoch": 5.94,
      "learning_rate": 0.00017829842622174891,
      "loss": 0.2005,
      "step": 1860
    },
    {
      "epoch": 5.97,
      "learning_rate": 0.00017806176783812566,
      "loss": 0.1027,
      "step": 1870
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.97,
      "eval_loss": 0.11499868333339691,
      "eval_runtime": 8.0385,
      "eval_samples_per_second": 49.761,
      "eval_steps_per_second": 6.22,
      "step": 1878
    },
    {
      "epoch": 6.01,
      "learning_rate": 0.00017782510945450243,
      "loss": 0.051,
      "step": 1880
    },
    {
      "epoch": 6.04,
      "learning_rate": 0.0001775884510708792,
      "loss": 0.0318,
      "step": 1890
    },
    {
      "epoch": 6.07,
      "learning_rate": 0.00017735179268725594,
      "loss": 0.0468,
      "step": 1900
    },
    {
      "epoch": 6.1,
      "learning_rate": 0.0001771151343036327,
      "loss": 0.074,
      "step": 1910
    },
    {
      "epoch": 6.13,
      "learning_rate": 0.00017687847592000948,
      "loss": 0.0653,
      "step": 1920
    },
    {
      "epoch": 6.17,
      "learning_rate": 0.00017664181753638625,
      "loss": 0.1136,
      "step": 1930
    },
    {
      "epoch": 6.2,
      "learning_rate": 0.000176405159152763,
      "loss": 0.0369,
      "step": 1940
    },
    {
      "epoch": 6.23,
      "learning_rate": 0.00017616850076913976,
      "loss": 0.0872,
      "step": 1950
    },
    {
      "epoch": 6.26,
      "learning_rate": 0.00017593184238551653,
      "loss": 0.1174,
      "step": 1960
    },
    {
      "epoch": 6.29,
      "learning_rate": 0.00017569518400189328,
      "loss": 0.0716,
      "step": 1970
    },
    {
      "epoch": 6.33,
      "learning_rate": 0.00017545852561827002,
      "loss": 0.1122,
      "step": 1980
    },
    {
      "epoch": 6.36,
      "learning_rate": 0.0001752218672346468,
      "loss": 0.1801,
      "step": 1990
    },
    {
      "epoch": 6.39,
      "learning_rate": 0.00017498520885102356,
      "loss": 0.0333,
      "step": 2000
    },
    {
      "epoch": 6.42,
      "learning_rate": 0.0001747485504674003,
      "loss": 0.0227,
      "step": 2010
    },
    {
      "epoch": 6.45,
      "learning_rate": 0.00017451189208377707,
      "loss": 0.0389,
      "step": 2020
    },
    {
      "epoch": 6.49,
      "learning_rate": 0.00017427523370015384,
      "loss": 0.0434,
      "step": 2030
    },
    {
      "epoch": 6.52,
      "learning_rate": 0.0001740385753165306,
      "loss": 0.0726,
      "step": 2040
    },
    {
      "epoch": 6.55,
      "learning_rate": 0.00017380191693290735,
      "loss": 0.0456,
      "step": 2050
    },
    {
      "epoch": 6.58,
      "learning_rate": 0.00017356525854928412,
      "loss": 0.0396,
      "step": 2060
    },
    {
      "epoch": 6.61,
      "learning_rate": 0.0001733286001656609,
      "loss": 0.0983,
      "step": 2070
    },
    {
      "epoch": 6.65,
      "learning_rate": 0.00017309194178203764,
      "loss": 0.0743,
      "step": 2080
    },
    {
      "epoch": 6.68,
      "learning_rate": 0.00017285528339841438,
      "loss": 0.055,
      "step": 2090
    },
    {
      "epoch": 6.71,
      "learning_rate": 0.00017261862501479115,
      "loss": 0.1489,
      "step": 2100
    },
    {
      "epoch": 6.74,
      "learning_rate": 0.00017238196663116792,
      "loss": 0.0918,
      "step": 2110
    },
    {
      "epoch": 6.77,
      "learning_rate": 0.00017214530824754466,
      "loss": 0.0382,
      "step": 2120
    },
    {
      "epoch": 6.81,
      "learning_rate": 0.00017190864986392143,
      "loss": 0.03,
      "step": 2130
    },
    {
      "epoch": 6.84,
      "learning_rate": 0.0001716719914802982,
      "loss": 0.035,
      "step": 2140
    },
    {
      "epoch": 6.87,
      "learning_rate": 0.00017143533309667497,
      "loss": 0.1244,
      "step": 2150
    },
    {
      "epoch": 6.9,
      "learning_rate": 0.00017119867471305172,
      "loss": 0.1426,
      "step": 2160
    },
    {
      "epoch": 6.93,
      "learning_rate": 0.00017096201632942849,
      "loss": 0.0794,
      "step": 2170
    },
    {
      "epoch": 6.96,
      "learning_rate": 0.00017072535794580526,
      "loss": 0.1954,
      "step": 2180
    },
    {
      "epoch": 7.0,
      "learning_rate": 0.000170488699562182,
      "loss": 0.0466,
      "step": 2190
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.955,
      "eval_loss": 0.15742188692092896,
      "eval_runtime": 6.8368,
      "eval_samples_per_second": 58.507,
      "eval_steps_per_second": 7.313,
      "step": 2191
    },
    {
      "epoch": 7.03,
      "learning_rate": 0.00017025204117855874,
      "loss": 0.0528,
      "step": 2200
    },
    {
      "epoch": 7.06,
      "learning_rate": 0.0001700153827949355,
      "loss": 0.0362,
      "step": 2210
    },
    {
      "epoch": 7.09,
      "learning_rate": 0.00016977872441131228,
      "loss": 0.1156,
      "step": 2220
    },
    {
      "epoch": 7.12,
      "learning_rate": 0.00016954206602768902,
      "loss": 0.0595,
      "step": 2230
    },
    {
      "epoch": 7.16,
      "learning_rate": 0.0001693054076440658,
      "loss": 0.0263,
      "step": 2240
    },
    {
      "epoch": 7.19,
      "learning_rate": 0.00016906874926044256,
      "loss": 0.071,
      "step": 2250
    },
    {
      "epoch": 7.22,
      "learning_rate": 0.00016883209087681933,
      "loss": 0.0282,
      "step": 2260
    },
    {
      "epoch": 7.25,
      "learning_rate": 0.00016859543249319608,
      "loss": 0.0367,
      "step": 2270
    },
    {
      "epoch": 7.28,
      "learning_rate": 0.00016835877410957285,
      "loss": 0.0475,
      "step": 2280
    },
    {
      "epoch": 7.32,
      "learning_rate": 0.0001681221157259496,
      "loss": 0.0636,
      "step": 2290
    },
    {
      "epoch": 7.35,
      "learning_rate": 0.00016788545734232636,
      "loss": 0.0471,
      "step": 2300
    },
    {
      "epoch": 7.38,
      "learning_rate": 0.0001676487989587031,
      "loss": 0.0322,
      "step": 2310
    },
    {
      "epoch": 7.41,
      "learning_rate": 0.00016741214057507987,
      "loss": 0.0763,
      "step": 2320
    },
    {
      "epoch": 7.44,
      "learning_rate": 0.00016717548219145664,
      "loss": 0.1087,
      "step": 2330
    },
    {
      "epoch": 7.48,
      "learning_rate": 0.0001669388238078334,
      "loss": 0.0612,
      "step": 2340
    },
    {
      "epoch": 7.51,
      "learning_rate": 0.00016670216542421016,
      "loss": 0.0327,
      "step": 2350
    },
    {
      "epoch": 7.54,
      "learning_rate": 0.00016646550704058693,
      "loss": 0.0609,
      "step": 2360
    },
    {
      "epoch": 7.57,
      "learning_rate": 0.0001662288486569637,
      "loss": 0.0506,
      "step": 2370
    },
    {
      "epoch": 7.6,
      "learning_rate": 0.00016599219027334044,
      "loss": 0.031,
      "step": 2380
    },
    {
      "epoch": 7.64,
      "learning_rate": 0.0001657555318897172,
      "loss": 0.0729,
      "step": 2390
    },
    {
      "epoch": 7.67,
      "learning_rate": 0.00016551887350609395,
      "loss": 0.0224,
      "step": 2400
    },
    {
      "epoch": 7.7,
      "learning_rate": 0.00016528221512247072,
      "loss": 0.0306,
      "step": 2410
    },
    {
      "epoch": 7.73,
      "learning_rate": 0.00016504555673884746,
      "loss": 0.0286,
      "step": 2420
    },
    {
      "epoch": 7.76,
      "learning_rate": 0.00016480889835522423,
      "loss": 0.075,
      "step": 2430
    },
    {
      "epoch": 7.8,
      "learning_rate": 0.000164572239971601,
      "loss": 0.074,
      "step": 2440
    },
    {
      "epoch": 7.83,
      "learning_rate": 0.00016433558158797777,
      "loss": 0.1059,
      "step": 2450
    },
    {
      "epoch": 7.86,
      "learning_rate": 0.00016409892320435452,
      "loss": 0.078,
      "step": 2460
    },
    {
      "epoch": 7.89,
      "learning_rate": 0.00016386226482073129,
      "loss": 0.0753,
      "step": 2470
    },
    {
      "epoch": 7.92,
      "learning_rate": 0.00016362560643710806,
      "loss": 0.0698,
      "step": 2480
    },
    {
      "epoch": 7.96,
      "learning_rate": 0.0001633889480534848,
      "loss": 0.1025,
      "step": 2490
    },
    {
      "epoch": 7.99,
      "learning_rate": 0.00016315228966986157,
      "loss": 0.1343,
      "step": 2500
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.955,
      "eval_loss": 0.2272537797689438,
      "eval_runtime": 5.1085,
      "eval_samples_per_second": 78.302,
      "eval_steps_per_second": 9.788,
      "step": 2504
    },
    {
      "epoch": 8.02,
      "learning_rate": 0.0001629156312862383,
      "loss": 0.038,
      "step": 2510
    },
    {
      "epoch": 8.05,
      "learning_rate": 0.00016267897290261508,
      "loss": 0.0652,
      "step": 2520
    },
    {
      "epoch": 8.08,
      "learning_rate": 0.00016244231451899182,
      "loss": 0.037,
      "step": 2530
    },
    {
      "epoch": 8.12,
      "learning_rate": 0.0001622056561353686,
      "loss": 0.0205,
      "step": 2540
    },
    {
      "epoch": 8.15,
      "learning_rate": 0.00016196899775174536,
      "loss": 0.0323,
      "step": 2550
    },
    {
      "epoch": 8.18,
      "learning_rate": 0.00016173233936812213,
      "loss": 0.0932,
      "step": 2560
    },
    {
      "epoch": 8.21,
      "learning_rate": 0.00016149568098449888,
      "loss": 0.0683,
      "step": 2570
    },
    {
      "epoch": 8.24,
      "learning_rate": 0.00016125902260087565,
      "loss": 0.0637,
      "step": 2580
    },
    {
      "epoch": 8.27,
      "learning_rate": 0.00016102236421725242,
      "loss": 0.0249,
      "step": 2590
    },
    {
      "epoch": 8.31,
      "learning_rate": 0.0001607857058336292,
      "loss": 0.132,
      "step": 2600
    },
    {
      "epoch": 8.34,
      "learning_rate": 0.00016054904745000593,
      "loss": 0.0318,
      "step": 2610
    },
    {
      "epoch": 8.37,
      "learning_rate": 0.00016031238906638267,
      "loss": 0.0479,
      "step": 2620
    },
    {
      "epoch": 8.4,
      "learning_rate": 0.00016007573068275944,
      "loss": 0.0313,
      "step": 2630
    },
    {
      "epoch": 8.43,
      "learning_rate": 0.00015983907229913619,
      "loss": 0.0409,
      "step": 2640
    },
    {
      "epoch": 8.47,
      "learning_rate": 0.00015960241391551296,
      "loss": 0.038,
      "step": 2650
    },
    {
      "epoch": 8.5,
      "learning_rate": 0.00015936575553188973,
      "loss": 0.0527,
      "step": 2660
    },
    {
      "epoch": 8.53,
      "learning_rate": 0.0001591290971482665,
      "loss": 0.0792,
      "step": 2670
    },
    {
      "epoch": 8.56,
      "learning_rate": 0.00015889243876464324,
      "loss": 0.0708,
      "step": 2680
    },
    {
      "epoch": 8.59,
      "learning_rate": 0.00015865578038102,
      "loss": 0.0673,
      "step": 2690
    },
    {
      "epoch": 8.63,
      "learning_rate": 0.00015841912199739678,
      "loss": 0.0326,
      "step": 2700
    },
    {
      "epoch": 8.66,
      "learning_rate": 0.00015818246361377352,
      "loss": 0.071,
      "step": 2710
    },
    {
      "epoch": 8.69,
      "learning_rate": 0.0001579458052301503,
      "loss": 0.1203,
      "step": 2720
    },
    {
      "epoch": 8.72,
      "learning_rate": 0.00015770914684652703,
      "loss": 0.0396,
      "step": 2730
    },
    {
      "epoch": 8.75,
      "learning_rate": 0.0001574724884629038,
      "loss": 0.0379,
      "step": 2740
    },
    {
      "epoch": 8.79,
      "learning_rate": 0.00015723583007928055,
      "loss": 0.0451,
      "step": 2750
    },
    {
      "epoch": 8.82,
      "learning_rate": 0.00015699917169565732,
      "loss": 0.0135,
      "step": 2760
    },
    {
      "epoch": 8.85,
      "learning_rate": 0.0001567625133120341,
      "loss": 0.0331,
      "step": 2770
    },
    {
      "epoch": 8.88,
      "learning_rate": 0.00015652585492841086,
      "loss": 0.0789,
      "step": 2780
    },
    {
      "epoch": 8.91,
      "learning_rate": 0.0001562891965447876,
      "loss": 0.0183,
      "step": 2790
    },
    {
      "epoch": 8.95,
      "learning_rate": 0.00015605253816116437,
      "loss": 0.0287,
      "step": 2800
    },
    {
      "epoch": 8.98,
      "learning_rate": 0.00015581587977754114,
      "loss": 0.0721,
      "step": 2810
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.985,
      "eval_loss": 0.07111229002475739,
      "eval_runtime": 6.7619,
      "eval_samples_per_second": 59.155,
      "eval_steps_per_second": 7.394,
      "step": 2817
    },
    {
      "epoch": 9.01,
      "learning_rate": 0.00015557922139391788,
      "loss": 0.0057,
      "step": 2820
    },
    {
      "epoch": 9.04,
      "learning_rate": 0.00015534256301029463,
      "loss": 0.0627,
      "step": 2830
    },
    {
      "epoch": 9.07,
      "learning_rate": 0.0001551059046266714,
      "loss": 0.0825,
      "step": 2840
    },
    {
      "epoch": 9.11,
      "learning_rate": 0.00015486924624304817,
      "loss": 0.0128,
      "step": 2850
    },
    {
      "epoch": 9.14,
      "learning_rate": 0.00015463258785942494,
      "loss": 0.011,
      "step": 2860
    },
    {
      "epoch": 9.17,
      "learning_rate": 0.00015439592947580168,
      "loss": 0.0174,
      "step": 2870
    },
    {
      "epoch": 9.2,
      "learning_rate": 0.00015415927109217845,
      "loss": 0.0184,
      "step": 2880
    },
    {
      "epoch": 9.23,
      "learning_rate": 0.00015392261270855522,
      "loss": 0.0184,
      "step": 2890
    },
    {
      "epoch": 9.27,
      "learning_rate": 0.00015368595432493196,
      "loss": 0.0292,
      "step": 2900
    },
    {
      "epoch": 9.3,
      "learning_rate": 0.00015344929594130873,
      "loss": 0.0538,
      "step": 2910
    },
    {
      "epoch": 9.33,
      "learning_rate": 0.0001532126375576855,
      "loss": 0.0304,
      "step": 2920
    },
    {
      "epoch": 9.36,
      "learning_rate": 0.00015297597917406224,
      "loss": 0.0354,
      "step": 2930
    },
    {
      "epoch": 9.39,
      "learning_rate": 0.000152739320790439,
      "loss": 0.038,
      "step": 2940
    },
    {
      "epoch": 9.42,
      "learning_rate": 0.00015250266240681576,
      "loss": 0.0028,
      "step": 2950
    },
    {
      "epoch": 9.46,
      "learning_rate": 0.00015226600402319253,
      "loss": 0.0448,
      "step": 2960
    },
    {
      "epoch": 9.49,
      "learning_rate": 0.0001520293456395693,
      "loss": 0.041,
      "step": 2970
    },
    {
      "epoch": 9.52,
      "learning_rate": 0.00015179268725594604,
      "loss": 0.0441,
      "step": 2980
    },
    {
      "epoch": 9.55,
      "learning_rate": 0.0001515560288723228,
      "loss": 0.0047,
      "step": 2990
    },
    {
      "epoch": 9.58,
      "learning_rate": 0.00015131937048869958,
      "loss": 0.0098,
      "step": 3000
    },
    {
      "epoch": 9.62,
      "learning_rate": 0.00015108271210507635,
      "loss": 0.0495,
      "step": 3010
    },
    {
      "epoch": 9.65,
      "learning_rate": 0.0001508460537214531,
      "loss": 0.0019,
      "step": 3020
    },
    {
      "epoch": 9.68,
      "learning_rate": 0.00015060939533782986,
      "loss": 0.0681,
      "step": 3030
    },
    {
      "epoch": 9.71,
      "learning_rate": 0.0001503727369542066,
      "loss": 0.0154,
      "step": 3040
    },
    {
      "epoch": 9.74,
      "learning_rate": 0.00015013607857058335,
      "loss": 0.033,
      "step": 3050
    },
    {
      "epoch": 9.78,
      "learning_rate": 0.00014989942018696012,
      "loss": 0.0211,
      "step": 3060
    },
    {
      "epoch": 9.81,
      "learning_rate": 0.0001496627618033369,
      "loss": 0.0246,
      "step": 3070
    },
    {
      "epoch": 9.84,
      "learning_rate": 0.00014942610341971366,
      "loss": 0.0153,
      "step": 3080
    },
    {
      "epoch": 9.87,
      "learning_rate": 0.0001491894450360904,
      "loss": 0.0035,
      "step": 3090
    },
    {
      "epoch": 9.9,
      "learning_rate": 0.00014895278665246717,
      "loss": 0.1033,
      "step": 3100
    },
    {
      "epoch": 9.94,
      "learning_rate": 0.00014871612826884394,
      "loss": 0.0397,
      "step": 3110
    },
    {
      "epoch": 9.97,
      "learning_rate": 0.0001484794698852207,
      "loss": 0.0303,
      "step": 3120
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.00014824281150159745,
      "loss": 0.0023,
      "step": 3130
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9775,
      "eval_loss": 0.16127361357212067,
      "eval_runtime": 5.6352,
      "eval_samples_per_second": 70.983,
      "eval_steps_per_second": 8.873,
      "step": 3130
    },
    {
      "epoch": 10.03,
      "learning_rate": 0.00014800615311797422,
      "loss": 0.0019,
      "step": 3140
    },
    {
      "epoch": 10.06,
      "learning_rate": 0.00014776949473435097,
      "loss": 0.0403,
      "step": 3150
    },
    {
      "epoch": 10.1,
      "learning_rate": 0.0001475328363507277,
      "loss": 0.0242,
      "step": 3160
    },
    {
      "epoch": 10.13,
      "learning_rate": 0.00014729617796710448,
      "loss": 0.0859,
      "step": 3170
    },
    {
      "epoch": 10.16,
      "learning_rate": 0.00014705951958348125,
      "loss": 0.0928,
      "step": 3180
    },
    {
      "epoch": 10.19,
      "learning_rate": 0.00014682286119985802,
      "loss": 0.0374,
      "step": 3190
    },
    {
      "epoch": 10.22,
      "learning_rate": 0.00014658620281623476,
      "loss": 0.0535,
      "step": 3200
    },
    {
      "epoch": 10.26,
      "learning_rate": 0.00014634954443261153,
      "loss": 0.0237,
      "step": 3210
    },
    {
      "epoch": 10.29,
      "learning_rate": 0.0001461128860489883,
      "loss": 0.0512,
      "step": 3220
    },
    {
      "epoch": 10.32,
      "learning_rate": 0.00014587622766536507,
      "loss": 0.0284,
      "step": 3230
    },
    {
      "epoch": 10.35,
      "learning_rate": 0.00014563956928174182,
      "loss": 0.014,
      "step": 3240
    },
    {
      "epoch": 10.38,
      "learning_rate": 0.00014540291089811856,
      "loss": 0.0281,
      "step": 3250
    },
    {
      "epoch": 10.42,
      "learning_rate": 0.00014516625251449533,
      "loss": 0.0441,
      "step": 3260
    },
    {
      "epoch": 10.45,
      "learning_rate": 0.0001449295941308721,
      "loss": 0.0165,
      "step": 3270
    },
    {
      "epoch": 10.48,
      "learning_rate": 0.00014469293574724884,
      "loss": 0.0239,
      "step": 3280
    },
    {
      "epoch": 10.51,
      "learning_rate": 0.0001444562773636256,
      "loss": 0.0202,
      "step": 3290
    },
    {
      "epoch": 10.54,
      "learning_rate": 0.00014421961898000238,
      "loss": 0.0323,
      "step": 3300
    },
    {
      "epoch": 10.58,
      "learning_rate": 0.00014398296059637912,
      "loss": 0.0383,
      "step": 3310
    },
    {
      "epoch": 10.61,
      "learning_rate": 0.0001437463022127559,
      "loss": 0.0405,
      "step": 3320
    },
    {
      "epoch": 10.64,
      "learning_rate": 0.00014350964382913266,
      "loss": 0.0575,
      "step": 3330
    },
    {
      "epoch": 10.67,
      "learning_rate": 0.00014327298544550943,
      "loss": 0.0177,
      "step": 3340
    },
    {
      "epoch": 10.7,
      "learning_rate": 0.00014303632706188618,
      "loss": 0.015,
      "step": 3350
    },
    {
      "epoch": 10.73,
      "learning_rate": 0.00014279966867826292,
      "loss": 0.0085,
      "step": 3360
    },
    {
      "epoch": 10.77,
      "learning_rate": 0.0001425630102946397,
      "loss": 0.0035,
      "step": 3370
    },
    {
      "epoch": 10.8,
      "learning_rate": 0.00014232635191101646,
      "loss": 0.035,
      "step": 3380
    },
    {
      "epoch": 10.83,
      "learning_rate": 0.0001420896935273932,
      "loss": 0.0257,
      "step": 3390
    },
    {
      "epoch": 10.86,
      "learning_rate": 0.00014185303514376997,
      "loss": 0.0435,
      "step": 3400
    },
    {
      "epoch": 10.89,
      "learning_rate": 0.00014161637676014674,
      "loss": 0.0202,
      "step": 3410
    },
    {
      "epoch": 10.93,
      "learning_rate": 0.00014137971837652349,
      "loss": 0.0196,
      "step": 3420
    },
    {
      "epoch": 10.96,
      "learning_rate": 0.00014114305999290026,
      "loss": 0.0119,
      "step": 3430
    },
    {
      "epoch": 10.99,
      "learning_rate": 0.00014090640160927703,
      "loss": 0.0037,
      "step": 3440
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.9875,
      "eval_loss": 0.06855019927024841,
      "eval_runtime": 6.4278,
      "eval_samples_per_second": 62.229,
      "eval_steps_per_second": 7.779,
      "step": 3443
    },
    {
      "epoch": 11.02,
      "learning_rate": 0.0001406697432256538,
      "loss": 0.0182,
      "step": 3450
    },
    {
      "epoch": 11.05,
      "learning_rate": 0.00014043308484203054,
      "loss": 0.0152,
      "step": 3460
    },
    {
      "epoch": 11.09,
      "learning_rate": 0.00014019642645840728,
      "loss": 0.0647,
      "step": 3470
    },
    {
      "epoch": 11.12,
      "learning_rate": 0.00013995976807478405,
      "loss": 0.0197,
      "step": 3480
    },
    {
      "epoch": 11.15,
      "learning_rate": 0.00013972310969116082,
      "loss": 0.0118,
      "step": 3490
    },
    {
      "epoch": 11.18,
      "learning_rate": 0.00013948645130753756,
      "loss": 0.0111,
      "step": 3500
    },
    {
      "epoch": 11.21,
      "learning_rate": 0.00013924979292391433,
      "loss": 0.0521,
      "step": 3510
    },
    {
      "epoch": 11.25,
      "learning_rate": 0.0001390131345402911,
      "loss": 0.024,
      "step": 3520
    },
    {
      "epoch": 11.28,
      "learning_rate": 0.00013877647615666787,
      "loss": 0.0609,
      "step": 3530
    },
    {
      "epoch": 11.31,
      "learning_rate": 0.00013853981777304462,
      "loss": 0.0066,
      "step": 3540
    },
    {
      "epoch": 11.34,
      "learning_rate": 0.0001383031593894214,
      "loss": 0.0029,
      "step": 3550
    },
    {
      "epoch": 11.37,
      "learning_rate": 0.00013806650100579813,
      "loss": 0.0616,
      "step": 3560
    },
    {
      "epoch": 11.41,
      "learning_rate": 0.0001378298426221749,
      "loss": 0.026,
      "step": 3570
    },
    {
      "epoch": 11.44,
      "learning_rate": 0.00013759318423855164,
      "loss": 0.0362,
      "step": 3580
    },
    {
      "epoch": 11.47,
      "learning_rate": 0.0001373565258549284,
      "loss": 0.0199,
      "step": 3590
    },
    {
      "epoch": 11.5,
      "learning_rate": 0.00013711986747130518,
      "loss": 0.0463,
      "step": 3600
    },
    {
      "epoch": 11.53,
      "learning_rate": 0.00013688320908768193,
      "loss": 0.0164,
      "step": 3610
    },
    {
      "epoch": 11.57,
      "learning_rate": 0.0001366465507040587,
      "loss": 0.0013,
      "step": 3620
    },
    {
      "epoch": 11.6,
      "learning_rate": 0.00013640989232043547,
      "loss": 0.0082,
      "step": 3630
    },
    {
      "epoch": 11.63,
      "learning_rate": 0.00013617323393681224,
      "loss": 0.0024,
      "step": 3640
    },
    {
      "epoch": 11.66,
      "learning_rate": 0.00013593657555318898,
      "loss": 0.0384,
      "step": 3650
    },
    {
      "epoch": 11.69,
      "learning_rate": 0.00013569991716956575,
      "loss": 0.0283,
      "step": 3660
    },
    {
      "epoch": 11.73,
      "learning_rate": 0.0001354632587859425,
      "loss": 0.0261,
      "step": 3670
    },
    {
      "epoch": 11.76,
      "learning_rate": 0.00013522660040231926,
      "loss": 0.0245,
      "step": 3680
    },
    {
      "epoch": 11.79,
      "learning_rate": 0.000134989942018696,
      "loss": 0.0789,
      "step": 3690
    },
    {
      "epoch": 11.82,
      "learning_rate": 0.00013475328363507277,
      "loss": 0.0312,
      "step": 3700
    },
    {
      "epoch": 11.85,
      "learning_rate": 0.00013451662525144954,
      "loss": 0.053,
      "step": 3710
    },
    {
      "epoch": 11.88,
      "learning_rate": 0.0001342799668678263,
      "loss": 0.0202,
      "step": 3720
    },
    {
      "epoch": 11.92,
      "learning_rate": 0.00013404330848420306,
      "loss": 0.0081,
      "step": 3730
    },
    {
      "epoch": 11.95,
      "learning_rate": 0.00013380665010057983,
      "loss": 0.0511,
      "step": 3740
    },
    {
      "epoch": 11.98,
      "learning_rate": 0.0001335699917169566,
      "loss": 0.0369,
      "step": 3750
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9625,
      "eval_loss": 0.1564856767654419,
      "eval_runtime": 5.157,
      "eval_samples_per_second": 77.564,
      "eval_steps_per_second": 9.695,
      "step": 3756
    },
    {
      "epoch": 12.01,
      "learning_rate": 0.00013333333333333334,
      "loss": 0.0043,
      "step": 3760
    },
    {
      "epoch": 12.04,
      "learning_rate": 0.0001330966749497101,
      "loss": 0.0015,
      "step": 3770
    },
    {
      "epoch": 12.08,
      "learning_rate": 0.00013286001656608685,
      "loss": 0.0085,
      "step": 3780
    },
    {
      "epoch": 12.11,
      "learning_rate": 0.00013262335818246362,
      "loss": 0.0498,
      "step": 3790
    },
    {
      "epoch": 12.14,
      "learning_rate": 0.00013238669979884037,
      "loss": 0.0111,
      "step": 3800
    },
    {
      "epoch": 12.17,
      "learning_rate": 0.00013215004141521714,
      "loss": 0.0255,
      "step": 3810
    },
    {
      "epoch": 12.2,
      "learning_rate": 0.0001319133830315939,
      "loss": 0.0061,
      "step": 3820
    },
    {
      "epoch": 12.24,
      "learning_rate": 0.00013167672464797065,
      "loss": 0.0162,
      "step": 3830
    },
    {
      "epoch": 12.27,
      "learning_rate": 0.00013144006626434742,
      "loss": 0.0032,
      "step": 3840
    },
    {
      "epoch": 12.3,
      "learning_rate": 0.0001312034078807242,
      "loss": 0.0193,
      "step": 3850
    },
    {
      "epoch": 12.33,
      "learning_rate": 0.00013096674949710096,
      "loss": 0.0082,
      "step": 3860
    },
    {
      "epoch": 12.36,
      "learning_rate": 0.0001307300911134777,
      "loss": 0.0054,
      "step": 3870
    },
    {
      "epoch": 12.4,
      "learning_rate": 0.00013049343272985447,
      "loss": 0.0226,
      "step": 3880
    },
    {
      "epoch": 12.43,
      "learning_rate": 0.00013025677434623121,
      "loss": 0.0042,
      "step": 3890
    },
    {
      "epoch": 12.46,
      "learning_rate": 0.00013002011596260798,
      "loss": 0.0065,
      "step": 3900
    },
    {
      "epoch": 12.49,
      "learning_rate": 0.00012978345757898473,
      "loss": 0.008,
      "step": 3910
    },
    {
      "epoch": 12.52,
      "learning_rate": 0.0001295467991953615,
      "loss": 0.0209,
      "step": 3920
    },
    {
      "epoch": 12.56,
      "learning_rate": 0.00012931014081173827,
      "loss": 0.0192,
      "step": 3930
    },
    {
      "epoch": 12.59,
      "learning_rate": 0.000129073482428115,
      "loss": 0.0098,
      "step": 3940
    },
    {
      "epoch": 12.62,
      "learning_rate": 0.00012883682404449178,
      "loss": 0.0154,
      "step": 3950
    },
    {
      "epoch": 12.65,
      "learning_rate": 0.00012860016566086855,
      "loss": 0.0675,
      "step": 3960
    },
    {
      "epoch": 12.68,
      "learning_rate": 0.00012836350727724532,
      "loss": 0.0252,
      "step": 3970
    },
    {
      "epoch": 12.72,
      "learning_rate": 0.00012812684889362206,
      "loss": 0.037,
      "step": 3980
    },
    {
      "epoch": 12.75,
      "learning_rate": 0.00012789019050999883,
      "loss": 0.0148,
      "step": 3990
    },
    {
      "epoch": 12.78,
      "learning_rate": 0.00012765353212637558,
      "loss": 0.0041,
      "step": 4000
    },
    {
      "epoch": 12.81,
      "learning_rate": 0.00012741687374275235,
      "loss": 0.0283,
      "step": 4010
    },
    {
      "epoch": 12.84,
      "learning_rate": 0.0001271802153591291,
      "loss": 0.0428,
      "step": 4020
    },
    {
      "epoch": 12.88,
      "learning_rate": 0.00012694355697550586,
      "loss": 0.0165,
      "step": 4030
    },
    {
      "epoch": 12.91,
      "learning_rate": 0.00012670689859188263,
      "loss": 0.0263,
      "step": 4040
    },
    {
      "epoch": 12.94,
      "learning_rate": 0.0001264702402082594,
      "loss": 0.0365,
      "step": 4050
    },
    {
      "epoch": 12.97,
      "learning_rate": 0.00012623358182463614,
      "loss": 0.0016,
      "step": 4060
    },
    {
      "epoch": 13.0,
      "eval_accuracy": 0.9725,
      "eval_loss": 0.14009271562099457,
      "eval_runtime": 6.8166,
      "eval_samples_per_second": 58.68,
      "eval_steps_per_second": 7.335,
      "step": 4069
    },
    {
      "epoch": 13.0,
      "learning_rate": 0.0001259969234410129,
      "loss": 0.0087,
      "step": 4070
    },
    {
      "epoch": 13.04,
      "learning_rate": 0.00012576026505738968,
      "loss": 0.0028,
      "step": 4080
    },
    {
      "epoch": 13.07,
      "learning_rate": 0.00012552360667376642,
      "loss": 0.0354,
      "step": 4090
    },
    {
      "epoch": 13.1,
      "learning_rate": 0.00012528694829014317,
      "loss": 0.0096,
      "step": 4100
    },
    {
      "epoch": 13.13,
      "learning_rate": 0.00012505028990651994,
      "loss": 0.0302,
      "step": 4110
    },
    {
      "epoch": 13.16,
      "learning_rate": 0.0001248136315228967,
      "loss": 0.0679,
      "step": 4120
    },
    {
      "epoch": 13.19,
      "learning_rate": 0.00012457697313927345,
      "loss": 0.0313,
      "step": 4130
    },
    {
      "epoch": 13.23,
      "learning_rate": 0.00012434031475565022,
      "loss": 0.048,
      "step": 4140
    },
    {
      "epoch": 13.26,
      "learning_rate": 0.000124103656372027,
      "loss": 0.014,
      "step": 4150
    },
    {
      "epoch": 13.29,
      "learning_rate": 0.00012386699798840376,
      "loss": 0.0143,
      "step": 4160
    },
    {
      "epoch": 13.32,
      "learning_rate": 0.0001236303396047805,
      "loss": 0.0013,
      "step": 4170
    },
    {
      "epoch": 13.35,
      "learning_rate": 0.00012339368122115727,
      "loss": 0.0005,
      "step": 4180
    },
    {
      "epoch": 13.39,
      "learning_rate": 0.00012315702283753404,
      "loss": 0.0017,
      "step": 4190
    },
    {
      "epoch": 13.42,
      "learning_rate": 0.00012292036445391079,
      "loss": 0.0012,
      "step": 4200
    },
    {
      "epoch": 13.45,
      "learning_rate": 0.00012268370607028753,
      "loss": 0.0055,
      "step": 4210
    },
    {
      "epoch": 13.48,
      "learning_rate": 0.0001224470476866643,
      "loss": 0.0103,
      "step": 4220
    },
    {
      "epoch": 13.51,
      "learning_rate": 0.00012221038930304107,
      "loss": 0.0095,
      "step": 4230
    },
    {
      "epoch": 13.55,
      "learning_rate": 0.00012197373091941781,
      "loss": 0.0175,
      "step": 4240
    },
    {
      "epoch": 13.58,
      "learning_rate": 0.00012173707253579458,
      "loss": 0.0155,
      "step": 4250
    },
    {
      "epoch": 13.61,
      "learning_rate": 0.00012150041415217135,
      "loss": 0.0169,
      "step": 4260
    },
    {
      "epoch": 13.64,
      "learning_rate": 0.00012126375576854812,
      "loss": 0.0419,
      "step": 4270
    },
    {
      "epoch": 13.67,
      "learning_rate": 0.00012102709738492486,
      "loss": 0.0281,
      "step": 4280
    },
    {
      "epoch": 13.71,
      "learning_rate": 0.00012079043900130162,
      "loss": 0.0124,
      "step": 4290
    },
    {
      "epoch": 13.74,
      "learning_rate": 0.00012055378061767839,
      "loss": 0.0054,
      "step": 4300
    },
    {
      "epoch": 13.77,
      "learning_rate": 0.00012031712223405516,
      "loss": 0.0682,
      "step": 4310
    },
    {
      "epoch": 13.8,
      "learning_rate": 0.0001200804638504319,
      "loss": 0.0139,
      "step": 4320
    },
    {
      "epoch": 13.83,
      "learning_rate": 0.00011984380546680867,
      "loss": 0.0018,
      "step": 4330
    },
    {
      "epoch": 13.87,
      "learning_rate": 0.00011960714708318543,
      "loss": 0.0258,
      "step": 4340
    },
    {
      "epoch": 13.9,
      "learning_rate": 0.00011937048869956217,
      "loss": 0.0194,
      "step": 4350
    },
    {
      "epoch": 13.93,
      "learning_rate": 0.00011913383031593894,
      "loss": 0.0132,
      "step": 4360
    },
    {
      "epoch": 13.96,
      "learning_rate": 0.00011889717193231571,
      "loss": 0.0024,
      "step": 4370
    },
    {
      "epoch": 13.99,
      "learning_rate": 0.00011866051354869248,
      "loss": 0.0006,
      "step": 4380
    },
    {
      "epoch": 14.0,
      "eval_accuracy": 0.9925,
      "eval_loss": 0.1054958701133728,
      "eval_runtime": 5.0787,
      "eval_samples_per_second": 78.761,
      "eval_steps_per_second": 9.845,
      "step": 4382
    },
    {
      "epoch": 14.03,
      "learning_rate": 0.00011842385516506922,
      "loss": 0.0259,
      "step": 4390
    },
    {
      "epoch": 14.06,
      "learning_rate": 0.00011818719678144598,
      "loss": 0.0004,
      "step": 4400
    },
    {
      "epoch": 14.09,
      "learning_rate": 0.00011795053839782275,
      "loss": 0.0318,
      "step": 4410
    },
    {
      "epoch": 14.12,
      "learning_rate": 0.00011771388001419952,
      "loss": 0.0292,
      "step": 4420
    },
    {
      "epoch": 14.15,
      "learning_rate": 0.00011747722163057626,
      "loss": 0.0068,
      "step": 4430
    },
    {
      "epoch": 14.19,
      "learning_rate": 0.00011724056324695303,
      "loss": 0.0118,
      "step": 4440
    },
    {
      "epoch": 14.22,
      "learning_rate": 0.00011700390486332979,
      "loss": 0.0136,
      "step": 4450
    },
    {
      "epoch": 14.25,
      "learning_rate": 0.00011676724647970656,
      "loss": 0.0937,
      "step": 4460
    },
    {
      "epoch": 14.28,
      "learning_rate": 0.0001165305880960833,
      "loss": 0.0369,
      "step": 4470
    },
    {
      "epoch": 14.31,
      "learning_rate": 0.00011629392971246007,
      "loss": 0.0098,
      "step": 4480
    },
    {
      "epoch": 14.35,
      "learning_rate": 0.00011605727132883683,
      "loss": 0.0144,
      "step": 4490
    },
    {
      "epoch": 14.38,
      "learning_rate": 0.00011582061294521359,
      "loss": 0.0939,
      "step": 4500
    },
    {
      "epoch": 14.41,
      "learning_rate": 0.00011558395456159034,
      "loss": 0.0064,
      "step": 4510
    },
    {
      "epoch": 14.44,
      "learning_rate": 0.00011534729617796711,
      "loss": 0.0376,
      "step": 4520
    },
    {
      "epoch": 14.47,
      "learning_rate": 0.00011511063779434388,
      "loss": 0.0066,
      "step": 4530
    },
    {
      "epoch": 14.5,
      "learning_rate": 0.00011487397941072063,
      "loss": 0.0348,
      "step": 4540
    },
    {
      "epoch": 14.54,
      "learning_rate": 0.00011463732102709738,
      "loss": 0.0144,
      "step": 4550
    },
    {
      "epoch": 14.57,
      "learning_rate": 0.00011440066264347415,
      "loss": 0.005,
      "step": 4560
    },
    {
      "epoch": 14.6,
      "learning_rate": 0.00011416400425985092,
      "loss": 0.0041,
      "step": 4570
    },
    {
      "epoch": 14.63,
      "learning_rate": 0.00011392734587622766,
      "loss": 0.086,
      "step": 4580
    },
    {
      "epoch": 14.66,
      "learning_rate": 0.00011369068749260443,
      "loss": 0.0031,
      "step": 4590
    },
    {
      "epoch": 14.7,
      "learning_rate": 0.00011345402910898119,
      "loss": 0.0158,
      "step": 4600
    },
    {
      "epoch": 14.73,
      "learning_rate": 0.00011321737072535793,
      "loss": 0.0056,
      "step": 4610
    },
    {
      "epoch": 14.76,
      "learning_rate": 0.0001129807123417347,
      "loss": 0.0021,
      "step": 4620
    },
    {
      "epoch": 14.79,
      "learning_rate": 0.00011274405395811147,
      "loss": 0.012,
      "step": 4630
    },
    {
      "epoch": 14.82,
      "learning_rate": 0.00011250739557448824,
      "loss": 0.0001,
      "step": 4640
    },
    {
      "epoch": 14.86,
      "learning_rate": 0.00011227073719086499,
      "loss": 0.0003,
      "step": 4650
    },
    {
      "epoch": 14.89,
      "learning_rate": 0.00011203407880724174,
      "loss": 0.001,
      "step": 4660
    },
    {
      "epoch": 14.92,
      "learning_rate": 0.00011179742042361851,
      "loss": 0.0003,
      "step": 4670
    },
    {
      "epoch": 14.95,
      "learning_rate": 0.00011156076203999528,
      "loss": 0.0132,
      "step": 4680
    },
    {
      "epoch": 14.98,
      "learning_rate": 0.00011132410365637203,
      "loss": 0.0001,
      "step": 4690
    },
    {
      "epoch": 15.0,
      "eval_accuracy": 0.9925,
      "eval_loss": 0.08063908666372299,
      "eval_runtime": 5.1103,
      "eval_samples_per_second": 78.274,
      "eval_steps_per_second": 9.784,
      "step": 4695
    },
    {
      "epoch": 15.02,
      "learning_rate": 0.0001110874452727488,
      "loss": 0.0015,
      "step": 4700
    },
    {
      "epoch": 15.05,
      "learning_rate": 0.00011085078688912555,
      "loss": 0.0056,
      "step": 4710
    },
    {
      "epoch": 15.08,
      "learning_rate": 0.00011061412850550232,
      "loss": 0.0119,
      "step": 4720
    },
    {
      "epoch": 15.11,
      "learning_rate": 0.00011037747012187907,
      "loss": 0.0031,
      "step": 4730
    },
    {
      "epoch": 15.14,
      "learning_rate": 0.00011014081173825584,
      "loss": 0.031,
      "step": 4740
    },
    {
      "epoch": 15.18,
      "learning_rate": 0.0001099041533546326,
      "loss": 0.0055,
      "step": 4750
    },
    {
      "epoch": 15.21,
      "learning_rate": 0.00010966749497100935,
      "loss": 0.0176,
      "step": 4760
    },
    {
      "epoch": 15.24,
      "learning_rate": 0.0001094308365873861,
      "loss": 0.0115,
      "step": 4770
    },
    {
      "epoch": 15.27,
      "learning_rate": 0.00010919417820376287,
      "loss": 0.0084,
      "step": 4780
    },
    {
      "epoch": 15.3,
      "learning_rate": 0.00010895751982013964,
      "loss": 0.0255,
      "step": 4790
    },
    {
      "epoch": 15.34,
      "learning_rate": 0.00010872086143651639,
      "loss": 0.0007,
      "step": 4800
    },
    {
      "epoch": 15.37,
      "learning_rate": 0.00010848420305289316,
      "loss": 0.003,
      "step": 4810
    },
    {
      "epoch": 15.4,
      "learning_rate": 0.00010824754466926991,
      "loss": 0.0016,
      "step": 4820
    },
    {
      "epoch": 15.43,
      "learning_rate": 0.00010801088628564668,
      "loss": 0.0156,
      "step": 4830
    },
    {
      "epoch": 15.46,
      "learning_rate": 0.00010777422790202343,
      "loss": 0.0531,
      "step": 4840
    },
    {
      "epoch": 15.5,
      "learning_rate": 0.0001075375695184002,
      "loss": 0.0154,
      "step": 4850
    },
    {
      "epoch": 15.53,
      "learning_rate": 0.00010730091113477697,
      "loss": 0.0105,
      "step": 4860
    },
    {
      "epoch": 15.56,
      "learning_rate": 0.00010706425275115371,
      "loss": 0.0094,
      "step": 4870
    },
    {
      "epoch": 15.59,
      "learning_rate": 0.00010682759436753047,
      "loss": 0.0165,
      "step": 4880
    },
    {
      "epoch": 15.62,
      "learning_rate": 0.00010659093598390724,
      "loss": 0.061,
      "step": 4890
    },
    {
      "epoch": 15.65,
      "learning_rate": 0.000106354277600284,
      "loss": 0.0035,
      "step": 4900
    },
    {
      "epoch": 15.69,
      "learning_rate": 0.00010611761921666075,
      "loss": 0.0003,
      "step": 4910
    },
    {
      "epoch": 15.72,
      "learning_rate": 0.00010588096083303752,
      "loss": 0.0038,
      "step": 4920
    },
    {
      "epoch": 15.75,
      "learning_rate": 0.00010564430244941428,
      "loss": 0.0002,
      "step": 4930
    },
    {
      "epoch": 15.78,
      "learning_rate": 0.00010540764406579105,
      "loss": 0.0002,
      "step": 4940
    },
    {
      "epoch": 15.81,
      "learning_rate": 0.00010517098568216779,
      "loss": 0.0001,
      "step": 4950
    },
    {
      "epoch": 15.85,
      "learning_rate": 0.00010493432729854456,
      "loss": 0.0005,
      "step": 4960
    },
    {
      "epoch": 15.88,
      "learning_rate": 0.00010469766891492131,
      "loss": 0.0058,
      "step": 4970
    },
    {
      "epoch": 15.91,
      "learning_rate": 0.00010446101053129808,
      "loss": 0.005,
      "step": 4980
    },
    {
      "epoch": 15.94,
      "learning_rate": 0.00010422435214767483,
      "loss": 0.0057,
      "step": 4990
    },
    {
      "epoch": 15.97,
      "learning_rate": 0.0001039876937640516,
      "loss": 0.0003,
      "step": 5000
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.9925,
      "eval_loss": 0.08654958754777908,
      "eval_runtime": 6.5529,
      "eval_samples_per_second": 61.042,
      "eval_steps_per_second": 7.63,
      "step": 5008
    },
    {
      "epoch": 16.01,
      "learning_rate": 0.00010375103538042837,
      "loss": 0.0026,
      "step": 5010
    },
    {
      "epoch": 16.04,
      "learning_rate": 0.00010351437699680511,
      "loss": 0.0072,
      "step": 5020
    },
    {
      "epoch": 16.07,
      "learning_rate": 0.00010327771861318187,
      "loss": 0.001,
      "step": 5030
    },
    {
      "epoch": 16.1,
      "learning_rate": 0.00010304106022955864,
      "loss": 0.0035,
      "step": 5040
    },
    {
      "epoch": 16.13,
      "learning_rate": 0.0001028044018459354,
      "loss": 0.0001,
      "step": 5050
    },
    {
      "epoch": 16.17,
      "learning_rate": 0.00010259140930067448,
      "loss": 0.0183,
      "step": 5060
    },
    {
      "epoch": 16.2,
      "learning_rate": 0.00010235475091705124,
      "loss": 0.011,
      "step": 5070
    },
    {
      "epoch": 16.23,
      "learning_rate": 0.00010211809253342801,
      "loss": 0.0002,
      "step": 5080
    },
    {
      "epoch": 16.26,
      "learning_rate": 0.00010188143414980475,
      "loss": 0.0245,
      "step": 5090
    },
    {
      "epoch": 16.29,
      "learning_rate": 0.00010164477576618152,
      "loss": 0.0009,
      "step": 5100
    },
    {
      "epoch": 16.33,
      "learning_rate": 0.00010140811738255829,
      "loss": 0.0003,
      "step": 5110
    },
    {
      "epoch": 16.36,
      "learning_rate": 0.00010117145899893504,
      "loss": 0.0031,
      "step": 5120
    },
    {
      "epoch": 16.39,
      "learning_rate": 0.00010093480061531179,
      "loss": 0.0001,
      "step": 5130
    },
    {
      "epoch": 16.42,
      "learning_rate": 0.00010069814223168856,
      "loss": 0.0132,
      "step": 5140
    },
    {
      "epoch": 16.45,
      "learning_rate": 0.00010046148384806533,
      "loss": 0.0185,
      "step": 5150
    },
    {
      "epoch": 16.49,
      "learning_rate": 0.00010022482546444207,
      "loss": 0.0048,
      "step": 5160
    },
    {
      "epoch": 16.52,
      "learning_rate": 9.998816708081884e-05,
      "loss": 0.0159,
      "step": 5170
    },
    {
      "epoch": 16.55,
      "learning_rate": 9.97515086971956e-05,
      "loss": 0.0681,
      "step": 5180
    },
    {
      "epoch": 16.58,
      "learning_rate": 9.951485031357236e-05,
      "loss": 0.0237,
      "step": 5190
    },
    {
      "epoch": 16.61,
      "learning_rate": 9.927819192994913e-05,
      "loss": 0.0213,
      "step": 5200
    },
    {
      "epoch": 16.65,
      "learning_rate": 9.904153354632588e-05,
      "loss": 0.0139,
      "step": 5210
    },
    {
      "epoch": 16.68,
      "learning_rate": 9.880487516270265e-05,
      "loss": 0.0224,
      "step": 5220
    },
    {
      "epoch": 16.71,
      "learning_rate": 9.85682167790794e-05,
      "loss": 0.008,
      "step": 5230
    },
    {
      "epoch": 16.74,
      "learning_rate": 9.833155839545617e-05,
      "loss": 0.0013,
      "step": 5240
    },
    {
      "epoch": 16.77,
      "learning_rate": 9.809490001183292e-05,
      "loss": 0.0317,
      "step": 5250
    },
    {
      "epoch": 16.81,
      "learning_rate": 9.785824162820968e-05,
      "loss": 0.0001,
      "step": 5260
    },
    {
      "epoch": 16.84,
      "learning_rate": 9.762158324458645e-05,
      "loss": 0.0058,
      "step": 5270
    },
    {
      "epoch": 16.87,
      "learning_rate": 9.73849248609632e-05,
      "loss": 0.0044,
      "step": 5280
    },
    {
      "epoch": 16.9,
      "learning_rate": 9.714826647733996e-05,
      "loss": 0.0175,
      "step": 5290
    },
    {
      "epoch": 16.93,
      "learning_rate": 9.691160809371672e-05,
      "loss": 0.0028,
      "step": 5300
    },
    {
      "epoch": 16.96,
      "learning_rate": 9.667494971009349e-05,
      "loss": 0.0001,
      "step": 5310
    },
    {
      "epoch": 17.0,
      "learning_rate": 9.643829132647025e-05,
      "loss": 0.0035,
      "step": 5320
    },
    {
      "epoch": 17.0,
      "eval_accuracy": 0.9825,
      "eval_loss": 0.09907230734825134,
      "eval_runtime": 5.2571,
      "eval_samples_per_second": 76.088,
      "eval_steps_per_second": 9.511,
      "step": 5321
    },
    {
      "epoch": 17.03,
      "learning_rate": 9.620163294284702e-05,
      "loss": 0.0286,
      "step": 5330
    },
    {
      "epoch": 17.06,
      "learning_rate": 9.596497455922376e-05,
      "loss": 0.0002,
      "step": 5340
    },
    {
      "epoch": 17.09,
      "learning_rate": 9.572831617560053e-05,
      "loss": 0.0009,
      "step": 5350
    },
    {
      "epoch": 17.12,
      "learning_rate": 9.549165779197728e-05,
      "loss": 0.0016,
      "step": 5360
    },
    {
      "epoch": 17.16,
      "learning_rate": 9.525499940835405e-05,
      "loss": 0.0021,
      "step": 5370
    },
    {
      "epoch": 17.19,
      "learning_rate": 9.501834102473081e-05,
      "loss": 0.0001,
      "step": 5380
    },
    {
      "epoch": 17.22,
      "learning_rate": 9.478168264110757e-05,
      "loss": 0.0012,
      "step": 5390
    },
    {
      "epoch": 17.25,
      "learning_rate": 9.454502425748432e-05,
      "loss": 0.0209,
      "step": 5400
    },
    {
      "epoch": 17.28,
      "learning_rate": 9.430836587386108e-05,
      "loss": 0.0003,
      "step": 5410
    },
    {
      "epoch": 17.32,
      "learning_rate": 9.407170749023785e-05,
      "loss": 0.0002,
      "step": 5420
    },
    {
      "epoch": 17.35,
      "learning_rate": 9.38350491066146e-05,
      "loss": 0.0002,
      "step": 5430
    },
    {
      "epoch": 17.38,
      "learning_rate": 9.359839072299136e-05,
      "loss": 0.0076,
      "step": 5440
    },
    {
      "epoch": 17.41,
      "learning_rate": 9.336173233936812e-05,
      "loss": 0.0001,
      "step": 5450
    },
    {
      "epoch": 17.44,
      "learning_rate": 9.312507395574489e-05,
      "loss": 0.0001,
      "step": 5460
    },
    {
      "epoch": 17.48,
      "learning_rate": 9.288841557212165e-05,
      "loss": 0.0026,
      "step": 5470
    },
    {
      "epoch": 17.51,
      "learning_rate": 9.265175718849842e-05,
      "loss": 0.0004,
      "step": 5480
    },
    {
      "epoch": 17.54,
      "learning_rate": 9.241509880487517e-05,
      "loss": 0.0,
      "step": 5490
    },
    {
      "epoch": 17.57,
      "learning_rate": 9.217844042125193e-05,
      "loss": 0.0017,
      "step": 5500
    },
    {
      "epoch": 17.6,
      "learning_rate": 9.194178203762869e-05,
      "loss": 0.0119,
      "step": 5510
    },
    {
      "epoch": 17.64,
      "learning_rate": 9.170512365400544e-05,
      "loss": 0.0033,
      "step": 5520
    },
    {
      "epoch": 17.67,
      "learning_rate": 9.146846527038221e-05,
      "loss": 0.0004,
      "step": 5530
    },
    {
      "epoch": 17.7,
      "learning_rate": 9.123180688675897e-05,
      "loss": 0.0104,
      "step": 5540
    },
    {
      "epoch": 17.73,
      "learning_rate": 9.099514850313572e-05,
      "loss": 0.0018,
      "step": 5550
    },
    {
      "epoch": 17.76,
      "learning_rate": 9.075849011951248e-05,
      "loss": 0.0203,
      "step": 5560
    },
    {
      "epoch": 17.8,
      "learning_rate": 9.052183173588925e-05,
      "loss": 0.0005,
      "step": 5570
    },
    {
      "epoch": 17.83,
      "learning_rate": 9.028517335226601e-05,
      "loss": 0.0108,
      "step": 5580
    },
    {
      "epoch": 17.86,
      "learning_rate": 9.004851496864278e-05,
      "loss": 0.0192,
      "step": 5590
    },
    {
      "epoch": 17.89,
      "learning_rate": 8.981185658501953e-05,
      "loss": 0.0023,
      "step": 5600
    },
    {
      "epoch": 17.92,
      "learning_rate": 8.957519820139629e-05,
      "loss": 0.0508,
      "step": 5610
    },
    {
      "epoch": 17.96,
      "learning_rate": 8.933853981777305e-05,
      "loss": 0.0007,
      "step": 5620
    },
    {
      "epoch": 17.99,
      "learning_rate": 8.910188143414982e-05,
      "loss": 0.0001,
      "step": 5630
    },
    {
      "epoch": 18.0,
      "eval_accuracy": 0.9875,
      "eval_loss": 0.12689071893692017,
      "eval_runtime": 6.9648,
      "eval_samples_per_second": 57.431,
      "eval_steps_per_second": 7.179,
      "step": 5634
    },
    {
      "epoch": 18.02,
      "learning_rate": 8.886522305052657e-05,
      "loss": 0.001,
      "step": 5640
    },
    {
      "epoch": 18.05,
      "learning_rate": 8.862856466690333e-05,
      "loss": 0.0042,
      "step": 5650
    },
    {
      "epoch": 18.08,
      "learning_rate": 8.839190628328009e-05,
      "loss": 0.0254,
      "step": 5660
    },
    {
      "epoch": 18.12,
      "learning_rate": 8.815524789965684e-05,
      "loss": 0.0141,
      "step": 5670
    },
    {
      "epoch": 18.15,
      "learning_rate": 8.791858951603361e-05,
      "loss": 0.0008,
      "step": 5680
    },
    {
      "epoch": 18.18,
      "learning_rate": 8.768193113241037e-05,
      "loss": 0.0015,
      "step": 5690
    },
    {
      "epoch": 18.21,
      "learning_rate": 8.744527274878714e-05,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 18.24,
      "learning_rate": 8.720861436516388e-05,
      "loss": 0.0038,
      "step": 5710
    },
    {
      "epoch": 18.27,
      "learning_rate": 8.697195598154065e-05,
      "loss": 0.0111,
      "step": 5720
    },
    {
      "epoch": 18.31,
      "learning_rate": 8.673529759791741e-05,
      "loss": 0.001,
      "step": 5730
    },
    {
      "epoch": 18.34,
      "learning_rate": 8.649863921429418e-05,
      "loss": 0.0003,
      "step": 5740
    },
    {
      "epoch": 18.37,
      "learning_rate": 8.626198083067093e-05,
      "loss": 0.0009,
      "step": 5750
    },
    {
      "epoch": 18.4,
      "learning_rate": 8.602532244704769e-05,
      "loss": 0.0002,
      "step": 5760
    },
    {
      "epoch": 18.43,
      "learning_rate": 8.578866406342445e-05,
      "loss": 0.0002,
      "step": 5770
    },
    {
      "epoch": 18.47,
      "learning_rate": 8.55520056798012e-05,
      "loss": 0.0001,
      "step": 5780
    },
    {
      "epoch": 18.5,
      "learning_rate": 8.531534729617797e-05,
      "loss": 0.0117,
      "step": 5790
    },
    {
      "epoch": 18.53,
      "learning_rate": 8.507868891255473e-05,
      "loss": 0.0009,
      "step": 5800
    },
    {
      "epoch": 18.56,
      "learning_rate": 8.48420305289315e-05,
      "loss": 0.0185,
      "step": 5810
    },
    {
      "epoch": 18.59,
      "learning_rate": 8.460537214530824e-05,
      "loss": 0.001,
      "step": 5820
    },
    {
      "epoch": 18.63,
      "learning_rate": 8.436871376168501e-05,
      "loss": 0.0014,
      "step": 5830
    },
    {
      "epoch": 18.66,
      "learning_rate": 8.413205537806177e-05,
      "loss": 0.0001,
      "step": 5840
    },
    {
      "epoch": 18.69,
      "learning_rate": 8.389539699443854e-05,
      "loss": 0.0014,
      "step": 5850
    },
    {
      "epoch": 18.72,
      "learning_rate": 8.36587386108153e-05,
      "loss": 0.0001,
      "step": 5860
    },
    {
      "epoch": 18.75,
      "learning_rate": 8.342208022719205e-05,
      "loss": 0.0053,
      "step": 5870
    },
    {
      "epoch": 18.79,
      "learning_rate": 8.318542184356881e-05,
      "loss": 0.0055,
      "step": 5880
    },
    {
      "epoch": 18.82,
      "learning_rate": 8.294876345994558e-05,
      "loss": 0.0,
      "step": 5890
    },
    {
      "epoch": 18.85,
      "learning_rate": 8.271210507632233e-05,
      "loss": 0.0011,
      "step": 5900
    },
    {
      "epoch": 18.88,
      "learning_rate": 8.247544669269909e-05,
      "loss": 0.0,
      "step": 5910
    },
    {
      "epoch": 18.91,
      "learning_rate": 8.223878830907585e-05,
      "loss": 0.0003,
      "step": 5920
    },
    {
      "epoch": 18.95,
      "learning_rate": 8.20021299254526e-05,
      "loss": 0.0001,
      "step": 5930
    },
    {
      "epoch": 18.98,
      "learning_rate": 8.176547154182937e-05,
      "loss": 0.0088,
      "step": 5940
    },
    {
      "epoch": 19.0,
      "eval_accuracy": 0.9925,
      "eval_loss": 0.10001015663146973,
      "eval_runtime": 5.2536,
      "eval_samples_per_second": 76.138,
      "eval_steps_per_second": 9.517,
      "step": 5947
    },
    {
      "epoch": 19.01,
      "learning_rate": 8.152881315820613e-05,
      "loss": 0.0001,
      "step": 5950
    },
    {
      "epoch": 19.04,
      "learning_rate": 8.12921547745829e-05,
      "loss": 0.0,
      "step": 5960
    },
    {
      "epoch": 19.07,
      "learning_rate": 8.105549639095966e-05,
      "loss": 0.0045,
      "step": 5970
    },
    {
      "epoch": 19.11,
      "learning_rate": 8.081883800733641e-05,
      "loss": 0.0,
      "step": 5980
    },
    {
      "epoch": 19.14,
      "learning_rate": 8.058217962371317e-05,
      "loss": 0.0,
      "step": 5990
    },
    {
      "epoch": 19.17,
      "learning_rate": 8.034552124008994e-05,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 19.2,
      "learning_rate": 8.01088628564667e-05,
      "loss": 0.0,
      "step": 6010
    },
    {
      "epoch": 19.23,
      "learning_rate": 7.987220447284345e-05,
      "loss": 0.0,
      "step": 6020
    },
    {
      "epoch": 19.27,
      "learning_rate": 7.963554608922021e-05,
      "loss": 0.0,
      "step": 6030
    },
    {
      "epoch": 19.3,
      "learning_rate": 7.939888770559697e-05,
      "loss": 0.0,
      "step": 6040
    },
    {
      "epoch": 19.33,
      "learning_rate": 7.916222932197374e-05,
      "loss": 0.0005,
      "step": 6050
    },
    {
      "epoch": 19.36,
      "learning_rate": 7.892557093835049e-05,
      "loss": 0.0,
      "step": 6060
    },
    {
      "epoch": 19.39,
      "learning_rate": 7.868891255472726e-05,
      "loss": 0.0224,
      "step": 6070
    },
    {
      "epoch": 19.42,
      "learning_rate": 7.845225417110402e-05,
      "loss": 0.0,
      "step": 6080
    },
    {
      "epoch": 19.46,
      "learning_rate": 7.821559578748077e-05,
      "loss": 0.0,
      "step": 6090
    },
    {
      "epoch": 19.49,
      "learning_rate": 7.797893740385753e-05,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 19.52,
      "learning_rate": 7.77422790202343e-05,
      "loss": 0.0,
      "step": 6110
    },
    {
      "epoch": 19.55,
      "learning_rate": 7.750562063661106e-05,
      "loss": 0.0,
      "step": 6120
    },
    {
      "epoch": 19.58,
      "learning_rate": 7.726896225298781e-05,
      "loss": 0.0,
      "step": 6130
    },
    {
      "epoch": 19.62,
      "learning_rate": 7.703230386936457e-05,
      "loss": 0.0,
      "step": 6140
    },
    {
      "epoch": 19.65,
      "learning_rate": 7.679564548574134e-05,
      "loss": 0.0005,
      "step": 6150
    },
    {
      "epoch": 19.68,
      "learning_rate": 7.65589871021181e-05,
      "loss": 0.0001,
      "step": 6160
    },
    {
      "epoch": 19.71,
      "learning_rate": 7.632232871849485e-05,
      "loss": 0.0,
      "step": 6170
    },
    {
      "epoch": 19.74,
      "learning_rate": 7.608567033487162e-05,
      "loss": 0.0,
      "step": 6180
    },
    {
      "epoch": 19.78,
      "learning_rate": 7.584901195124837e-05,
      "loss": 0.0005,
      "step": 6190
    },
    {
      "epoch": 19.81,
      "learning_rate": 7.561235356762514e-05,
      "loss": 0.0001,
      "step": 6200
    },
    {
      "epoch": 19.84,
      "learning_rate": 7.537569518400189e-05,
      "loss": 0.0001,
      "step": 6210
    },
    {
      "epoch": 19.87,
      "learning_rate": 7.513903680037866e-05,
      "loss": 0.0064,
      "step": 6220
    },
    {
      "epoch": 19.9,
      "learning_rate": 7.490237841675542e-05,
      "loss": 0.0204,
      "step": 6230
    },
    {
      "epoch": 19.94,
      "learning_rate": 7.466572003313218e-05,
      "loss": 0.0061,
      "step": 6240
    },
    {
      "epoch": 19.97,
      "learning_rate": 7.442906164950893e-05,
      "loss": 0.0001,
      "step": 6250
    },
    {
      "epoch": 20.0,
      "learning_rate": 7.41924032658857e-05,
      "loss": 0.0,
      "step": 6260
    },
    {
      "epoch": 20.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.09493375569581985,
      "eval_runtime": 6.1355,
      "eval_samples_per_second": 65.194,
      "eval_steps_per_second": 8.149,
      "step": 6260
    },
    {
      "epoch": 20.03,
      "learning_rate": 7.395574488226246e-05,
      "loss": 0.0001,
      "step": 6270
    },
    {
      "epoch": 20.06,
      "learning_rate": 7.371908649863923e-05,
      "loss": 0.0442,
      "step": 6280
    },
    {
      "epoch": 20.1,
      "learning_rate": 7.348242811501597e-05,
      "loss": 0.0001,
      "step": 6290
    },
    {
      "epoch": 20.13,
      "learning_rate": 7.324576973139273e-05,
      "loss": 0.0,
      "step": 6300
    },
    {
      "epoch": 20.16,
      "learning_rate": 7.30091113477695e-05,
      "loss": 0.0011,
      "step": 6310
    },
    {
      "epoch": 20.19,
      "learning_rate": 7.277245296414625e-05,
      "loss": 0.0,
      "step": 6320
    },
    {
      "epoch": 20.22,
      "learning_rate": 7.253579458052302e-05,
      "loss": 0.0,
      "step": 6330
    },
    {
      "epoch": 20.26,
      "learning_rate": 7.229913619689978e-05,
      "loss": 0.0,
      "step": 6340
    },
    {
      "epoch": 20.29,
      "learning_rate": 7.206247781327654e-05,
      "loss": 0.0016,
      "step": 6350
    },
    {
      "epoch": 20.32,
      "learning_rate": 7.182581942965329e-05,
      "loss": 0.0,
      "step": 6360
    },
    {
      "epoch": 20.35,
      "learning_rate": 7.158916104603006e-05,
      "loss": 0.0,
      "step": 6370
    },
    {
      "epoch": 20.38,
      "learning_rate": 7.135250266240682e-05,
      "loss": 0.0,
      "step": 6380
    },
    {
      "epoch": 20.42,
      "learning_rate": 7.111584427878359e-05,
      "loss": 0.0071,
      "step": 6390
    },
    {
      "epoch": 20.45,
      "learning_rate": 7.087918589516033e-05,
      "loss": 0.0,
      "step": 6400
    },
    {
      "epoch": 20.48,
      "learning_rate": 7.06425275115371e-05,
      "loss": 0.0014,
      "step": 6410
    },
    {
      "epoch": 20.51,
      "learning_rate": 7.040586912791386e-05,
      "loss": 0.0077,
      "step": 6420
    },
    {
      "epoch": 20.54,
      "learning_rate": 7.016921074429062e-05,
      "loss": 0.0022,
      "step": 6430
    },
    {
      "epoch": 20.58,
      "learning_rate": 6.993255236066739e-05,
      "loss": 0.0004,
      "step": 6440
    },
    {
      "epoch": 20.61,
      "learning_rate": 6.969589397704414e-05,
      "loss": 0.0,
      "step": 6450
    },
    {
      "epoch": 20.64,
      "learning_rate": 6.94592355934209e-05,
      "loss": 0.0,
      "step": 6460
    },
    {
      "epoch": 20.67,
      "learning_rate": 6.922257720979765e-05,
      "loss": 0.0014,
      "step": 6470
    },
    {
      "epoch": 20.7,
      "learning_rate": 6.898591882617442e-05,
      "loss": 0.0102,
      "step": 6480
    },
    {
      "epoch": 20.73,
      "learning_rate": 6.874926044255118e-05,
      "loss": 0.0305,
      "step": 6490
    },
    {
      "epoch": 20.77,
      "learning_rate": 6.851260205892794e-05,
      "loss": 0.0079,
      "step": 6500
    },
    {
      "epoch": 20.8,
      "learning_rate": 6.82759436753047e-05,
      "loss": 0.0,
      "step": 6510
    },
    {
      "epoch": 20.83,
      "learning_rate": 6.803928529168146e-05,
      "loss": 0.0208,
      "step": 6520
    },
    {
      "epoch": 20.86,
      "learning_rate": 6.780262690805822e-05,
      "loss": 0.0001,
      "step": 6530
    },
    {
      "epoch": 20.89,
      "learning_rate": 6.756596852443499e-05,
      "loss": 0.0,
      "step": 6540
    },
    {
      "epoch": 20.93,
      "learning_rate": 6.732931014081175e-05,
      "loss": 0.0001,
      "step": 6550
    },
    {
      "epoch": 20.96,
      "learning_rate": 6.70926517571885e-05,
      "loss": 0.0001,
      "step": 6560
    },
    {
      "epoch": 20.99,
      "learning_rate": 6.685599337356526e-05,
      "loss": 0.0,
      "step": 6570
    },
    {
      "epoch": 21.0,
      "eval_accuracy": 0.985,
      "eval_loss": 0.1263246089220047,
      "eval_runtime": 5.756,
      "eval_samples_per_second": 69.493,
      "eval_steps_per_second": 8.687,
      "step": 6573
    },
    {
      "epoch": 21.02,
      "learning_rate": 6.661933498994202e-05,
      "loss": 0.0,
      "step": 6580
    },
    {
      "epoch": 21.05,
      "learning_rate": 6.638267660631879e-05,
      "loss": 0.0015,
      "step": 6590
    },
    {
      "epoch": 21.09,
      "learning_rate": 6.614601822269554e-05,
      "loss": 0.0168,
      "step": 6600
    },
    {
      "epoch": 21.12,
      "learning_rate": 6.59093598390723e-05,
      "loss": 0.0,
      "step": 6610
    },
    {
      "epoch": 21.15,
      "learning_rate": 6.567270145544905e-05,
      "loss": 0.0004,
      "step": 6620
    },
    {
      "epoch": 21.18,
      "learning_rate": 6.543604307182582e-05,
      "loss": 0.0,
      "step": 6630
    },
    {
      "epoch": 21.21,
      "learning_rate": 6.519938468820258e-05,
      "loss": 0.0019,
      "step": 6640
    },
    {
      "epoch": 21.25,
      "learning_rate": 6.496272630457935e-05,
      "loss": 0.0436,
      "step": 6650
    },
    {
      "epoch": 21.28,
      "learning_rate": 6.472606792095611e-05,
      "loss": 0.0003,
      "step": 6660
    },
    {
      "epoch": 21.31,
      "learning_rate": 6.448940953733286e-05,
      "loss": 0.0002,
      "step": 6670
    },
    {
      "epoch": 21.34,
      "learning_rate": 6.425275115370962e-05,
      "loss": 0.0091,
      "step": 6680
    },
    {
      "epoch": 21.37,
      "learning_rate": 6.401609277008639e-05,
      "loss": 0.0003,
      "step": 6690
    },
    {
      "epoch": 21.41,
      "learning_rate": 6.377943438646315e-05,
      "loss": 0.0,
      "step": 6700
    },
    {
      "epoch": 21.44,
      "learning_rate": 6.35427760028399e-05,
      "loss": 0.0143,
      "step": 6710
    },
    {
      "epoch": 21.47,
      "learning_rate": 6.330611761921666e-05,
      "loss": 0.0002,
      "step": 6720
    },
    {
      "epoch": 21.5,
      "learning_rate": 6.306945923559342e-05,
      "loss": 0.0,
      "step": 6730
    },
    {
      "epoch": 21.53,
      "learning_rate": 6.283280085197019e-05,
      "loss": 0.0,
      "step": 6740
    },
    {
      "epoch": 21.57,
      "learning_rate": 6.259614246834694e-05,
      "loss": 0.0,
      "step": 6750
    },
    {
      "epoch": 21.6,
      "learning_rate": 6.235948408472371e-05,
      "loss": 0.0,
      "step": 6760
    },
    {
      "epoch": 21.63,
      "learning_rate": 6.212282570110046e-05,
      "loss": 0.0006,
      "step": 6770
    },
    {
      "epoch": 21.66,
      "learning_rate": 6.188616731747723e-05,
      "loss": 0.0,
      "step": 6780
    },
    {
      "epoch": 21.69,
      "learning_rate": 6.164950893385398e-05,
      "loss": 0.0002,
      "step": 6790
    },
    {
      "epoch": 21.73,
      "learning_rate": 6.141285055023075e-05,
      "loss": 0.0001,
      "step": 6800
    },
    {
      "epoch": 21.76,
      "learning_rate": 6.117619216660751e-05,
      "loss": 0.0,
      "step": 6810
    },
    {
      "epoch": 21.79,
      "learning_rate": 6.093953378298427e-05,
      "loss": 0.0,
      "step": 6820
    },
    {
      "epoch": 21.82,
      "learning_rate": 6.070287539936103e-05,
      "loss": 0.0038,
      "step": 6830
    },
    {
      "epoch": 21.85,
      "learning_rate": 6.046621701573778e-05,
      "loss": 0.072,
      "step": 6840
    },
    {
      "epoch": 21.88,
      "learning_rate": 6.022955863211455e-05,
      "loss": 0.0467,
      "step": 6850
    },
    {
      "epoch": 21.92,
      "learning_rate": 5.9992900248491304e-05,
      "loss": 0.0134,
      "step": 6860
    },
    {
      "epoch": 21.95,
      "learning_rate": 5.975624186486807e-05,
      "loss": 0.0026,
      "step": 6870
    },
    {
      "epoch": 21.98,
      "learning_rate": 5.9519583481244824e-05,
      "loss": 0.0239,
      "step": 6880
    },
    {
      "epoch": 22.0,
      "eval_accuracy": 0.9725,
      "eval_loss": 0.20680759847164154,
      "eval_runtime": 6.3851,
      "eval_samples_per_second": 62.646,
      "eval_steps_per_second": 7.831,
      "step": 6886
    },
    {
      "epoch": 22.01,
      "learning_rate": 5.928292509762159e-05,
      "loss": 0.0005,
      "step": 6890
    },
    {
      "epoch": 22.04,
      "learning_rate": 5.904626671399834e-05,
      "loss": 0.0217,
      "step": 6900
    },
    {
      "epoch": 22.08,
      "learning_rate": 5.880960833037511e-05,
      "loss": 0.0035,
      "step": 6910
    },
    {
      "epoch": 22.11,
      "learning_rate": 5.857294994675186e-05,
      "loss": 0.005,
      "step": 6920
    },
    {
      "epoch": 22.14,
      "learning_rate": 5.833629156312863e-05,
      "loss": 0.0001,
      "step": 6930
    },
    {
      "epoch": 22.17,
      "learning_rate": 5.809963317950539e-05,
      "loss": 0.0003,
      "step": 6940
    },
    {
      "epoch": 22.2,
      "learning_rate": 5.786297479588215e-05,
      "loss": 0.0007,
      "step": 6950
    },
    {
      "epoch": 22.24,
      "learning_rate": 5.762631641225891e-05,
      "loss": 0.0,
      "step": 6960
    },
    {
      "epoch": 22.27,
      "learning_rate": 5.7389658028635665e-05,
      "loss": 0.0001,
      "step": 6970
    },
    {
      "epoch": 22.3,
      "learning_rate": 5.715299964501243e-05,
      "loss": 0.0,
      "step": 6980
    },
    {
      "epoch": 22.33,
      "learning_rate": 5.6916341261389185e-05,
      "loss": 0.0052,
      "step": 6990
    },
    {
      "epoch": 22.36,
      "learning_rate": 5.667968287776595e-05,
      "loss": 0.0325,
      "step": 7000
    },
    {
      "epoch": 22.4,
      "learning_rate": 5.6443024494142705e-05,
      "loss": 0.0003,
      "step": 7010
    },
    {
      "epoch": 22.43,
      "learning_rate": 5.620636611051947e-05,
      "loss": 0.0037,
      "step": 7020
    },
    {
      "epoch": 22.46,
      "learning_rate": 5.5969707726896224e-05,
      "loss": 0.0002,
      "step": 7030
    },
    {
      "epoch": 22.49,
      "learning_rate": 5.5733049343272994e-05,
      "loss": 0.0245,
      "step": 7040
    },
    {
      "epoch": 22.52,
      "learning_rate": 5.5496390959649744e-05,
      "loss": 0.0013,
      "step": 7050
    },
    {
      "epoch": 22.56,
      "learning_rate": 5.5259732576026514e-05,
      "loss": 0.0097,
      "step": 7060
    },
    {
      "epoch": 22.59,
      "learning_rate": 5.502307419240327e-05,
      "loss": 0.0006,
      "step": 7070
    },
    {
      "epoch": 22.62,
      "learning_rate": 5.4786415808780034e-05,
      "loss": 0.0,
      "step": 7080
    },
    {
      "epoch": 22.65,
      "learning_rate": 5.454975742515679e-05,
      "loss": 0.0004,
      "step": 7090
    },
    {
      "epoch": 22.68,
      "learning_rate": 5.4313099041533546e-05,
      "loss": 0.0007,
      "step": 7100
    },
    {
      "epoch": 22.72,
      "learning_rate": 5.407644065791031e-05,
      "loss": 0.0044,
      "step": 7110
    },
    {
      "epoch": 22.75,
      "learning_rate": 5.3839782274287066e-05,
      "loss": 0.0003,
      "step": 7120
    },
    {
      "epoch": 22.78,
      "learning_rate": 5.360312389066383e-05,
      "loss": 0.0,
      "step": 7130
    },
    {
      "epoch": 22.81,
      "learning_rate": 5.3366465507040586e-05,
      "loss": 0.0026,
      "step": 7140
    },
    {
      "epoch": 22.84,
      "learning_rate": 5.312980712341735e-05,
      "loss": 0.0,
      "step": 7150
    },
    {
      "epoch": 22.88,
      "learning_rate": 5.2893148739794105e-05,
      "loss": 0.0019,
      "step": 7160
    },
    {
      "epoch": 22.91,
      "learning_rate": 5.2656490356170875e-05,
      "loss": 0.0,
      "step": 7170
    },
    {
      "epoch": 22.94,
      "learning_rate": 5.241983197254763e-05,
      "loss": 0.0001,
      "step": 7180
    },
    {
      "epoch": 22.97,
      "learning_rate": 5.2183173588924395e-05,
      "loss": 0.0,
      "step": 7190
    },
    {
      "epoch": 23.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.10589393973350525,
      "eval_runtime": 5.2362,
      "eval_samples_per_second": 76.391,
      "eval_steps_per_second": 9.549,
      "step": 7199
    },
    {
      "epoch": 23.0,
      "learning_rate": 5.194651520530115e-05,
      "loss": 0.0,
      "step": 7200
    },
    {
      "epoch": 23.04,
      "learning_rate": 5.1709856821677915e-05,
      "loss": 0.0003,
      "step": 7210
    },
    {
      "epoch": 23.07,
      "learning_rate": 5.147319843805467e-05,
      "loss": 0.0003,
      "step": 7220
    },
    {
      "epoch": 23.1,
      "learning_rate": 5.123654005443143e-05,
      "loss": 0.0,
      "step": 7230
    },
    {
      "epoch": 23.13,
      "learning_rate": 5.099988167080819e-05,
      "loss": 0.0,
      "step": 7240
    },
    {
      "epoch": 23.16,
      "learning_rate": 5.076322328718495e-05,
      "loss": 0.0016,
      "step": 7250
    },
    {
      "epoch": 23.19,
      "learning_rate": 5.052656490356171e-05,
      "loss": 0.0003,
      "step": 7260
    },
    {
      "epoch": 23.23,
      "learning_rate": 5.028990651993847e-05,
      "loss": 0.0002,
      "step": 7270
    },
    {
      "epoch": 23.26,
      "learning_rate": 5.0053248136315237e-05,
      "loss": 0.0058,
      "step": 7280
    },
    {
      "epoch": 23.29,
      "learning_rate": 4.9816589752691986e-05,
      "loss": 0.0,
      "step": 7290
    },
    {
      "epoch": 23.32,
      "learning_rate": 4.957993136906875e-05,
      "loss": 0.0001,
      "step": 7300
    },
    {
      "epoch": 23.35,
      "learning_rate": 4.934327298544551e-05,
      "loss": 0.0085,
      "step": 7310
    },
    {
      "epoch": 23.39,
      "learning_rate": 4.910661460182227e-05,
      "loss": 0.0073,
      "step": 7320
    },
    {
      "epoch": 23.42,
      "learning_rate": 4.886995621819903e-05,
      "loss": 0.0076,
      "step": 7330
    },
    {
      "epoch": 23.45,
      "learning_rate": 4.863329783457579e-05,
      "loss": 0.0,
      "step": 7340
    },
    {
      "epoch": 23.48,
      "learning_rate": 4.839663945095255e-05,
      "loss": 0.0,
      "step": 7350
    },
    {
      "epoch": 23.51,
      "learning_rate": 4.8159981067329315e-05,
      "loss": 0.0,
      "step": 7360
    },
    {
      "epoch": 23.55,
      "learning_rate": 4.792332268370607e-05,
      "loss": 0.0002,
      "step": 7370
    },
    {
      "epoch": 23.58,
      "learning_rate": 4.7686664300082835e-05,
      "loss": 0.0,
      "step": 7380
    },
    {
      "epoch": 23.61,
      "learning_rate": 4.745000591645959e-05,
      "loss": 0.0,
      "step": 7390
    },
    {
      "epoch": 23.64,
      "learning_rate": 4.7213347532836354e-05,
      "loss": 0.0,
      "step": 7400
    },
    {
      "epoch": 23.67,
      "learning_rate": 4.697668914921312e-05,
      "loss": 0.0001,
      "step": 7410
    },
    {
      "epoch": 23.71,
      "learning_rate": 4.674003076558987e-05,
      "loss": 0.0027,
      "step": 7420
    },
    {
      "epoch": 23.74,
      "learning_rate": 4.650337238196663e-05,
      "loss": 0.0,
      "step": 7430
    },
    {
      "epoch": 23.77,
      "learning_rate": 4.6266713998343394e-05,
      "loss": 0.0,
      "step": 7440
    },
    {
      "epoch": 23.8,
      "learning_rate": 4.603005561472015e-05,
      "loss": 0.0,
      "step": 7450
    },
    {
      "epoch": 23.83,
      "learning_rate": 4.579339723109691e-05,
      "loss": 0.0003,
      "step": 7460
    },
    {
      "epoch": 23.87,
      "learning_rate": 4.5556738847473676e-05,
      "loss": 0.0,
      "step": 7470
    },
    {
      "epoch": 23.9,
      "learning_rate": 4.532008046385043e-05,
      "loss": 0.0,
      "step": 7480
    },
    {
      "epoch": 23.93,
      "learning_rate": 4.5083422080227196e-05,
      "loss": 0.0,
      "step": 7490
    },
    {
      "epoch": 23.96,
      "learning_rate": 4.484676369660395e-05,
      "loss": 0.0,
      "step": 7500
    },
    {
      "epoch": 23.99,
      "learning_rate": 4.4610105312980716e-05,
      "loss": 0.0,
      "step": 7510
    },
    {
      "epoch": 24.0,
      "eval_accuracy": 0.9875,
      "eval_loss": 0.14845281839370728,
      "eval_runtime": 6.865,
      "eval_samples_per_second": 58.267,
      "eval_steps_per_second": 7.283,
      "step": 7512
    },
    {
      "epoch": 24.03,
      "learning_rate": 4.437344692935748e-05,
      "loss": 0.0,
      "step": 7520
    },
    {
      "epoch": 24.06,
      "learning_rate": 4.4136788545734235e-05,
      "loss": 0.0001,
      "step": 7530
    },
    {
      "epoch": 24.09,
      "learning_rate": 4.3900130162111e-05,
      "loss": 0.0008,
      "step": 7540
    },
    {
      "epoch": 24.12,
      "learning_rate": 4.3663471778487755e-05,
      "loss": 0.0,
      "step": 7550
    },
    {
      "epoch": 24.15,
      "learning_rate": 4.342681339486451e-05,
      "loss": 0.0,
      "step": 7560
    },
    {
      "epoch": 24.19,
      "learning_rate": 4.3190155011241275e-05,
      "loss": 0.0,
      "step": 7570
    },
    {
      "epoch": 24.22,
      "learning_rate": 4.295349662761803e-05,
      "loss": 0.0,
      "step": 7580
    },
    {
      "epoch": 24.25,
      "learning_rate": 4.2716838243994794e-05,
      "loss": 0.0,
      "step": 7590
    },
    {
      "epoch": 24.28,
      "learning_rate": 4.248017986037156e-05,
      "loss": 0.0001,
      "step": 7600
    },
    {
      "epoch": 24.31,
      "learning_rate": 4.2243521476748314e-05,
      "loss": 0.0059,
      "step": 7610
    },
    {
      "epoch": 24.35,
      "learning_rate": 4.200686309312508e-05,
      "loss": 0.0,
      "step": 7620
    },
    {
      "epoch": 24.38,
      "learning_rate": 4.1770204709501833e-05,
      "loss": 0.0038,
      "step": 7630
    },
    {
      "epoch": 24.41,
      "learning_rate": 4.15335463258786e-05,
      "loss": 0.0,
      "step": 7640
    },
    {
      "epoch": 24.44,
      "learning_rate": 4.129688794225536e-05,
      "loss": 0.0,
      "step": 7650
    },
    {
      "epoch": 24.47,
      "learning_rate": 4.1060229558632116e-05,
      "loss": 0.0,
      "step": 7660
    },
    {
      "epoch": 24.5,
      "learning_rate": 4.082357117500888e-05,
      "loss": 0.0,
      "step": 7670
    },
    {
      "epoch": 24.54,
      "learning_rate": 4.058691279138564e-05,
      "loss": 0.0002,
      "step": 7680
    },
    {
      "epoch": 24.57,
      "learning_rate": 4.035025440776239e-05,
      "loss": 0.0,
      "step": 7690
    },
    {
      "epoch": 24.6,
      "learning_rate": 4.0113596024139156e-05,
      "loss": 0.0,
      "step": 7700
    },
    {
      "epoch": 24.63,
      "learning_rate": 3.987693764051592e-05,
      "loss": 0.0,
      "step": 7710
    },
    {
      "epoch": 24.66,
      "learning_rate": 3.9640279256892675e-05,
      "loss": 0.0,
      "step": 7720
    },
    {
      "epoch": 24.7,
      "learning_rate": 3.940362087326944e-05,
      "loss": 0.0001,
      "step": 7730
    },
    {
      "epoch": 24.73,
      "learning_rate": 3.9166962489646195e-05,
      "loss": 0.0005,
      "step": 7740
    },
    {
      "epoch": 24.76,
      "learning_rate": 3.893030410602296e-05,
      "loss": 0.0,
      "step": 7750
    },
    {
      "epoch": 24.79,
      "learning_rate": 3.869364572239972e-05,
      "loss": 0.0001,
      "step": 7760
    },
    {
      "epoch": 24.82,
      "learning_rate": 3.845698733877648e-05,
      "loss": 0.0,
      "step": 7770
    },
    {
      "epoch": 24.86,
      "learning_rate": 3.822032895515324e-05,
      "loss": 0.0,
      "step": 7780
    },
    {
      "epoch": 24.89,
      "learning_rate": 3.798367057153e-05,
      "loss": 0.0,
      "step": 7790
    },
    {
      "epoch": 24.92,
      "learning_rate": 3.774701218790676e-05,
      "loss": 0.0,
      "step": 7800
    },
    {
      "epoch": 24.95,
      "learning_rate": 3.7510353804283524e-05,
      "loss": 0.0245,
      "step": 7810
    },
    {
      "epoch": 24.98,
      "learning_rate": 3.727369542066027e-05,
      "loss": 0.0,
      "step": 7820
    },
    {
      "epoch": 25.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.15507297217845917,
      "eval_runtime": 5.1221,
      "eval_samples_per_second": 78.093,
      "eval_steps_per_second": 9.762,
      "step": 7825
    },
    {
      "epoch": 25.02,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.0,
      "step": 7830
    },
    {
      "epoch": 25.05,
      "learning_rate": 3.68003786534138e-05,
      "loss": 0.0,
      "step": 7840
    },
    {
      "epoch": 25.08,
      "learning_rate": 3.6563720269790556e-05,
      "loss": 0.0,
      "step": 7850
    },
    {
      "epoch": 25.11,
      "learning_rate": 3.632706188616732e-05,
      "loss": 0.0,
      "step": 7860
    },
    {
      "epoch": 25.14,
      "learning_rate": 3.6090403502544076e-05,
      "loss": 0.0,
      "step": 7870
    },
    {
      "epoch": 25.18,
      "learning_rate": 3.585374511892084e-05,
      "loss": 0.0,
      "step": 7880
    },
    {
      "epoch": 25.21,
      "learning_rate": 3.56170867352976e-05,
      "loss": 0.0,
      "step": 7890
    },
    {
      "epoch": 25.24,
      "learning_rate": 3.538042835167436e-05,
      "loss": 0.0,
      "step": 7900
    },
    {
      "epoch": 25.27,
      "learning_rate": 3.514376996805112e-05,
      "loss": 0.0,
      "step": 7910
    },
    {
      "epoch": 25.3,
      "learning_rate": 3.4907111584427885e-05,
      "loss": 0.0,
      "step": 7920
    },
    {
      "epoch": 25.34,
      "learning_rate": 3.467045320080464e-05,
      "loss": 0.0,
      "step": 7930
    },
    {
      "epoch": 25.37,
      "learning_rate": 3.4433794817181405e-05,
      "loss": 0.0,
      "step": 7940
    },
    {
      "epoch": 25.4,
      "learning_rate": 3.419713643355816e-05,
      "loss": 0.0033,
      "step": 7950
    },
    {
      "epoch": 25.43,
      "learning_rate": 3.396047804993492e-05,
      "loss": 0.0,
      "step": 7960
    },
    {
      "epoch": 25.46,
      "learning_rate": 3.372381966631168e-05,
      "loss": 0.0,
      "step": 7970
    },
    {
      "epoch": 25.5,
      "learning_rate": 3.348716128268844e-05,
      "loss": 0.0,
      "step": 7980
    },
    {
      "epoch": 25.53,
      "learning_rate": 3.32505028990652e-05,
      "loss": 0.0,
      "step": 7990
    },
    {
      "epoch": 25.56,
      "learning_rate": 3.3013844515441964e-05,
      "loss": 0.0071,
      "step": 8000
    },
    {
      "epoch": 25.59,
      "learning_rate": 3.277718613181872e-05,
      "loss": 0.0,
      "step": 8010
    },
    {
      "epoch": 25.62,
      "learning_rate": 3.254052774819548e-05,
      "loss": 0.0001,
      "step": 8020
    },
    {
      "epoch": 25.65,
      "learning_rate": 3.230386936457224e-05,
      "loss": 0.0,
      "step": 8030
    },
    {
      "epoch": 25.69,
      "learning_rate": 3.2067210980949e-05,
      "loss": 0.0,
      "step": 8040
    },
    {
      "epoch": 25.72,
      "learning_rate": 3.1830552597325766e-05,
      "loss": 0.0,
      "step": 8050
    },
    {
      "epoch": 25.75,
      "learning_rate": 3.159389421370252e-05,
      "loss": 0.0,
      "step": 8060
    },
    {
      "epoch": 25.78,
      "learning_rate": 3.1357235830079286e-05,
      "loss": 0.0,
      "step": 8070
    },
    {
      "epoch": 25.81,
      "learning_rate": 3.112057744645604e-05,
      "loss": 0.0,
      "step": 8080
    },
    {
      "epoch": 25.85,
      "learning_rate": 3.08839190628328e-05,
      "loss": 0.0001,
      "step": 8090
    },
    {
      "epoch": 25.88,
      "learning_rate": 3.064726067920956e-05,
      "loss": 0.0002,
      "step": 8100
    },
    {
      "epoch": 25.91,
      "learning_rate": 3.041060229558632e-05,
      "loss": 0.0,
      "step": 8110
    },
    {
      "epoch": 25.94,
      "learning_rate": 3.017394391196308e-05,
      "loss": 0.0,
      "step": 8120
    },
    {
      "epoch": 25.97,
      "learning_rate": 2.993728552833984e-05,
      "loss": 0.0,
      "step": 8130
    },
    {
      "epoch": 26.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.14341458678245544,
      "eval_runtime": 6.8347,
      "eval_samples_per_second": 58.525,
      "eval_steps_per_second": 7.316,
      "step": 8138
    },
    {
      "epoch": 26.01,
      "learning_rate": 2.9700627144716604e-05,
      "loss": 0.0,
      "step": 8140
    },
    {
      "epoch": 26.04,
      "learning_rate": 2.9463968761093364e-05,
      "loss": 0.0,
      "step": 8150
    },
    {
      "epoch": 26.07,
      "learning_rate": 2.9227310377470124e-05,
      "loss": 0.0,
      "step": 8160
    },
    {
      "epoch": 26.1,
      "learning_rate": 2.8990651993846884e-05,
      "loss": 0.0,
      "step": 8170
    },
    {
      "epoch": 26.13,
      "learning_rate": 2.8753993610223644e-05,
      "loss": 0.0,
      "step": 8180
    },
    {
      "epoch": 26.17,
      "learning_rate": 2.8517335226600407e-05,
      "loss": 0.0,
      "step": 8190
    },
    {
      "epoch": 26.2,
      "learning_rate": 2.8280676842977167e-05,
      "loss": 0.0,
      "step": 8200
    },
    {
      "epoch": 26.23,
      "learning_rate": 2.804401845935392e-05,
      "loss": 0.0,
      "step": 8210
    },
    {
      "epoch": 26.26,
      "learning_rate": 2.7807360075730683e-05,
      "loss": 0.0,
      "step": 8220
    },
    {
      "epoch": 26.29,
      "learning_rate": 2.7570701692107443e-05,
      "loss": 0.0,
      "step": 8230
    },
    {
      "epoch": 26.33,
      "learning_rate": 2.7334043308484202e-05,
      "loss": 0.0,
      "step": 8240
    },
    {
      "epoch": 26.36,
      "learning_rate": 2.7097384924860962e-05,
      "loss": 0.0,
      "step": 8250
    },
    {
      "epoch": 26.39,
      "learning_rate": 2.6860726541237726e-05,
      "loss": 0.0,
      "step": 8260
    },
    {
      "epoch": 26.42,
      "learning_rate": 2.6624068157614485e-05,
      "loss": 0.0001,
      "step": 8270
    },
    {
      "epoch": 26.45,
      "learning_rate": 2.6387409773991245e-05,
      "loss": 0.0,
      "step": 8280
    },
    {
      "epoch": 26.49,
      "learning_rate": 2.6150751390368005e-05,
      "loss": 0.0,
      "step": 8290
    },
    {
      "epoch": 26.52,
      "learning_rate": 2.5914093006744765e-05,
      "loss": 0.0,
      "step": 8300
    },
    {
      "epoch": 26.55,
      "learning_rate": 2.5677434623121528e-05,
      "loss": 0.0,
      "step": 8310
    },
    {
      "epoch": 26.58,
      "learning_rate": 2.5440776239498288e-05,
      "loss": 0.0,
      "step": 8320
    },
    {
      "epoch": 26.61,
      "learning_rate": 2.5204117855875048e-05,
      "loss": 0.0001,
      "step": 8330
    },
    {
      "epoch": 26.65,
      "learning_rate": 2.4967459472251804e-05,
      "loss": 0.0,
      "step": 8340
    },
    {
      "epoch": 26.68,
      "learning_rate": 2.4730801088628567e-05,
      "loss": 0.0,
      "step": 8350
    },
    {
      "epoch": 26.71,
      "learning_rate": 2.4494142705005327e-05,
      "loss": 0.0,
      "step": 8360
    },
    {
      "epoch": 26.74,
      "learning_rate": 2.4257484321382087e-05,
      "loss": 0.0,
      "step": 8370
    },
    {
      "epoch": 26.77,
      "learning_rate": 2.4020825937758847e-05,
      "loss": 0.0,
      "step": 8380
    },
    {
      "epoch": 26.81,
      "learning_rate": 2.3784167554135606e-05,
      "loss": 0.0,
      "step": 8390
    },
    {
      "epoch": 26.84,
      "learning_rate": 2.3547509170512366e-05,
      "loss": 0.0,
      "step": 8400
    },
    {
      "epoch": 26.87,
      "learning_rate": 2.3310850786889126e-05,
      "loss": 0.0,
      "step": 8410
    },
    {
      "epoch": 26.9,
      "learning_rate": 2.3074192403265886e-05,
      "loss": 0.0,
      "step": 8420
    },
    {
      "epoch": 26.93,
      "learning_rate": 2.283753401964265e-05,
      "loss": 0.0,
      "step": 8430
    },
    {
      "epoch": 26.96,
      "learning_rate": 2.2600875636019406e-05,
      "loss": 0.0015,
      "step": 8440
    },
    {
      "epoch": 27.0,
      "learning_rate": 2.2364217252396165e-05,
      "loss": 0.0001,
      "step": 8450
    },
    {
      "epoch": 27.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.1398298293352127,
      "eval_runtime": 5.1804,
      "eval_samples_per_second": 77.213,
      "eval_steps_per_second": 9.652,
      "step": 8451
    },
    {
      "epoch": 27.03,
      "learning_rate": 2.2127558868772925e-05,
      "loss": 0.0,
      "step": 8460
    },
    {
      "epoch": 27.06,
      "learning_rate": 2.189090048514969e-05,
      "loss": 0.0,
      "step": 8470
    },
    {
      "epoch": 27.09,
      "learning_rate": 2.1654242101526448e-05,
      "loss": 0.0,
      "step": 8480
    },
    {
      "epoch": 27.12,
      "learning_rate": 2.1417583717903208e-05,
      "loss": 0.0007,
      "step": 8490
    },
    {
      "epoch": 27.16,
      "learning_rate": 2.1180925334279968e-05,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 27.19,
      "learning_rate": 2.0944266950656728e-05,
      "loss": 0.0,
      "step": 8510
    },
    {
      "epoch": 27.22,
      "learning_rate": 2.0707608567033487e-05,
      "loss": 0.0,
      "step": 8520
    },
    {
      "epoch": 27.25,
      "learning_rate": 2.0470950183410247e-05,
      "loss": 0.0,
      "step": 8530
    },
    {
      "epoch": 27.28,
      "learning_rate": 2.0234291799787007e-05,
      "loss": 0.0,
      "step": 8540
    },
    {
      "epoch": 27.32,
      "learning_rate": 1.999763341616377e-05,
      "loss": 0.0,
      "step": 8550
    },
    {
      "epoch": 27.35,
      "learning_rate": 1.976097503254053e-05,
      "loss": 0.0,
      "step": 8560
    },
    {
      "epoch": 27.38,
      "learning_rate": 1.952431664891729e-05,
      "loss": 0.0,
      "step": 8570
    },
    {
      "epoch": 27.41,
      "learning_rate": 1.9287658265294046e-05,
      "loss": 0.0019,
      "step": 8580
    },
    {
      "epoch": 27.44,
      "learning_rate": 1.905099988167081e-05,
      "loss": 0.0,
      "step": 8590
    },
    {
      "epoch": 27.48,
      "learning_rate": 1.881434149804757e-05,
      "loss": 0.0,
      "step": 8600
    },
    {
      "epoch": 27.51,
      "learning_rate": 1.857768311442433e-05,
      "loss": 0.0002,
      "step": 8610
    },
    {
      "epoch": 27.54,
      "learning_rate": 1.834102473080109e-05,
      "loss": 0.0,
      "step": 8620
    },
    {
      "epoch": 27.57,
      "learning_rate": 1.8104366347177852e-05,
      "loss": 0.0,
      "step": 8630
    },
    {
      "epoch": 27.6,
      "learning_rate": 1.786770796355461e-05,
      "loss": 0.0001,
      "step": 8640
    },
    {
      "epoch": 27.64,
      "learning_rate": 1.763104957993137e-05,
      "loss": 0.0031,
      "step": 8650
    },
    {
      "epoch": 27.67,
      "learning_rate": 1.7394391196308128e-05,
      "loss": 0.0,
      "step": 8660
    },
    {
      "epoch": 27.7,
      "learning_rate": 1.715773281268489e-05,
      "loss": 0.0,
      "step": 8670
    },
    {
      "epoch": 27.73,
      "learning_rate": 1.692107442906165e-05,
      "loss": 0.0,
      "step": 8680
    },
    {
      "epoch": 27.76,
      "learning_rate": 1.668441604543841e-05,
      "loss": 0.0,
      "step": 8690
    },
    {
      "epoch": 27.8,
      "learning_rate": 1.644775766181517e-05,
      "loss": 0.0,
      "step": 8700
    },
    {
      "epoch": 27.83,
      "learning_rate": 1.621109927819193e-05,
      "loss": 0.0,
      "step": 8710
    },
    {
      "epoch": 27.86,
      "learning_rate": 1.597444089456869e-05,
      "loss": 0.0,
      "step": 8720
    },
    {
      "epoch": 27.89,
      "learning_rate": 1.573778251094545e-05,
      "loss": 0.0,
      "step": 8730
    },
    {
      "epoch": 27.92,
      "learning_rate": 1.550112412732221e-05,
      "loss": 0.0,
      "step": 8740
    },
    {
      "epoch": 27.96,
      "learning_rate": 1.5264465743698973e-05,
      "loss": 0.0,
      "step": 8750
    },
    {
      "epoch": 27.99,
      "learning_rate": 1.5027807360075733e-05,
      "loss": 0.0,
      "step": 8760
    },
    {
      "epoch": 28.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.13574786484241486,
      "eval_runtime": 6.1395,
      "eval_samples_per_second": 65.152,
      "eval_steps_per_second": 8.144,
      "step": 8764
    },
    {
      "epoch": 28.02,
      "learning_rate": 1.479114897645249e-05,
      "loss": 0.0,
      "step": 8770
    },
    {
      "epoch": 28.05,
      "learning_rate": 1.4554490592829251e-05,
      "loss": 0.0,
      "step": 8780
    },
    {
      "epoch": 28.08,
      "learning_rate": 1.4317832209206011e-05,
      "loss": 0.0,
      "step": 8790
    },
    {
      "epoch": 28.12,
      "learning_rate": 1.4081173825582772e-05,
      "loss": 0.0,
      "step": 8800
    },
    {
      "epoch": 28.15,
      "learning_rate": 1.3844515441959532e-05,
      "loss": 0.0,
      "step": 8810
    },
    {
      "epoch": 28.18,
      "learning_rate": 1.3607857058336294e-05,
      "loss": 0.0,
      "step": 8820
    },
    {
      "epoch": 28.21,
      "learning_rate": 1.3371198674713054e-05,
      "loss": 0.0,
      "step": 8830
    },
    {
      "epoch": 28.24,
      "learning_rate": 1.3134540291089812e-05,
      "loss": 0.0033,
      "step": 8840
    },
    {
      "epoch": 28.27,
      "learning_rate": 1.2897881907466571e-05,
      "loss": 0.0,
      "step": 8850
    },
    {
      "epoch": 28.31,
      "learning_rate": 1.2661223523843333e-05,
      "loss": 0.0,
      "step": 8860
    },
    {
      "epoch": 28.34,
      "learning_rate": 1.2424565140220093e-05,
      "loss": 0.0,
      "step": 8870
    },
    {
      "epoch": 28.37,
      "learning_rate": 1.2187906756596854e-05,
      "loss": 0.0,
      "step": 8880
    },
    {
      "epoch": 28.4,
      "learning_rate": 1.1951248372973612e-05,
      "loss": 0.0,
      "step": 8890
    },
    {
      "epoch": 28.43,
      "learning_rate": 1.1714589989350374e-05,
      "loss": 0.0,
      "step": 8900
    },
    {
      "epoch": 28.47,
      "learning_rate": 1.1477931605727134e-05,
      "loss": 0.0,
      "step": 8910
    },
    {
      "epoch": 28.5,
      "learning_rate": 1.1241273222103894e-05,
      "loss": 0.0,
      "step": 8920
    },
    {
      "epoch": 28.53,
      "learning_rate": 1.1004614838480653e-05,
      "loss": 0.0,
      "step": 8930
    },
    {
      "epoch": 28.56,
      "learning_rate": 1.0767956454857415e-05,
      "loss": 0.0001,
      "step": 8940
    },
    {
      "epoch": 28.59,
      "learning_rate": 1.0531298071234175e-05,
      "loss": 0.0,
      "step": 8950
    },
    {
      "epoch": 28.63,
      "learning_rate": 1.0294639687610935e-05,
      "loss": 0.0,
      "step": 8960
    },
    {
      "epoch": 28.66,
      "learning_rate": 1.0057981303987694e-05,
      "loss": 0.0,
      "step": 8970
    },
    {
      "epoch": 28.69,
      "learning_rate": 9.821322920364454e-06,
      "loss": 0.0,
      "step": 8980
    },
    {
      "epoch": 28.72,
      "learning_rate": 9.584664536741214e-06,
      "loss": 0.0,
      "step": 8990
    },
    {
      "epoch": 28.75,
      "learning_rate": 9.348006153117975e-06,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 28.79,
      "learning_rate": 9.111347769494735e-06,
      "loss": 0.0,
      "step": 9010
    },
    {
      "epoch": 28.82,
      "learning_rate": 8.874689385871495e-06,
      "loss": 0.0,
      "step": 9020
    },
    {
      "epoch": 28.85,
      "learning_rate": 8.638031002248255e-06,
      "loss": 0.0,
      "step": 9030
    },
    {
      "epoch": 28.88,
      "learning_rate": 8.401372618625015e-06,
      "loss": 0.0,
      "step": 9040
    },
    {
      "epoch": 28.91,
      "learning_rate": 8.164714235001775e-06,
      "loss": 0.0,
      "step": 9050
    },
    {
      "epoch": 28.95,
      "learning_rate": 7.928055851378536e-06,
      "loss": 0.0,
      "step": 9060
    },
    {
      "epoch": 28.98,
      "learning_rate": 7.691397467755296e-06,
      "loss": 0.0,
      "step": 9070
    },
    {
      "epoch": 29.0,
      "eval_accuracy": 0.99,
      "eval_loss": 0.13164377212524414,
      "eval_runtime": 5.2001,
      "eval_samples_per_second": 76.922,
      "eval_steps_per_second": 9.615,
      "step": 9077
    },
    {
      "epoch": 29.01,
      "learning_rate": 7.4547390841320565e-06,
      "loss": 0.0,
      "step": 9080
    },
    {
      "epoch": 29.04,
      "learning_rate": 7.2180807005088155e-06,
      "loss": 0.0,
      "step": 9090
    },
    {
      "epoch": 29.07,
      "learning_rate": 6.981422316885576e-06,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 29.11,
      "learning_rate": 6.744763933262337e-06,
      "loss": 0.0,
      "step": 9110
    },
    {
      "epoch": 29.14,
      "learning_rate": 6.508105549639096e-06,
      "loss": 0.0,
      "step": 9120
    },
    {
      "epoch": 29.17,
      "learning_rate": 6.2714471660158565e-06,
      "loss": 0.0016,
      "step": 9130
    },
    {
      "epoch": 29.2,
      "learning_rate": 6.034788782392616e-06,
      "loss": 0.0,
      "step": 9140
    },
    {
      "epoch": 29.23,
      "learning_rate": 5.798130398769377e-06,
      "loss": 0.0,
      "step": 9150
    },
    {
      "epoch": 29.27,
      "learning_rate": 5.561472015146137e-06,
      "loss": 0.0,
      "step": 9160
    },
    {
      "epoch": 29.3,
      "learning_rate": 5.324813631522897e-06,
      "loss": 0.0,
      "step": 9170
    },
    {
      "epoch": 29.33,
      "learning_rate": 5.088155247899657e-06,
      "loss": 0.0,
      "step": 9180
    },
    {
      "epoch": 29.36,
      "learning_rate": 4.851496864276417e-06,
      "loss": 0.0,
      "step": 9190
    },
    {
      "epoch": 29.39,
      "learning_rate": 4.614838480653178e-06,
      "loss": 0.0,
      "step": 9200
    },
    {
      "epoch": 29.42,
      "learning_rate": 4.3781800970299375e-06,
      "loss": 0.0,
      "step": 9210
    },
    {
      "epoch": 29.46,
      "learning_rate": 4.141521713406698e-06,
      "loss": 0.0,
      "step": 9220
    },
    {
      "epoch": 29.49,
      "learning_rate": 3.904863329783458e-06,
      "loss": 0.0,
      "step": 9230
    },
    {
      "epoch": 29.52,
      "learning_rate": 3.668204946160218e-06,
      "loss": 0.0,
      "step": 9240
    },
    {
      "epoch": 29.55,
      "learning_rate": 3.431546562536978e-06,
      "loss": 0.0,
      "step": 9250
    },
    {
      "epoch": 29.58,
      "learning_rate": 3.194888178913738e-06,
      "loss": 0.0,
      "step": 9260
    },
    {
      "epoch": 29.62,
      "learning_rate": 2.958229795290498e-06,
      "loss": 0.0,
      "step": 9270
    },
    {
      "epoch": 29.65,
      "learning_rate": 2.7215714116672583e-06,
      "loss": 0.0,
      "step": 9280
    },
    {
      "epoch": 29.68,
      "learning_rate": 2.4849130280440186e-06,
      "loss": 0.0,
      "step": 9290
    },
    {
      "epoch": 29.71,
      "learning_rate": 2.248254644420779e-06,
      "loss": 0.0,
      "step": 9300
    },
    {
      "epoch": 29.74,
      "learning_rate": 2.0115962607975386e-06,
      "loss": 0.0,
      "step": 9310
    },
    {
      "epoch": 29.78,
      "learning_rate": 1.7749378771742989e-06,
      "loss": 0.0,
      "step": 9320
    },
    {
      "epoch": 29.81,
      "learning_rate": 1.538279493551059e-06,
      "loss": 0.0,
      "step": 9330
    },
    {
      "epoch": 29.84,
      "learning_rate": 1.3016211099278191e-06,
      "loss": 0.0,
      "step": 9340
    },
    {
      "epoch": 29.87,
      "learning_rate": 1.0649627263045794e-06,
      "loss": 0.0,
      "step": 9350
    },
    {
      "epoch": 29.9,
      "learning_rate": 8.283043426813395e-07,
      "loss": 0.0,
      "step": 9360
    },
    {
      "epoch": 29.94,
      "learning_rate": 5.916459590580996e-07,
      "loss": 0.0,
      "step": 9370
    },
    {
      "epoch": 29.97,
      "learning_rate": 3.549875754348598e-07,
      "loss": 0.0,
      "step": 9380
    },
    {
      "epoch": 30.0,
      "learning_rate": 1.1832919181161993e-07,
      "loss": 0.0,
      "step": 9390
    },
    {
      "epoch": 30.0,
      "eval_accuracy": 0.9925,
      "eval_loss": 0.12477295100688934,
      "eval_runtime": 5.3362,
      "eval_samples_per_second": 74.96,
      "eval_steps_per_second": 9.37,
      "step": 9390
    },
    {
      "epoch": 30.0,
      "step": 9390,
      "total_flos": 2.3515974247797965e+19,
      "train_loss": 0.07091858377127398,
      "train_runtime": 7238.3055,
      "train_samples_per_second": 41.467,
      "train_steps_per_second": 1.297
    }
  ],
  "logging_steps": 10,
  "max_steps": 9390,
  "num_train_epochs": 30,
  "save_steps": 500,
  "total_flos": 2.3515974247797965e+19,
  "trial_name": null,
  "trial_params": null
}