{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 1000,
  "global_step": 43943,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00022756753066472475,
      "grad_norm": 812.0,
      "learning_rate": 2.0477815699658704e-07,
      "loss": 11.2718,
      "step": 10
    },
    {
      "epoch": 0.0004551350613294495,
      "grad_norm": 884.0,
      "learning_rate": 4.3230944254835047e-07,
      "loss": 10.3364,
      "step": 20
    },
    {
      "epoch": 0.0006827025919941743,
      "grad_norm": 752.0,
      "learning_rate": 6.598407281001138e-07,
      "loss": 7.8486,
      "step": 30
    },
    {
      "epoch": 0.000910270122658899,
      "grad_norm": 540.0,
      "learning_rate": 8.873720136518772e-07,
      "loss": 7.4714,
      "step": 40
    },
    {
      "epoch": 0.0011378376533236238,
      "grad_norm": 544.0,
      "learning_rate": 1.1149032992036406e-06,
      "loss": 8.5469,
      "step": 50
    },
    {
      "epoch": 0.0013654051839883485,
      "grad_norm": 1192.0,
      "learning_rate": 1.342434584755404e-06,
      "loss": 10.141,
      "step": 60
    },
    {
      "epoch": 0.0015929727146530733,
      "grad_norm": 916.0,
      "learning_rate": 1.5699658703071675e-06,
      "loss": 6.4641,
      "step": 70
    },
    {
      "epoch": 0.001820540245317798,
      "grad_norm": 482.0,
      "learning_rate": 1.7974971558589307e-06,
      "loss": 5.5834,
      "step": 80
    },
    {
      "epoch": 0.002048107775982523,
      "grad_norm": 1480.0,
      "learning_rate": 2.025028441410694e-06,
      "loss": 5.2911,
      "step": 90
    },
    {
      "epoch": 0.0022756753066472475,
      "grad_norm": 660.0,
      "learning_rate": 2.2525597269624573e-06,
      "loss": 3.8162,
      "step": 100
    },
    {
      "epoch": 0.0025032428373119725,
      "grad_norm": 312.0,
      "learning_rate": 2.4800910125142208e-06,
      "loss": 3.4865,
      "step": 110
    },
    {
      "epoch": 0.002730810367976697,
      "grad_norm": 332.0,
      "learning_rate": 2.7076222980659846e-06,
      "loss": 1.789,
      "step": 120
    },
    {
      "epoch": 0.002958377898641422,
      "grad_norm": 772.0,
      "learning_rate": 2.9351535836177476e-06,
      "loss": 2.034,
      "step": 130
    },
    {
      "epoch": 0.0031859454293061465,
      "grad_norm": 251.0,
      "learning_rate": 3.162684869169511e-06,
      "loss": 2.5315,
      "step": 140
    },
    {
      "epoch": 0.0034135129599708715,
      "grad_norm": 256.0,
      "learning_rate": 3.390216154721274e-06,
      "loss": 2.2956,
      "step": 150
    },
    {
      "epoch": 0.003641080490635596,
      "grad_norm": 334.0,
      "learning_rate": 3.617747440273038e-06,
      "loss": 1.6865,
      "step": 160
    },
    {
      "epoch": 0.003868648021300321,
      "grad_norm": 828.0,
      "learning_rate": 3.845278725824802e-06,
      "loss": 1.9043,
      "step": 170
    },
    {
      "epoch": 0.004096215551965046,
      "grad_norm": 76.0,
      "learning_rate": 4.072810011376565e-06,
      "loss": 1.1723,
      "step": 180
    },
    {
      "epoch": 0.0043237830826297705,
      "grad_norm": 756.0,
      "learning_rate": 4.300341296928328e-06,
      "loss": 1.1716,
      "step": 190
    },
    {
      "epoch": 0.004551350613294495,
      "grad_norm": 53.75,
      "learning_rate": 4.527872582480091e-06,
      "loss": 1.0108,
      "step": 200
    },
    {
      "epoch": 0.00477891814395922,
      "grad_norm": 976.0,
      "learning_rate": 4.755403868031855e-06,
      "loss": 1.2443,
      "step": 210
    },
    {
      "epoch": 0.005006485674623945,
      "grad_norm": 201.0,
      "learning_rate": 4.982935153583618e-06,
      "loss": 1.0298,
      "step": 220
    },
    {
      "epoch": 0.0052340532052886695,
      "grad_norm": 556.0,
      "learning_rate": 5.210466439135382e-06,
      "loss": 1.5186,
      "step": 230
    },
    {
      "epoch": 0.005461620735953394,
      "grad_norm": 20.0,
      "learning_rate": 5.437997724687145e-06,
      "loss": 0.6934,
      "step": 240
    },
    {
      "epoch": 0.005689188266618119,
      "grad_norm": 234.0,
      "learning_rate": 5.665529010238908e-06,
      "loss": 1.6481,
      "step": 250
    },
    {
      "epoch": 0.005916755797282844,
      "grad_norm": 98.0,
      "learning_rate": 5.893060295790671e-06,
      "loss": 1.8,
      "step": 260
    },
    {
      "epoch": 0.0061443233279475686,
      "grad_norm": 1.453125,
      "learning_rate": 6.120591581342435e-06,
      "loss": 1.9409,
      "step": 270
    },
    {
      "epoch": 0.006371890858612293,
      "grad_norm": 452.0,
      "learning_rate": 6.348122866894198e-06,
      "loss": 1.3056,
      "step": 280
    },
    {
      "epoch": 0.006599458389277018,
      "grad_norm": 212.0,
      "learning_rate": 6.575654152445961e-06,
      "loss": 1.5522,
      "step": 290
    },
    {
      "epoch": 0.006827025919941743,
      "grad_norm": 944.0,
      "learning_rate": 6.803185437997726e-06,
      "loss": 0.947,
      "step": 300
    },
    {
      "epoch": 0.0070545934506064676,
      "grad_norm": 728.0,
      "learning_rate": 7.030716723549489e-06,
      "loss": 1.137,
      "step": 310
    },
    {
      "epoch": 0.007282160981271192,
      "grad_norm": 126.0,
      "learning_rate": 7.258248009101252e-06,
      "loss": 0.692,
      "step": 320
    },
    {
      "epoch": 0.007509728511935917,
      "grad_norm": 215.0,
      "learning_rate": 7.485779294653015e-06,
      "loss": 0.93,
      "step": 330
    },
    {
      "epoch": 0.007737296042600642,
      "grad_norm": 732.0,
      "learning_rate": 7.713310580204779e-06,
      "loss": 1.4648,
      "step": 340
    },
    {
      "epoch": 0.007964863573265367,
      "grad_norm": 1.0625,
      "learning_rate": 7.940841865756542e-06,
      "loss": 1.1252,
      "step": 350
    },
    {
      "epoch": 0.008192431103930092,
      "grad_norm": 13.9375,
      "learning_rate": 8.168373151308305e-06,
      "loss": 1.1186,
      "step": 360
    },
    {
      "epoch": 0.008419998634594816,
      "grad_norm": 108.5,
      "learning_rate": 8.395904436860068e-06,
      "loss": 1.6202,
      "step": 370
    },
    {
      "epoch": 0.008647566165259541,
      "grad_norm": 0.033447265625,
      "learning_rate": 8.623435722411833e-06,
      "loss": 1.4176,
      "step": 380
    },
    {
      "epoch": 0.008875133695924265,
      "grad_norm": 450.0,
      "learning_rate": 8.850967007963596e-06,
      "loss": 1.8093,
      "step": 390
    },
    {
      "epoch": 0.00910270122658899,
      "grad_norm": 95.0,
      "learning_rate": 9.078498293515359e-06,
      "loss": 1.8751,
      "step": 400
    },
    {
      "epoch": 0.009330268757253716,
      "grad_norm": 185.0,
      "learning_rate": 9.306029579067123e-06,
      "loss": 1.7214,
      "step": 410
    },
    {
      "epoch": 0.00955783628791844,
      "grad_norm": 7.59375,
      "learning_rate": 9.533560864618886e-06,
      "loss": 0.9538,
      "step": 420
    },
    {
      "epoch": 0.009785403818583165,
      "grad_norm": 370.0,
      "learning_rate": 9.76109215017065e-06,
      "loss": 1.9393,
      "step": 430
    },
    {
      "epoch": 0.01001297134924789,
      "grad_norm": 178.0,
      "learning_rate": 9.988623435722412e-06,
      "loss": 1.9089,
      "step": 440
    },
    {
      "epoch": 0.010240538879912614,
      "grad_norm": 342.0,
      "learning_rate": 1.0216154721274177e-05,
      "loss": 0.6716,
      "step": 450
    },
    {
      "epoch": 0.010468106410577339,
      "grad_norm": 241.0,
      "learning_rate": 1.044368600682594e-05,
      "loss": 2.0107,
      "step": 460
    },
    {
      "epoch": 0.010695673941242063,
      "grad_norm": 202.0,
      "learning_rate": 1.0671217292377703e-05,
      "loss": 1.106,
      "step": 470
    },
    {
      "epoch": 0.010923241471906788,
      "grad_norm": 272.0,
      "learning_rate": 1.0898748577929466e-05,
      "loss": 1.1183,
      "step": 480
    },
    {
      "epoch": 0.011150809002571514,
      "grad_norm": 187.0,
      "learning_rate": 1.1126279863481229e-05,
      "loss": 0.7857,
      "step": 490
    },
    {
      "epoch": 0.011378376533236237,
      "grad_norm": 80.5,
      "learning_rate": 1.1353811149032994e-05,
      "loss": 1.2253,
      "step": 500
    },
    {
      "epoch": 0.011605944063900963,
      "grad_norm": 280.0,
      "learning_rate": 1.1581342434584757e-05,
      "loss": 0.9281,
      "step": 510
    },
    {
      "epoch": 0.011833511594565688,
      "grad_norm": 178.0,
      "learning_rate": 1.180887372013652e-05,
      "loss": 1.0197,
      "step": 520
    },
    {
      "epoch": 0.012061079125230412,
      "grad_norm": 45.25,
      "learning_rate": 1.2036405005688283e-05,
      "loss": 0.924,
      "step": 530
    },
    {
      "epoch": 0.012288646655895137,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2263936291240046e-05,
      "loss": 1.0238,
      "step": 540
    },
    {
      "epoch": 0.01251621418655986,
      "grad_norm": 0.01190185546875,
      "learning_rate": 1.2491467576791809e-05,
      "loss": 0.7098,
      "step": 550
    },
    {
      "epoch": 0.012743781717224586,
      "grad_norm": 24.625,
      "learning_rate": 1.2718998862343572e-05,
      "loss": 0.7466,
      "step": 560
    },
    {
      "epoch": 0.012971349247889312,
      "grad_norm": 96.0,
      "learning_rate": 1.2946530147895338e-05,
      "loss": 0.7371,
      "step": 570
    },
    {
      "epoch": 0.013198916778554035,
      "grad_norm": 233.0,
      "learning_rate": 1.3174061433447101e-05,
      "loss": 2.01,
      "step": 580
    },
    {
      "epoch": 0.01342648430921876,
      "grad_norm": 5.4375,
      "learning_rate": 1.3401592718998864e-05,
      "loss": 0.6927,
      "step": 590
    },
    {
      "epoch": 0.013654051839883486,
      "grad_norm": 164.0,
      "learning_rate": 1.3629124004550627e-05,
      "loss": 1.0115,
      "step": 600
    },
    {
      "epoch": 0.01388161937054821,
      "grad_norm": 278.0,
      "learning_rate": 1.385665529010239e-05,
      "loss": 1.4222,
      "step": 610
    },
    {
      "epoch": 0.014109186901212935,
      "grad_norm": 20.5,
      "learning_rate": 1.4084186575654153e-05,
      "loss": 1.1939,
      "step": 620
    },
    {
      "epoch": 0.01433675443187766,
      "grad_norm": 184.0,
      "learning_rate": 1.4311717861205916e-05,
      "loss": 1.2414,
      "step": 630
    },
    {
      "epoch": 0.014564321962542384,
      "grad_norm": 78.5,
      "learning_rate": 1.4539249146757681e-05,
      "loss": 1.2714,
      "step": 640
    },
    {
      "epoch": 0.01479188949320711,
      "grad_norm": 300.0,
      "learning_rate": 1.4766780432309444e-05,
      "loss": 1.0122,
      "step": 650
    },
    {
      "epoch": 0.015019457023871833,
      "grad_norm": 318.0,
      "learning_rate": 1.4994311717861207e-05,
      "loss": 1.2019,
      "step": 660
    },
    {
      "epoch": 0.015247024554536559,
      "grad_norm": 692.0,
      "learning_rate": 1.522184300341297e-05,
      "loss": 1.1806,
      "step": 670
    },
    {
      "epoch": 0.015474592085201284,
      "grad_norm": 5.90625,
      "learning_rate": 1.5449374288964735e-05,
      "loss": 0.9984,
      "step": 680
    },
    {
      "epoch": 0.015702159615866008,
      "grad_norm": 524.0,
      "learning_rate": 1.5676905574516498e-05,
      "loss": 1.6459,
      "step": 690
    },
    {
      "epoch": 0.015929727146530733,
      "grad_norm": 53.0,
      "learning_rate": 1.590443686006826e-05,
      "loss": 0.3876,
      "step": 700
    },
    {
      "epoch": 0.01615729467719546,
      "grad_norm": 150.0,
      "learning_rate": 1.6131968145620024e-05,
      "loss": 1.0511,
      "step": 710
    },
    {
      "epoch": 0.016384862207860184,
      "grad_norm": 532.0,
      "learning_rate": 1.6359499431171787e-05,
      "loss": 1.1115,
      "step": 720
    },
    {
      "epoch": 0.016612429738524906,
      "grad_norm": 212.0,
      "learning_rate": 1.658703071672355e-05,
      "loss": 1.2553,
      "step": 730
    },
    {
      "epoch": 0.01683999726918963,
      "grad_norm": 161.0,
      "learning_rate": 1.6814562002275313e-05,
      "loss": 1.1006,
      "step": 740
    },
    {
      "epoch": 0.017067564799854357,
      "grad_norm": 17.625,
      "learning_rate": 1.7042093287827076e-05,
      "loss": 1.7845,
      "step": 750
    },
    {
      "epoch": 0.017295132330519082,
      "grad_norm": 158.0,
      "learning_rate": 1.726962457337884e-05,
      "loss": 1.8899,
      "step": 760
    },
    {
      "epoch": 0.017522699861183808,
      "grad_norm": 61.75,
      "learning_rate": 1.7497155858930602e-05,
      "loss": 0.6635,
      "step": 770
    },
    {
      "epoch": 0.01775026739184853,
      "grad_norm": 380.0,
      "learning_rate": 1.7724687144482368e-05,
      "loss": 0.9673,
      "step": 780
    },
    {
      "epoch": 0.017977834922513255,
      "grad_norm": 270.0,
      "learning_rate": 1.795221843003413e-05,
      "loss": 2.6468,
      "step": 790
    },
    {
      "epoch": 0.01820540245317798,
      "grad_norm": 211.0,
      "learning_rate": 1.8179749715585894e-05,
      "loss": 0.5992,
      "step": 800
    },
    {
      "epoch": 0.018432969983842706,
      "grad_norm": 176.0,
      "learning_rate": 1.8407281001137657e-05,
      "loss": 0.5747,
      "step": 810
    },
    {
      "epoch": 0.01866053751450743,
      "grad_norm": 692.0,
      "learning_rate": 1.863481228668942e-05,
      "loss": 0.9565,
      "step": 820
    },
    {
      "epoch": 0.018888105045172156,
      "grad_norm": 366.0,
      "learning_rate": 1.8862343572241183e-05,
      "loss": 2.7505,
      "step": 830
    },
    {
      "epoch": 0.01911567257583688,
      "grad_norm": 80.0,
      "learning_rate": 1.9089874857792946e-05,
      "loss": 1.6672,
      "step": 840
    },
    {
      "epoch": 0.019343240106501604,
      "grad_norm": 0.052734375,
      "learning_rate": 1.9317406143344713e-05,
      "loss": 1.2575,
      "step": 850
    },
    {
      "epoch": 0.01957080763716633,
      "grad_norm": 48.75,
      "learning_rate": 1.9544937428896476e-05,
      "loss": 1.5053,
      "step": 860
    },
    {
      "epoch": 0.019798375167831055,
      "grad_norm": 4.71875,
      "learning_rate": 1.977246871444824e-05,
      "loss": 0.8211,
      "step": 870
    },
    {
      "epoch": 0.02002594269849578,
      "grad_norm": 35.5,
      "learning_rate": 2e-05,
      "loss": 1.5695,
      "step": 880
    },
    {
      "epoch": 0.020253510229160502,
      "grad_norm": 62.25,
      "learning_rate": 1.999535574958202e-05,
      "loss": 1.2563,
      "step": 890
    },
    {
      "epoch": 0.020481077759825227,
      "grad_norm": 184.0,
      "learning_rate": 1.9990711499164036e-05,
      "loss": 1.0591,
      "step": 900
    },
    {
      "epoch": 0.020708645290489953,
      "grad_norm": 494.0,
      "learning_rate": 1.9986067248746056e-05,
      "loss": 1.5684,
      "step": 910
    },
    {
      "epoch": 0.020936212821154678,
      "grad_norm": 50.0,
      "learning_rate": 1.998142299832807e-05,
      "loss": 1.0449,
      "step": 920
    },
    {
      "epoch": 0.021163780351819404,
      "grad_norm": 99.0,
      "learning_rate": 1.9976778747910087e-05,
      "loss": 1.009,
      "step": 930
    },
    {
      "epoch": 0.021391347882484125,
      "grad_norm": 132.0,
      "learning_rate": 1.9972134497492107e-05,
      "loss": 1.0228,
      "step": 940
    },
    {
      "epoch": 0.02161891541314885,
      "grad_norm": 532.0,
      "learning_rate": 1.9967490247074124e-05,
      "loss": 2.0096,
      "step": 950
    },
    {
      "epoch": 0.021846482943813576,
      "grad_norm": 61.0,
      "learning_rate": 1.996284599665614e-05,
      "loss": 1.0553,
      "step": 960
    },
    {
      "epoch": 0.0220740504744783,
      "grad_norm": 104.5,
      "learning_rate": 1.995820174623816e-05,
      "loss": 1.25,
      "step": 970
    },
    {
      "epoch": 0.022301618005143027,
      "grad_norm": 42.75,
      "learning_rate": 1.9953557495820176e-05,
      "loss": 0.8325,
      "step": 980
    },
    {
      "epoch": 0.022529185535807753,
      "grad_norm": 0.1728515625,
      "learning_rate": 1.9948913245402193e-05,
      "loss": 0.5951,
      "step": 990
    },
    {
      "epoch": 0.022756753066472474,
      "grad_norm": 143.0,
      "learning_rate": 1.994426899498421e-05,
      "loss": 0.3057,
      "step": 1000
    },
    {
      "epoch": 0.0229843205971372,
      "grad_norm": 106.0,
      "learning_rate": 1.9939624744566227e-05,
      "loss": 1.384,
      "step": 1010
    },
    {
      "epoch": 0.023211888127801925,
      "grad_norm": 211.0,
      "learning_rate": 1.9934980494148247e-05,
      "loss": 2.4723,
      "step": 1020
    },
    {
      "epoch": 0.02343945565846665,
      "grad_norm": 564.0,
      "learning_rate": 1.9930336243730264e-05,
      "loss": 1.3514,
      "step": 1030
    },
    {
      "epoch": 0.023667023189131376,
      "grad_norm": 7.5,
      "learning_rate": 1.992569199331228e-05,
      "loss": 2.2944,
      "step": 1040
    },
    {
      "epoch": 0.023894590719796098,
      "grad_norm": 0.09423828125,
      "learning_rate": 1.99210477428943e-05,
      "loss": 0.3816,
      "step": 1050
    },
    {
      "epoch": 0.024122158250460823,
      "grad_norm": 0.001495361328125,
      "learning_rate": 1.9916403492476315e-05,
      "loss": 0.9169,
      "step": 1060
    },
    {
      "epoch": 0.02434972578112555,
      "grad_norm": 118.0,
      "learning_rate": 1.9911759242058332e-05,
      "loss": 1.8365,
      "step": 1070
    },
    {
      "epoch": 0.024577293311790274,
      "grad_norm": 113.5,
      "learning_rate": 1.990711499164035e-05,
      "loss": 1.2963,
      "step": 1080
    },
    {
      "epoch": 0.024804860842455,
      "grad_norm": 0.083984375,
      "learning_rate": 1.990247074122237e-05,
      "loss": 0.5726,
      "step": 1090
    },
    {
      "epoch": 0.02503242837311972,
      "grad_norm": 148.0,
      "learning_rate": 1.9897826490804387e-05,
      "loss": 0.8862,
      "step": 1100
    },
    {
      "epoch": 0.025259995903784447,
      "grad_norm": 133.0,
      "learning_rate": 1.9893182240386404e-05,
      "loss": 0.4664,
      "step": 1110
    },
    {
      "epoch": 0.025487563434449172,
      "grad_norm": 127.0,
      "learning_rate": 1.988853798996842e-05,
      "loss": 1.0277,
      "step": 1120
    },
    {
      "epoch": 0.025715130965113898,
      "grad_norm": 141.0,
      "learning_rate": 1.9883893739550438e-05,
      "loss": 2.3615,
      "step": 1130
    },
    {
      "epoch": 0.025942698495778623,
      "grad_norm": 0.26171875,
      "learning_rate": 1.9879249489132455e-05,
      "loss": 0.6918,
      "step": 1140
    },
    {
      "epoch": 0.02617026602644335,
      "grad_norm": 212.0,
      "learning_rate": 1.9874605238714472e-05,
      "loss": 1.9012,
      "step": 1150
    },
    {
      "epoch": 0.02639783355710807,
      "grad_norm": 0.0908203125,
      "learning_rate": 1.986996098829649e-05,
      "loss": 1.6964,
      "step": 1160
    },
    {
      "epoch": 0.026625401087772796,
      "grad_norm": 494.0,
      "learning_rate": 1.986531673787851e-05,
      "loss": 1.7084,
      "step": 1170
    },
    {
      "epoch": 0.02685296861843752,
      "grad_norm": 412.0,
      "learning_rate": 1.9860672487460527e-05,
      "loss": 1.8972,
      "step": 1180
    },
    {
      "epoch": 0.027080536149102247,
      "grad_norm": 157.0,
      "learning_rate": 1.9856028237042544e-05,
      "loss": 1.0771,
      "step": 1190
    },
    {
      "epoch": 0.027308103679766972,
      "grad_norm": 89.5,
      "learning_rate": 1.985138398662456e-05,
      "loss": 1.5232,
      "step": 1200
    },
    {
      "epoch": 0.027535671210431694,
      "grad_norm": 2.046875,
      "learning_rate": 1.9846739736206578e-05,
      "loss": 0.568,
      "step": 1210
    },
    {
      "epoch": 0.02776323874109642,
      "grad_norm": 0.00019168853759765625,
      "learning_rate": 1.9842095485788595e-05,
      "loss": 1.0279,
      "step": 1220
    },
    {
      "epoch": 0.027990806271761145,
      "grad_norm": 314.0,
      "learning_rate": 1.9837451235370612e-05,
      "loss": 1.6179,
      "step": 1230
    },
    {
      "epoch": 0.02821837380242587,
      "grad_norm": 920.0,
      "learning_rate": 1.983280698495263e-05,
      "loss": 1.7443,
      "step": 1240
    },
    {
      "epoch": 0.028445941333090596,
      "grad_norm": 93.0,
      "learning_rate": 1.982816273453465e-05,
      "loss": 0.8336,
      "step": 1250
    },
    {
      "epoch": 0.02867350886375532,
      "grad_norm": 0.0016937255859375,
      "learning_rate": 1.9823518484116667e-05,
      "loss": 1.2344,
      "step": 1260
    },
    {
      "epoch": 0.028901076394420043,
      "grad_norm": 146.0,
      "learning_rate": 1.981887423369868e-05,
      "loss": 1.2899,
      "step": 1270
    },
    {
      "epoch": 0.02912864392508477,
      "grad_norm": 145.0,
      "learning_rate": 1.98142299832807e-05,
      "loss": 1.1792,
      "step": 1280
    },
    {
      "epoch": 0.029356211455749494,
      "grad_norm": 131.0,
      "learning_rate": 1.9809585732862718e-05,
      "loss": 1.18,
      "step": 1290
    },
    {
      "epoch": 0.02958377898641422,
      "grad_norm": 324.0,
      "learning_rate": 1.9804941482444735e-05,
      "loss": 1.3393,
      "step": 1300
    },
    {
      "epoch": 0.029811346517078945,
      "grad_norm": 134.0,
      "learning_rate": 1.9800297232026752e-05,
      "loss": 1.2237,
      "step": 1310
    },
    {
      "epoch": 0.030038914047743667,
      "grad_norm": 13.3125,
      "learning_rate": 1.979565298160877e-05,
      "loss": 1.9049,
      "step": 1320
    },
    {
      "epoch": 0.030266481578408392,
      "grad_norm": 0.0030517578125,
      "learning_rate": 1.979100873119079e-05,
      "loss": 0.7462,
      "step": 1330
    },
    {
      "epoch": 0.030494049109073117,
      "grad_norm": 252.0,
      "learning_rate": 1.9786364480772803e-05,
      "loss": 2.214,
      "step": 1340
    },
    {
      "epoch": 0.030721616639737843,
      "grad_norm": 440.0,
      "learning_rate": 1.978172023035482e-05,
      "loss": 1.1145,
      "step": 1350
    },
    {
      "epoch": 0.030949184170402568,
      "grad_norm": 169.0,
      "learning_rate": 1.977707597993684e-05,
      "loss": 1.1504,
      "step": 1360
    },
    {
      "epoch": 0.03117675170106729,
      "grad_norm": 51.5,
      "learning_rate": 1.9772431729518858e-05,
      "loss": 1.1276,
      "step": 1370
    },
    {
      "epoch": 0.031404319231732016,
      "grad_norm": 202.0,
      "learning_rate": 1.9767787479100875e-05,
      "loss": 1.4093,
      "step": 1380
    },
    {
      "epoch": 0.03163188676239674,
      "grad_norm": 8.5,
      "learning_rate": 1.9763143228682892e-05,
      "loss": 0.7489,
      "step": 1390
    },
    {
      "epoch": 0.031859454293061466,
      "grad_norm": 49.0,
      "learning_rate": 1.9758498978264912e-05,
      "loss": 1.858,
      "step": 1400
    },
    {
      "epoch": 0.03208702182372619,
      "grad_norm": 80.5,
      "learning_rate": 1.9753854727846926e-05,
      "loss": 0.9938,
      "step": 1410
    },
    {
      "epoch": 0.03231458935439092,
      "grad_norm": 16.375,
      "learning_rate": 1.9749210477428943e-05,
      "loss": 1.2759,
      "step": 1420
    },
    {
      "epoch": 0.03254215688505564,
      "grad_norm": 46.75,
      "learning_rate": 1.9744566227010963e-05,
      "loss": 1.0914,
      "step": 1430
    },
    {
      "epoch": 0.03276972441572037,
      "grad_norm": 0.484375,
      "learning_rate": 1.973992197659298e-05,
      "loss": 0.8914,
      "step": 1440
    },
    {
      "epoch": 0.032997291946385086,
      "grad_norm": 196.0,
      "learning_rate": 1.9735277726174997e-05,
      "loss": 1.3525,
      "step": 1450
    },
    {
      "epoch": 0.03322485947704981,
      "grad_norm": 216.0,
      "learning_rate": 1.9730633475757015e-05,
      "loss": 1.2875,
      "step": 1460
    },
    {
      "epoch": 0.03345242700771454,
      "grad_norm": 130.0,
      "learning_rate": 1.972598922533903e-05,
      "loss": 0.8692,
      "step": 1470
    },
    {
      "epoch": 0.03367999453837926,
      "grad_norm": 34.5,
      "learning_rate": 1.972134497492105e-05,
      "loss": 1.2848,
      "step": 1480
    },
    {
      "epoch": 0.03390756206904399,
      "grad_norm": 1.3046875,
      "learning_rate": 1.9716700724503066e-05,
      "loss": 1.1641,
      "step": 1490
    },
    {
      "epoch": 0.03413512959970871,
      "grad_norm": 95.0,
      "learning_rate": 1.9712056474085083e-05,
      "loss": 0.8262,
      "step": 1500
    },
    {
      "epoch": 0.03436269713037344,
      "grad_norm": 0.1142578125,
      "learning_rate": 1.9707412223667103e-05,
      "loss": 1.7248,
      "step": 1510
    },
    {
      "epoch": 0.034590264661038164,
      "grad_norm": 330.0,
      "learning_rate": 1.970276797324912e-05,
      "loss": 0.7362,
      "step": 1520
    },
    {
      "epoch": 0.03481783219170289,
      "grad_norm": 135.0,
      "learning_rate": 1.9698123722831137e-05,
      "loss": 0.5457,
      "step": 1530
    },
    {
      "epoch": 0.035045399722367615,
      "grad_norm": 23.75,
      "learning_rate": 1.9693479472413154e-05,
      "loss": 0.5933,
      "step": 1540
    },
    {
      "epoch": 0.03527296725303234,
      "grad_norm": 188.0,
      "learning_rate": 1.968883522199517e-05,
      "loss": 1.4926,
      "step": 1550
    },
    {
      "epoch": 0.03550053478369706,
      "grad_norm": 0.5625,
      "learning_rate": 1.968419097157719e-05,
      "loss": 0.7648,
      "step": 1560
    },
    {
      "epoch": 0.035728102314361784,
      "grad_norm": 0.06005859375,
      "learning_rate": 1.9679546721159206e-05,
      "loss": 1.7118,
      "step": 1570
    },
    {
      "epoch": 0.03595566984502651,
      "grad_norm": 616.0,
      "learning_rate": 1.9674902470741223e-05,
      "loss": 1.1693,
      "step": 1580
    },
    {
      "epoch": 0.036183237375691235,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9670258220323243e-05,
      "loss": 0.4716,
      "step": 1590
    },
    {
      "epoch": 0.03641080490635596,
      "grad_norm": 253.0,
      "learning_rate": 1.966561396990526e-05,
      "loss": 0.9428,
      "step": 1600
    },
    {
      "epoch": 0.036638372437020686,
      "grad_norm": 450.0,
      "learning_rate": 1.9660969719487277e-05,
      "loss": 1.91,
      "step": 1610
    },
    {
      "epoch": 0.03686593996768541,
      "grad_norm": 218.0,
      "learning_rate": 1.9656325469069294e-05,
      "loss": 0.4805,
      "step": 1620
    },
    {
      "epoch": 0.03709350749835014,
      "grad_norm": 2.09375,
      "learning_rate": 1.965168121865131e-05,
      "loss": 1.2511,
      "step": 1630
    },
    {
      "epoch": 0.03732107502901486,
      "grad_norm": 7.65625,
      "learning_rate": 1.9647036968233328e-05,
      "loss": 0.8677,
      "step": 1640
    },
    {
      "epoch": 0.03754864255967959,
      "grad_norm": 7.84375,
      "learning_rate": 1.9642392717815345e-05,
      "loss": 0.6325,
      "step": 1650
    },
    {
      "epoch": 0.03777621009034431,
      "grad_norm": 1.203125,
      "learning_rate": 1.9637748467397362e-05,
      "loss": 0.6305,
      "step": 1660
    },
    {
      "epoch": 0.03800377762100903,
      "grad_norm": 0.0186767578125,
      "learning_rate": 1.9633104216979383e-05,
      "loss": 0.8823,
      "step": 1670
    },
    {
      "epoch": 0.03823134515167376,
      "grad_norm": 218.0,
      "learning_rate": 1.96284599665614e-05,
      "loss": 1.872,
      "step": 1680
    },
    {
      "epoch": 0.03845891268233848,
      "grad_norm": 80.0,
      "learning_rate": 1.9623815716143414e-05,
      "loss": 2.0605,
      "step": 1690
    },
    {
      "epoch": 0.03868648021300321,
      "grad_norm": 300.0,
      "learning_rate": 1.9619171465725434e-05,
      "loss": 1.421,
      "step": 1700
    },
    {
      "epoch": 0.03891404774366793,
      "grad_norm": 1.25,
      "learning_rate": 1.961452721530745e-05,
      "loss": 0.7913,
      "step": 1710
    },
    {
      "epoch": 0.03914161527433266,
      "grad_norm": 60.0,
      "learning_rate": 1.9609882964889468e-05,
      "loss": 0.6761,
      "step": 1720
    },
    {
      "epoch": 0.039369182804997384,
      "grad_norm": 162.0,
      "learning_rate": 1.9605238714471485e-05,
      "loss": 1.4394,
      "step": 1730
    },
    {
      "epoch": 0.03959675033566211,
      "grad_norm": 74.5,
      "learning_rate": 1.9600594464053506e-05,
      "loss": 1.4238,
      "step": 1740
    },
    {
      "epoch": 0.039824317866326835,
      "grad_norm": 8.5,
      "learning_rate": 1.9595950213635523e-05,
      "loss": 0.8131,
      "step": 1750
    },
    {
      "epoch": 0.04005188539699156,
      "grad_norm": 38.5,
      "learning_rate": 1.9591305963217536e-05,
      "loss": 0.4427,
      "step": 1760
    },
    {
      "epoch": 0.040279452927656285,
      "grad_norm": 155.0,
      "learning_rate": 1.9586661712799553e-05,
      "loss": 0.8234,
      "step": 1770
    },
    {
      "epoch": 0.040507020458321004,
      "grad_norm": 14.8125,
      "learning_rate": 1.9582017462381574e-05,
      "loss": 1.557,
      "step": 1780
    },
    {
      "epoch": 0.04073458798898573,
      "grad_norm": 239.0,
      "learning_rate": 1.957737321196359e-05,
      "loss": 1.9888,
      "step": 1790
    },
    {
      "epoch": 0.040962155519650455,
      "grad_norm": 23.5,
      "learning_rate": 1.9572728961545608e-05,
      "loss": 0.8928,
      "step": 1800
    },
    {
      "epoch": 0.04118972305031518,
      "grad_norm": 27.25,
      "learning_rate": 1.9568084711127625e-05,
      "loss": 1.3141,
      "step": 1810
    },
    {
      "epoch": 0.041417290580979906,
      "grad_norm": 1.125,
      "learning_rate": 1.9563440460709645e-05,
      "loss": 1.3772,
      "step": 1820
    },
    {
      "epoch": 0.04164485811164463,
      "grad_norm": 201.0,
      "learning_rate": 1.955879621029166e-05,
      "loss": 2.2851,
      "step": 1830
    },
    {
      "epoch": 0.041872425642309356,
      "grad_norm": 134.0,
      "learning_rate": 1.9554151959873676e-05,
      "loss": 1.3275,
      "step": 1840
    },
    {
      "epoch": 0.04209999317297408,
      "grad_norm": 143.0,
      "learning_rate": 1.9549507709455697e-05,
      "loss": 1.6835,
      "step": 1850
    },
    {
      "epoch": 0.04232756070363881,
      "grad_norm": 105.5,
      "learning_rate": 1.9544863459037714e-05,
      "loss": 1.5849,
      "step": 1860
    },
    {
      "epoch": 0.04255512823430353,
      "grad_norm": 55.25,
      "learning_rate": 1.954021920861973e-05,
      "loss": 0.785,
      "step": 1870
    },
    {
      "epoch": 0.04278269576496825,
      "grad_norm": 239.0,
      "learning_rate": 1.9535574958201748e-05,
      "loss": 1.3229,
      "step": 1880
    },
    {
      "epoch": 0.043010263295632976,
      "grad_norm": 11.0625,
      "learning_rate": 1.9530930707783765e-05,
      "loss": 0.8176,
      "step": 1890
    },
    {
      "epoch": 0.0432378308262977,
      "grad_norm": 0.0126953125,
      "learning_rate": 1.9526286457365782e-05,
      "loss": 0.9469,
      "step": 1900
    },
    {
      "epoch": 0.04346539835696243,
      "grad_norm": 251.0,
      "learning_rate": 1.95216422069478e-05,
      "loss": 1.5199,
      "step": 1910
    },
    {
      "epoch": 0.04369296588762715,
      "grad_norm": 227.0,
      "learning_rate": 1.9516997956529816e-05,
      "loss": 0.6017,
      "step": 1920
    },
    {
      "epoch": 0.04392053341829188,
      "grad_norm": 223.0,
      "learning_rate": 1.9512353706111836e-05,
      "loss": 0.2719,
      "step": 1930
    },
    {
      "epoch": 0.0441481009489566,
      "grad_norm": 188.0,
      "learning_rate": 1.9507709455693854e-05,
      "loss": 3.0157,
      "step": 1940
    },
    {
      "epoch": 0.04437566847962133,
      "grad_norm": 50.5,
      "learning_rate": 1.950306520527587e-05,
      "loss": 0.5819,
      "step": 1950
    },
    {
      "epoch": 0.044603236010286054,
      "grad_norm": 0.185546875,
      "learning_rate": 1.9498420954857888e-05,
      "loss": 0.272,
      "step": 1960
    },
    {
      "epoch": 0.04483080354095078,
      "grad_norm": 572.0,
      "learning_rate": 1.9493776704439905e-05,
      "loss": 1.2198,
      "step": 1970
    },
    {
      "epoch": 0.045058371071615505,
      "grad_norm": 13.3125,
      "learning_rate": 1.9489132454021922e-05,
      "loss": 1.1429,
      "step": 1980
    },
    {
      "epoch": 0.045285938602280223,
      "grad_norm": 0.09130859375,
      "learning_rate": 1.948448820360394e-05,
      "loss": 0.8416,
      "step": 1990
    },
    {
      "epoch": 0.04551350613294495,
      "grad_norm": 64.0,
      "learning_rate": 1.9479843953185956e-05,
      "loss": 0.7706,
      "step": 2000
    },
    {
      "epoch": 0.045741073663609674,
      "grad_norm": 147.0,
      "learning_rate": 1.9475199702767976e-05,
      "loss": 0.6702,
      "step": 2010
    },
    {
      "epoch": 0.0459686411942744,
      "grad_norm": 142.0,
      "learning_rate": 1.9470555452349993e-05,
      "loss": 2.0384,
      "step": 2020
    },
    {
      "epoch": 0.046196208724939125,
      "grad_norm": 87.0,
      "learning_rate": 1.946591120193201e-05,
      "loss": 0.8487,
      "step": 2030
    },
    {
      "epoch": 0.04642377625560385,
      "grad_norm": 0.2314453125,
      "learning_rate": 1.9461266951514027e-05,
      "loss": 0.8028,
      "step": 2040
    },
    {
      "epoch": 0.046651343786268576,
      "grad_norm": 100.0,
      "learning_rate": 1.9456622701096044e-05,
      "loss": 1.4738,
      "step": 2050
    },
    {
      "epoch": 0.0468789113169333,
      "grad_norm": 96.5,
      "learning_rate": 1.945197845067806e-05,
      "loss": 0.7392,
      "step": 2060
    },
    {
      "epoch": 0.04710647884759803,
      "grad_norm": 476.0,
      "learning_rate": 1.944733420026008e-05,
      "loss": 1.2594,
      "step": 2070
    },
    {
      "epoch": 0.04733404637826275,
      "grad_norm": 152.0,
      "learning_rate": 1.9442689949842096e-05,
      "loss": 1.0102,
      "step": 2080
    },
    {
      "epoch": 0.04756161390892748,
      "grad_norm": 0.19140625,
      "learning_rate": 1.9438045699424116e-05,
      "loss": 1.063,
      "step": 2090
    },
    {
      "epoch": 0.047789181439592196,
      "grad_norm": 180.0,
      "learning_rate": 1.9433401449006133e-05,
      "loss": 1.1975,
      "step": 2100
    },
    {
      "epoch": 0.04801674897025692,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9428757198588147e-05,
      "loss": 1.7231,
      "step": 2110
    },
    {
      "epoch": 0.04824431650092165,
      "grad_norm": 0.251953125,
      "learning_rate": 1.9424112948170167e-05,
      "loss": 1.8199,
      "step": 2120
    },
    {
      "epoch": 0.04847188403158637,
      "grad_norm": 0.58203125,
      "learning_rate": 1.9419468697752184e-05,
      "loss": 0.9709,
      "step": 2130
    },
    {
      "epoch": 0.0486994515622511,
      "grad_norm": 201.0,
      "learning_rate": 1.94148244473342e-05,
      "loss": 0.8072,
      "step": 2140
    },
    {
      "epoch": 0.04892701909291582,
      "grad_norm": 52.25,
      "learning_rate": 1.941018019691622e-05,
      "loss": 0.488,
      "step": 2150
    },
    {
      "epoch": 0.04915458662358055,
      "grad_norm": 196.0,
      "learning_rate": 1.940553594649824e-05,
      "loss": 0.6251,
      "step": 2160
    },
    {
      "epoch": 0.049382154154245274,
      "grad_norm": 6.375,
      "learning_rate": 1.9400891696080256e-05,
      "loss": 0.5646,
      "step": 2170
    },
    {
      "epoch": 0.04960972168491,
      "grad_norm": 294.0,
      "learning_rate": 1.939624744566227e-05,
      "loss": 1.6046,
      "step": 2180
    },
    {
      "epoch": 0.049837289215574725,
      "grad_norm": 336.0,
      "learning_rate": 1.939160319524429e-05,
      "loss": 2.6019,
      "step": 2190
    },
    {
      "epoch": 0.05006485674623944,
      "grad_norm": 356.0,
      "learning_rate": 1.9386958944826307e-05,
      "loss": 0.6041,
      "step": 2200
    },
    {
      "epoch": 0.05029242427690417,
      "grad_norm": 48.25,
      "learning_rate": 1.9382314694408324e-05,
      "loss": 0.6656,
      "step": 2210
    },
    {
      "epoch": 0.050519991807568894,
      "grad_norm": 94.5,
      "learning_rate": 1.937767044399034e-05,
      "loss": 1.0116,
      "step": 2220
    },
    {
      "epoch": 0.05074755933823362,
      "grad_norm": 396.0,
      "learning_rate": 1.9373026193572358e-05,
      "loss": 1.7477,
      "step": 2230
    },
    {
      "epoch": 0.050975126868898345,
      "grad_norm": 122.5,
      "learning_rate": 1.936838194315438e-05,
      "loss": 1.3148,
      "step": 2240
    },
    {
      "epoch": 0.05120269439956307,
      "grad_norm": 322.0,
      "learning_rate": 1.9363737692736392e-05,
      "loss": 0.99,
      "step": 2250
    },
    {
      "epoch": 0.051430261930227796,
      "grad_norm": 4.25,
      "learning_rate": 1.935909344231841e-05,
      "loss": 0.9586,
      "step": 2260
    },
    {
      "epoch": 0.05165782946089252,
      "grad_norm": 386.0,
      "learning_rate": 1.935444919190043e-05,
      "loss": 1.4372,
      "step": 2270
    },
    {
      "epoch": 0.051885396991557246,
      "grad_norm": 226.0,
      "learning_rate": 1.9349804941482447e-05,
      "loss": 1.2,
      "step": 2280
    },
    {
      "epoch": 0.05211296452222197,
      "grad_norm": 0.09765625,
      "learning_rate": 1.9345160691064464e-05,
      "loss": 0.7493,
      "step": 2290
    },
    {
      "epoch": 0.0523405320528867,
      "grad_norm": 450.0,
      "learning_rate": 1.934051644064648e-05,
      "loss": 1.3263,
      "step": 2300
    },
    {
      "epoch": 0.052568099583551416,
      "grad_norm": 249.0,
      "learning_rate": 1.9335872190228498e-05,
      "loss": 1.5568,
      "step": 2310
    },
    {
      "epoch": 0.05279566711421614,
      "grad_norm": 111.5,
      "learning_rate": 1.9331227939810515e-05,
      "loss": 0.9231,
      "step": 2320
    },
    {
      "epoch": 0.053023234644880866,
      "grad_norm": 0.095703125,
      "learning_rate": 1.9326583689392532e-05,
      "loss": 0.8617,
      "step": 2330
    },
    {
      "epoch": 0.05325080217554559,
      "grad_norm": 66.0,
      "learning_rate": 1.932193943897455e-05,
      "loss": 1.2371,
      "step": 2340
    },
    {
      "epoch": 0.05347836970621032,
      "grad_norm": 123.5,
      "learning_rate": 1.931729518855657e-05,
      "loss": 1.4119,
      "step": 2350
    },
    {
      "epoch": 0.05370593723687504,
      "grad_norm": 0.06103515625,
      "learning_rate": 1.9312650938138587e-05,
      "loss": 1.1572,
      "step": 2360
    },
    {
      "epoch": 0.05393350476753977,
      "grad_norm": 91.5,
      "learning_rate": 1.9308006687720604e-05,
      "loss": 1.5039,
      "step": 2370
    },
    {
      "epoch": 0.05416107229820449,
      "grad_norm": 2.296875,
      "learning_rate": 1.930336243730262e-05,
      "loss": 0.7795,
      "step": 2380
    },
    {
      "epoch": 0.05438863982886922,
      "grad_norm": 0.00537109375,
      "learning_rate": 1.9298718186884638e-05,
      "loss": 0.9257,
      "step": 2390
    },
    {
      "epoch": 0.054616207359533944,
      "grad_norm": 304.0,
      "learning_rate": 1.9294073936466655e-05,
      "loss": 1.897,
      "step": 2400
    },
    {
      "epoch": 0.05484377489019867,
      "grad_norm": 326.0,
      "learning_rate": 1.9289429686048672e-05,
      "loss": 0.6469,
      "step": 2410
    },
    {
      "epoch": 0.05507134242086339,
      "grad_norm": 119.5,
      "learning_rate": 1.928478543563069e-05,
      "loss": 1.1891,
      "step": 2420
    },
    {
      "epoch": 0.055298909951528114,
      "grad_norm": 65.0,
      "learning_rate": 1.928014118521271e-05,
      "loss": 0.588,
      "step": 2430
    },
    {
      "epoch": 0.05552647748219284,
      "grad_norm": 232.0,
      "learning_rate": 1.9275496934794727e-05,
      "loss": 1.1786,
      "step": 2440
    },
    {
      "epoch": 0.055754045012857564,
      "grad_norm": 132.0,
      "learning_rate": 1.9270852684376744e-05,
      "loss": 2.1528,
      "step": 2450
    },
    {
      "epoch": 0.05598161254352229,
      "grad_norm": 118.5,
      "learning_rate": 1.926620843395876e-05,
      "loss": 0.8023,
      "step": 2460
    },
    {
      "epoch": 0.056209180074187015,
      "grad_norm": 44.5,
      "learning_rate": 1.9261564183540778e-05,
      "loss": 0.9334,
      "step": 2470
    },
    {
      "epoch": 0.05643674760485174,
      "grad_norm": 16.25,
      "learning_rate": 1.9256919933122795e-05,
      "loss": 1.6425,
      "step": 2480
    },
    {
      "epoch": 0.056664315135516466,
      "grad_norm": 89.0,
      "learning_rate": 1.9252275682704812e-05,
      "loss": 0.944,
      "step": 2490
    },
    {
      "epoch": 0.05689188266618119,
      "grad_norm": 73.5,
      "learning_rate": 1.9247631432286832e-05,
      "loss": 1.9675,
      "step": 2500
    },
    {
      "epoch": 0.05711945019684592,
      "grad_norm": 136.0,
      "learning_rate": 1.924298718186885e-05,
      "loss": 1.5524,
      "step": 2510
    },
    {
      "epoch": 0.05734701772751064,
      "grad_norm": 2.984375,
      "learning_rate": 1.9238342931450866e-05,
      "loss": 1.8072,
      "step": 2520
    },
    {
      "epoch": 0.05757458525817536,
      "grad_norm": 1.390625,
      "learning_rate": 1.9233698681032883e-05,
      "loss": 1.3017,
      "step": 2530
    },
    {
      "epoch": 0.057802152788840086,
      "grad_norm": 0.46484375,
      "learning_rate": 1.92290544306149e-05,
      "loss": 1.1158,
      "step": 2540
    },
    {
      "epoch": 0.05802972031950481,
      "grad_norm": 127.0,
      "learning_rate": 1.9224410180196918e-05,
      "loss": 1.4952,
      "step": 2550
    },
    {
      "epoch": 0.05825728785016954,
      "grad_norm": 3.078125,
      "learning_rate": 1.9219765929778935e-05,
      "loss": 1.5725,
      "step": 2560
    },
    {
      "epoch": 0.05848485538083426,
      "grad_norm": 0.9140625,
      "learning_rate": 1.921512167936095e-05,
      "loss": 0.8994,
      "step": 2570
    },
    {
      "epoch": 0.05871242291149899,
      "grad_norm": 98.0,
      "learning_rate": 1.9210477428942972e-05,
      "loss": 1.2764,
      "step": 2580
    },
    {
      "epoch": 0.05893999044216371,
      "grad_norm": 66.0,
      "learning_rate": 1.920583317852499e-05,
      "loss": 0.7414,
      "step": 2590
    },
    {
      "epoch": 0.05916755797282844,
      "grad_norm": 1.5703125,
      "learning_rate": 1.9201188928107003e-05,
      "loss": 1.7539,
      "step": 2600
    },
    {
      "epoch": 0.059395125503493164,
      "grad_norm": 135.0,
      "learning_rate": 1.9196544677689023e-05,
      "loss": 1.6336,
      "step": 2610
    },
    {
      "epoch": 0.05962269303415789,
      "grad_norm": 246.0,
      "learning_rate": 1.919190042727104e-05,
      "loss": 1.028,
      "step": 2620
    },
    {
      "epoch": 0.05985026056482261,
      "grad_norm": 0.0069580078125,
      "learning_rate": 1.9187256176853057e-05,
      "loss": 0.5558,
      "step": 2630
    },
    {
      "epoch": 0.06007782809548733,
      "grad_norm": 100.0,
      "learning_rate": 1.9182611926435074e-05,
      "loss": 1.0111,
      "step": 2640
    },
    {
      "epoch": 0.06030539562615206,
      "grad_norm": 26.0,
      "learning_rate": 1.917796767601709e-05,
      "loss": 0.4442,
      "step": 2650
    },
    {
      "epoch": 0.060532963156816784,
      "grad_norm": 147.0,
      "learning_rate": 1.9173323425599112e-05,
      "loss": 0.4586,
      "step": 2660
    },
    {
      "epoch": 0.06076053068748151,
      "grad_norm": 100.5,
      "learning_rate": 1.9168679175181126e-05,
      "loss": 0.4942,
      "step": 2670
    },
    {
      "epoch": 0.060988098218146235,
      "grad_norm": 247.0,
      "learning_rate": 1.9164034924763143e-05,
      "loss": 2.1476,
      "step": 2680
    },
    {
      "epoch": 0.06121566574881096,
      "grad_norm": 103.5,
      "learning_rate": 1.9159390674345163e-05,
      "loss": 1.1642,
      "step": 2690
    },
    {
      "epoch": 0.061443233279475686,
      "grad_norm": 740.0,
      "learning_rate": 1.915474642392718e-05,
      "loss": 1.7681,
      "step": 2700
    },
    {
      "epoch": 0.06167080081014041,
      "grad_norm": 0.000766754150390625,
      "learning_rate": 1.9150102173509197e-05,
      "loss": 1.0278,
      "step": 2710
    },
    {
      "epoch": 0.061898368340805136,
      "grad_norm": 0.034912109375,
      "learning_rate": 1.9145457923091214e-05,
      "loss": 0.5248,
      "step": 2720
    },
    {
      "epoch": 0.06212593587146986,
      "grad_norm": 210.0,
      "learning_rate": 1.914081367267323e-05,
      "loss": 0.9303,
      "step": 2730
    },
    {
      "epoch": 0.06235350340213458,
      "grad_norm": 164.0,
      "learning_rate": 1.913616942225525e-05,
      "loss": 1.0463,
      "step": 2740
    },
    {
      "epoch": 0.06258107093279931,
      "grad_norm": 21.0,
      "learning_rate": 1.9131525171837265e-05,
      "loss": 0.6941,
      "step": 2750
    },
    {
      "epoch": 0.06280863846346403,
      "grad_norm": 2.96875,
      "learning_rate": 1.9126880921419283e-05,
      "loss": 1.2098,
      "step": 2760
    },
    {
      "epoch": 0.06303620599412876,
      "grad_norm": 3.203125,
      "learning_rate": 1.9122236671001303e-05,
      "loss": 1.9144,
      "step": 2770
    },
    {
      "epoch": 0.06326377352479348,
      "grad_norm": 104.5,
      "learning_rate": 1.911759242058332e-05,
      "loss": 0.7588,
      "step": 2780
    },
    {
      "epoch": 0.06349134105545821,
      "grad_norm": 92.0,
      "learning_rate": 1.9112948170165337e-05,
      "loss": 1.9258,
      "step": 2790
    },
    {
      "epoch": 0.06371890858612293,
      "grad_norm": 0.06494140625,
      "learning_rate": 1.9108303919747354e-05,
      "loss": 0.3629,
      "step": 2800
    },
    {
      "epoch": 0.06394647611678765,
      "grad_norm": 0.01263427734375,
      "learning_rate": 1.910365966932937e-05,
      "loss": 0.5499,
      "step": 2810
    },
    {
      "epoch": 0.06417404364745238,
      "grad_norm": 0.06494140625,
      "learning_rate": 1.9099015418911388e-05,
      "loss": 0.6557,
      "step": 2820
    },
    {
      "epoch": 0.0644016111781171,
      "grad_norm": 0.07177734375,
      "learning_rate": 1.9094371168493405e-05,
      "loss": 1.2387,
      "step": 2830
    },
    {
      "epoch": 0.06462917870878183,
      "grad_norm": 840.0,
      "learning_rate": 1.9089726918075426e-05,
      "loss": 1.9772,
      "step": 2840
    },
    {
      "epoch": 0.06485674623944655,
      "grad_norm": 392.0,
      "learning_rate": 1.9085082667657443e-05,
      "loss": 2.0302,
      "step": 2850
    },
    {
      "epoch": 0.06508431377011129,
      "grad_norm": 122.0,
      "learning_rate": 1.908043841723946e-05,
      "loss": 0.9849,
      "step": 2860
    },
    {
      "epoch": 0.065311881300776,
      "grad_norm": 6.46875,
      "learning_rate": 1.9075794166821477e-05,
      "loss": 0.6327,
      "step": 2870
    },
    {
      "epoch": 0.06553944883144074,
      "grad_norm": 0.29296875,
      "learning_rate": 1.9071149916403494e-05,
      "loss": 0.2366,
      "step": 2880
    },
    {
      "epoch": 0.06576701636210545,
      "grad_norm": 0.000507354736328125,
      "learning_rate": 1.906650566598551e-05,
      "loss": 1.0514,
      "step": 2890
    },
    {
      "epoch": 0.06599458389277017,
      "grad_norm": 200.0,
      "learning_rate": 1.9061861415567528e-05,
      "loss": 0.6058,
      "step": 2900
    },
    {
      "epoch": 0.0662221514234349,
      "grad_norm": 207.0,
      "learning_rate": 1.9057217165149545e-05,
      "loss": 0.8723,
      "step": 2910
    },
    {
      "epoch": 0.06644971895409962,
      "grad_norm": 286.0,
      "learning_rate": 1.9052572914731566e-05,
      "loss": 0.734,
      "step": 2920
    },
    {
      "epoch": 0.06667728648476436,
      "grad_norm": 163.0,
      "learning_rate": 1.9047928664313583e-05,
      "loss": 2.5342,
      "step": 2930
    },
    {
      "epoch": 0.06690485401542907,
      "grad_norm": 340.0,
      "learning_rate": 1.90432844138956e-05,
      "loss": 1.2687,
      "step": 2940
    },
    {
      "epoch": 0.0671324215460938,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9038640163477617e-05,
      "loss": 0.5862,
      "step": 2950
    },
    {
      "epoch": 0.06735998907675853,
      "grad_norm": 101.0,
      "learning_rate": 1.9033995913059634e-05,
      "loss": 0.9355,
      "step": 2960
    },
    {
      "epoch": 0.06758755660742326,
      "grad_norm": 116.0,
      "learning_rate": 1.902935166264165e-05,
      "loss": 1.6136,
      "step": 2970
    },
    {
      "epoch": 0.06781512413808798,
      "grad_norm": 132.0,
      "learning_rate": 1.9024707412223668e-05,
      "loss": 0.8792,
      "step": 2980
    },
    {
      "epoch": 0.06804269166875271,
      "grad_norm": 0.0028533935546875,
      "learning_rate": 1.9020063161805685e-05,
      "loss": 1.4523,
      "step": 2990
    },
    {
      "epoch": 0.06827025919941743,
      "grad_norm": 145.0,
      "learning_rate": 1.9015418911387705e-05,
      "loss": 0.8868,
      "step": 3000
    },
    {
      "epoch": 0.06849782673008215,
      "grad_norm": 51.5,
      "learning_rate": 1.9010774660969722e-05,
      "loss": 0.7803,
      "step": 3010
    },
    {
      "epoch": 0.06872539426074688,
      "grad_norm": 63.25,
      "learning_rate": 1.9006130410551736e-05,
      "loss": 0.9399,
      "step": 3020
    },
    {
      "epoch": 0.0689529617914116,
      "grad_norm": 0.0634765625,
      "learning_rate": 1.9001486160133757e-05,
      "loss": 0.6319,
      "step": 3030
    },
    {
      "epoch": 0.06918052932207633,
      "grad_norm": 282.0,
      "learning_rate": 1.8996841909715774e-05,
      "loss": 1.5013,
      "step": 3040
    },
    {
      "epoch": 0.06940809685274105,
      "grad_norm": 66.5,
      "learning_rate": 1.899219765929779e-05,
      "loss": 0.6958,
      "step": 3050
    },
    {
      "epoch": 0.06963566438340578,
      "grad_norm": 1.4453125,
      "learning_rate": 1.8987553408879808e-05,
      "loss": 1.2576,
      "step": 3060
    },
    {
      "epoch": 0.0698632319140705,
      "grad_norm": 31.75,
      "learning_rate": 1.8982909158461825e-05,
      "loss": 0.9072,
      "step": 3070
    },
    {
      "epoch": 0.07009079944473523,
      "grad_norm": 0.498046875,
      "learning_rate": 1.8978264908043845e-05,
      "loss": 0.4992,
      "step": 3080
    },
    {
      "epoch": 0.07031836697539995,
      "grad_norm": 0.0003662109375,
      "learning_rate": 1.897362065762586e-05,
      "loss": 0.975,
      "step": 3090
    },
    {
      "epoch": 0.07054593450606468,
      "grad_norm": 19.375,
      "learning_rate": 1.8968976407207876e-05,
      "loss": 0.3622,
      "step": 3100
    },
    {
      "epoch": 0.0707735020367294,
      "grad_norm": 1.0703125,
      "learning_rate": 1.8964332156789896e-05,
      "loss": 1.5332,
      "step": 3110
    },
    {
      "epoch": 0.07100106956739412,
      "grad_norm": 158.0,
      "learning_rate": 1.8959687906371913e-05,
      "loss": 0.8343,
      "step": 3120
    },
    {
      "epoch": 0.07122863709805885,
      "grad_norm": 89.0,
      "learning_rate": 1.895504365595393e-05,
      "loss": 1.2094,
      "step": 3130
    },
    {
      "epoch": 0.07145620462872357,
      "grad_norm": 0.00084686279296875,
      "learning_rate": 1.8950399405535948e-05,
      "loss": 0.3673,
      "step": 3140
    },
    {
      "epoch": 0.0716837721593883,
      "grad_norm": 157.0,
      "learning_rate": 1.8945755155117968e-05,
      "loss": 0.5675,
      "step": 3150
    },
    {
      "epoch": 0.07191133969005302,
      "grad_norm": 0.08349609375,
      "learning_rate": 1.894111090469998e-05,
      "loss": 1.7348,
      "step": 3160
    },
    {
      "epoch": 0.07213890722071775,
      "grad_norm": 224.0,
      "learning_rate": 1.8936466654282e-05,
      "loss": 1.0345,
      "step": 3170
    },
    {
      "epoch": 0.07236647475138247,
      "grad_norm": 324.0,
      "learning_rate": 1.893182240386402e-05,
      "loss": 1.0265,
      "step": 3180
    },
    {
      "epoch": 0.0725940422820472,
      "grad_norm": 0.498046875,
      "learning_rate": 1.8927178153446036e-05,
      "loss": 1.011,
      "step": 3190
    },
    {
      "epoch": 0.07282160981271192,
      "grad_norm": 5.625,
      "learning_rate": 1.8922533903028053e-05,
      "loss": 0.8601,
      "step": 3200
    },
    {
      "epoch": 0.07304917734337665,
      "grad_norm": 0.0003604888916015625,
      "learning_rate": 1.891788965261007e-05,
      "loss": 0.9518,
      "step": 3210
    },
    {
      "epoch": 0.07327674487404137,
      "grad_norm": 0.859375,
      "learning_rate": 1.8913245402192087e-05,
      "loss": 0.5403,
      "step": 3220
    },
    {
      "epoch": 0.07350431240470609,
      "grad_norm": 97.5,
      "learning_rate": 1.8908601151774104e-05,
      "loss": 1.6403,
      "step": 3230
    },
    {
      "epoch": 0.07373187993537082,
      "grad_norm": 171.0,
      "learning_rate": 1.890395690135612e-05,
      "loss": 0.6887,
      "step": 3240
    },
    {
      "epoch": 0.07395944746603554,
      "grad_norm": 115.5,
      "learning_rate": 1.889931265093814e-05,
      "loss": 1.3209,
      "step": 3250
    },
    {
      "epoch": 0.07418701499670027,
      "grad_norm": 236.0,
      "learning_rate": 1.889466840052016e-05,
      "loss": 0.9023,
      "step": 3260
    },
    {
      "epoch": 0.07441458252736499,
      "grad_norm": 127.5,
      "learning_rate": 1.8890024150102176e-05,
      "loss": 0.8366,
      "step": 3270
    },
    {
      "epoch": 0.07464215005802972,
      "grad_norm": 251.0,
      "learning_rate": 1.8885379899684193e-05,
      "loss": 1.1234,
      "step": 3280
    },
    {
      "epoch": 0.07486971758869444,
      "grad_norm": 37.0,
      "learning_rate": 1.888073564926621e-05,
      "loss": 0.4003,
      "step": 3290
    },
    {
      "epoch": 0.07509728511935918,
      "grad_norm": 15.9375,
      "learning_rate": 1.8876091398848227e-05,
      "loss": 0.5239,
      "step": 3300
    },
    {
      "epoch": 0.0753248526500239,
      "grad_norm": 1.2265625,
      "learning_rate": 1.8871447148430244e-05,
      "loss": 0.9845,
      "step": 3310
    },
    {
      "epoch": 0.07555242018068863,
      "grad_norm": 74.5,
      "learning_rate": 1.886680289801226e-05,
      "loss": 0.9918,
      "step": 3320
    },
    {
      "epoch": 0.07577998771135334,
      "grad_norm": 15.375,
      "learning_rate": 1.886215864759428e-05,
      "loss": 1.9841,
      "step": 3330
    },
    {
      "epoch": 0.07600755524201806,
      "grad_norm": 11.4375,
      "learning_rate": 1.88575143971763e-05,
      "loss": 0.987,
      "step": 3340
    },
    {
      "epoch": 0.0762351227726828,
      "grad_norm": 181.0,
      "learning_rate": 1.8852870146758316e-05,
      "loss": 1.1012,
      "step": 3350
    },
    {
      "epoch": 0.07646269030334751,
      "grad_norm": 540.0,
      "learning_rate": 1.8848225896340333e-05,
      "loss": 1.1926,
      "step": 3360
    },
    {
      "epoch": 0.07669025783401225,
      "grad_norm": 79.0,
      "learning_rate": 1.884358164592235e-05,
      "loss": 0.4642,
      "step": 3370
    },
    {
      "epoch": 0.07691782536467696,
      "grad_norm": 266.0,
      "learning_rate": 1.8838937395504367e-05,
      "loss": 1.8232,
      "step": 3380
    },
    {
      "epoch": 0.0771453928953417,
      "grad_norm": 240.0,
      "learning_rate": 1.8834293145086384e-05,
      "loss": 2.7924,
      "step": 3390
    },
    {
      "epoch": 0.07737296042600642,
      "grad_norm": 38.75,
      "learning_rate": 1.88296488946684e-05,
      "loss": 0.8273,
      "step": 3400
    },
    {
      "epoch": 0.07760052795667115,
      "grad_norm": 80.0,
      "learning_rate": 1.8825004644250418e-05,
      "loss": 0.7285,
      "step": 3410
    },
    {
      "epoch": 0.07782809548733587,
      "grad_norm": 236.0,
      "learning_rate": 1.882036039383244e-05,
      "loss": 1.482,
      "step": 3420
    },
    {
      "epoch": 0.0780556630180006,
      "grad_norm": 274.0,
      "learning_rate": 1.8815716143414456e-05,
      "loss": 0.988,
      "step": 3430
    },
    {
      "epoch": 0.07828323054866532,
      "grad_norm": 8.8125,
      "learning_rate": 1.881107189299647e-05,
      "loss": 0.9761,
      "step": 3440
    },
    {
      "epoch": 0.07851079807933004,
      "grad_norm": 0.55859375,
      "learning_rate": 1.880642764257849e-05,
      "loss": 0.5996,
      "step": 3450
    },
    {
      "epoch": 0.07873836560999477,
      "grad_norm": 4.28125,
      "learning_rate": 1.8801783392160507e-05,
      "loss": 0.3362,
      "step": 3460
    },
    {
      "epoch": 0.07896593314065949,
      "grad_norm": 97.0,
      "learning_rate": 1.8797139141742524e-05,
      "loss": 0.9861,
      "step": 3470
    },
    {
      "epoch": 0.07919350067132422,
      "grad_norm": 57.75,
      "learning_rate": 1.879249489132454e-05,
      "loss": 0.6532,
      "step": 3480
    },
    {
      "epoch": 0.07942106820198894,
      "grad_norm": 127.5,
      "learning_rate": 1.878785064090656e-05,
      "loss": 1.1781,
      "step": 3490
    },
    {
      "epoch": 0.07964863573265367,
      "grad_norm": 432.0,
      "learning_rate": 1.878320639048858e-05,
      "loss": 0.8597,
      "step": 3500
    },
    {
      "epoch": 0.07987620326331839,
      "grad_norm": 121.5,
      "learning_rate": 1.8778562140070592e-05,
      "loss": 1.7989,
      "step": 3510
    },
    {
      "epoch": 0.08010377079398312,
      "grad_norm": 102.0,
      "learning_rate": 1.877391788965261e-05,
      "loss": 0.4543,
      "step": 3520
    },
    {
      "epoch": 0.08033133832464784,
      "grad_norm": 6.0625,
      "learning_rate": 1.876927363923463e-05,
      "loss": 1.2402,
      "step": 3530
    },
    {
      "epoch": 0.08055890585531257,
      "grad_norm": 149.0,
      "learning_rate": 1.8764629388816647e-05,
      "loss": 1.0314,
      "step": 3540
    },
    {
      "epoch": 0.08078647338597729,
      "grad_norm": 0.125,
      "learning_rate": 1.8759985138398664e-05,
      "loss": 0.7963,
      "step": 3550
    },
    {
      "epoch": 0.08101404091664201,
      "grad_norm": 6.8125,
      "learning_rate": 1.875534088798068e-05,
      "loss": 1.9317,
      "step": 3560
    },
    {
      "epoch": 0.08124160844730674,
      "grad_norm": 0.302734375,
      "learning_rate": 1.87506966375627e-05,
      "loss": 0.397,
      "step": 3570
    },
    {
      "epoch": 0.08146917597797146,
      "grad_norm": 173.0,
      "learning_rate": 1.8746052387144715e-05,
      "loss": 1.0142,
      "step": 3580
    },
    {
      "epoch": 0.08169674350863619,
      "grad_norm": 59.25,
      "learning_rate": 1.8741408136726732e-05,
      "loss": 0.4263,
      "step": 3590
    },
    {
      "epoch": 0.08192431103930091,
      "grad_norm": 0.00128173828125,
      "learning_rate": 1.8736763886308752e-05,
      "loss": 1.2497,
      "step": 3600
    },
    {
      "epoch": 0.08215187856996564,
      "grad_norm": 43.75,
      "learning_rate": 1.873211963589077e-05,
      "loss": 0.3371,
      "step": 3610
    },
    {
      "epoch": 0.08237944610063036,
      "grad_norm": 278.0,
      "learning_rate": 1.8727475385472786e-05,
      "loss": 1.0487,
      "step": 3620
    },
    {
      "epoch": 0.08260701363129509,
      "grad_norm": 72.5,
      "learning_rate": 1.8722831135054804e-05,
      "loss": 1.8185,
      "step": 3630
    },
    {
      "epoch": 0.08283458116195981,
      "grad_norm": 207.0,
      "learning_rate": 1.871818688463682e-05,
      "loss": 1.0284,
      "step": 3640
    },
    {
      "epoch": 0.08306214869262453,
      "grad_norm": 0.01336669921875,
      "learning_rate": 1.8713542634218838e-05,
      "loss": 1.8483,
      "step": 3650
    },
    {
      "epoch": 0.08328971622328926,
      "grad_norm": 154.0,
      "learning_rate": 1.8708898383800855e-05,
      "loss": 1.2936,
      "step": 3660
    },
    {
      "epoch": 0.08351728375395398,
      "grad_norm": 31.5,
      "learning_rate": 1.8704254133382872e-05,
      "loss": 0.6827,
      "step": 3670
    },
    {
      "epoch": 0.08374485128461871,
      "grad_norm": 0.0028839111328125,
      "learning_rate": 1.8699609882964892e-05,
      "loss": 1.1802,
      "step": 3680
    },
    {
      "epoch": 0.08397241881528343,
      "grad_norm": 216.0,
      "learning_rate": 1.869496563254691e-05,
      "loss": 0.9687,
      "step": 3690
    },
    {
      "epoch": 0.08419998634594816,
      "grad_norm": 0.275390625,
      "learning_rate": 1.8690321382128926e-05,
      "loss": 0.2817,
      "step": 3700
    },
    {
      "epoch": 0.08442755387661288,
      "grad_norm": 0.025146484375,
      "learning_rate": 1.8685677131710943e-05,
      "loss": 0.8772,
      "step": 3710
    },
    {
      "epoch": 0.08465512140727761,
      "grad_norm": 0.0859375,
      "learning_rate": 1.868103288129296e-05,
      "loss": 1.1622,
      "step": 3720
    },
    {
      "epoch": 0.08488268893794233,
      "grad_norm": 0.034423828125,
      "learning_rate": 1.8676388630874977e-05,
      "loss": 0.6207,
      "step": 3730
    },
    {
      "epoch": 0.08511025646860707,
      "grad_norm": 174.0,
      "learning_rate": 1.8671744380456995e-05,
      "loss": 1.2877,
      "step": 3740
    },
    {
      "epoch": 0.08533782399927178,
      "grad_norm": 234.0,
      "learning_rate": 1.866710013003901e-05,
      "loss": 0.8022,
      "step": 3750
    },
    {
      "epoch": 0.0855653915299365,
      "grad_norm": 77.5,
      "learning_rate": 1.8662455879621032e-05,
      "loss": 1.1684,
      "step": 3760
    },
    {
      "epoch": 0.08579295906060123,
      "grad_norm": 2.03125,
      "learning_rate": 1.865781162920305e-05,
      "loss": 0.6528,
      "step": 3770
    },
    {
      "epoch": 0.08602052659126595,
      "grad_norm": 506.0,
      "learning_rate": 1.8653167378785066e-05,
      "loss": 0.8789,
      "step": 3780
    },
    {
      "epoch": 0.08624809412193069,
      "grad_norm": 0.50390625,
      "learning_rate": 1.8648523128367083e-05,
      "loss": 0.1293,
      "step": 3790
    },
    {
      "epoch": 0.0864756616525954,
      "grad_norm": 0.0002651214599609375,
      "learning_rate": 1.86438788779491e-05,
      "loss": 0.8254,
      "step": 3800
    },
    {
      "epoch": 0.08670322918326014,
      "grad_norm": 217.0,
      "learning_rate": 1.8639234627531117e-05,
      "loss": 1.1668,
      "step": 3810
    },
    {
      "epoch": 0.08693079671392485,
      "grad_norm": 0.00701904296875,
      "learning_rate": 1.8634590377113134e-05,
      "loss": 0.6171,
      "step": 3820
    },
    {
      "epoch": 0.08715836424458959,
      "grad_norm": 109.0,
      "learning_rate": 1.8629946126695155e-05,
      "loss": 0.347,
      "step": 3830
    },
    {
      "epoch": 0.0873859317752543,
      "grad_norm": 0.1259765625,
      "learning_rate": 1.8625301876277172e-05,
      "loss": 0.9697,
      "step": 3840
    },
    {
      "epoch": 0.08761349930591904,
      "grad_norm": 108.0,
      "learning_rate": 1.862065762585919e-05,
      "loss": 1.4774,
      "step": 3850
    },
    {
      "epoch": 0.08784106683658376,
      "grad_norm": 70.0,
      "learning_rate": 1.8616013375441203e-05,
      "loss": 0.6104,
      "step": 3860
    },
    {
      "epoch": 0.08806863436724847,
      "grad_norm": 0.00185394287109375,
      "learning_rate": 1.8611369125023223e-05,
      "loss": 0.5244,
      "step": 3870
    },
    {
      "epoch": 0.0882962018979132,
      "grad_norm": 100.5,
      "learning_rate": 1.860672487460524e-05,
      "loss": 0.5807,
      "step": 3880
    },
    {
      "epoch": 0.08852376942857793,
      "grad_norm": 115.5,
      "learning_rate": 1.8602080624187257e-05,
      "loss": 1.3363,
      "step": 3890
    },
    {
      "epoch": 0.08875133695924266,
      "grad_norm": 976.0,
      "learning_rate": 1.8597436373769274e-05,
      "loss": 2.4015,
      "step": 3900
    },
    {
      "epoch": 0.08897890448990738,
      "grad_norm": 0.5078125,
      "learning_rate": 1.8592792123351295e-05,
      "loss": 0.8609,
      "step": 3910
    },
    {
      "epoch": 0.08920647202057211,
      "grad_norm": 0.02294921875,
      "learning_rate": 1.8588147872933312e-05,
      "loss": 1.0725,
      "step": 3920
    },
    {
      "epoch": 0.08943403955123683,
      "grad_norm": 0.00072479248046875,
      "learning_rate": 1.8583503622515325e-05,
      "loss": 1.5132,
      "step": 3930
    },
    {
      "epoch": 0.08966160708190156,
      "grad_norm": 111.0,
      "learning_rate": 1.8578859372097346e-05,
      "loss": 0.6439,
      "step": 3940
    },
    {
      "epoch": 0.08988917461256628,
      "grad_norm": 0.0302734375,
      "learning_rate": 1.8574215121679363e-05,
      "loss": 0.8196,
      "step": 3950
    },
    {
      "epoch": 0.09011674214323101,
      "grad_norm": 37.0,
      "learning_rate": 1.856957087126138e-05,
      "loss": 0.4857,
      "step": 3960
    },
    {
      "epoch": 0.09034430967389573,
      "grad_norm": 95.5,
      "learning_rate": 1.8564926620843397e-05,
      "loss": 0.6249,
      "step": 3970
    },
    {
      "epoch": 0.09057187720456045,
      "grad_norm": 48.75,
      "learning_rate": 1.8560282370425414e-05,
      "loss": 1.6335,
      "step": 3980
    },
    {
      "epoch": 0.09079944473522518,
      "grad_norm": 0.035400390625,
      "learning_rate": 1.8555638120007434e-05,
      "loss": 1.3521,
      "step": 3990
    },
    {
      "epoch": 0.0910270122658899,
      "grad_norm": 0.00341796875,
      "learning_rate": 1.8550993869589448e-05,
      "loss": 0.581,
      "step": 4000
    },
    {
      "epoch": 0.09125457979655463,
      "grad_norm": 46.5,
      "learning_rate": 1.8546349619171465e-05,
      "loss": 1.0666,
      "step": 4010
    },
    {
      "epoch": 0.09148214732721935,
      "grad_norm": 452.0,
      "learning_rate": 1.8541705368753486e-05,
      "loss": 0.8431,
      "step": 4020
    },
    {
      "epoch": 0.09170971485788408,
      "grad_norm": 97.0,
      "learning_rate": 1.8537061118335503e-05,
      "loss": 1.2089,
      "step": 4030
    },
    {
      "epoch": 0.0919372823885488,
      "grad_norm": 0.0067138671875,
      "learning_rate": 1.853241686791752e-05,
      "loss": 0.8273,
      "step": 4040
    },
    {
      "epoch": 0.09216484991921353,
      "grad_norm": 228.0,
      "learning_rate": 1.8527772617499537e-05,
      "loss": 1.6389,
      "step": 4050
    },
    {
      "epoch": 0.09239241744987825,
      "grad_norm": 115.0,
      "learning_rate": 1.8523128367081554e-05,
      "loss": 1.0782,
      "step": 4060
    },
    {
      "epoch": 0.09261998498054298,
      "grad_norm": 310.0,
      "learning_rate": 1.851848411666357e-05,
      "loss": 1.6441,
      "step": 4070
    },
    {
      "epoch": 0.0928475525112077,
      "grad_norm": 296.0,
      "learning_rate": 1.8513839866245588e-05,
      "loss": 0.9944,
      "step": 4080
    },
    {
      "epoch": 0.09307512004187242,
      "grad_norm": 7.0625,
      "learning_rate": 1.8509195615827605e-05,
      "loss": 0.9066,
      "step": 4090
    },
    {
      "epoch": 0.09330268757253715,
      "grad_norm": 158.0,
      "learning_rate": 1.8504551365409625e-05,
      "loss": 1.0299,
      "step": 4100
    },
    {
      "epoch": 0.09353025510320187,
      "grad_norm": 4.78125,
      "learning_rate": 1.8499907114991643e-05,
      "loss": 1.0905,
      "step": 4110
    },
    {
      "epoch": 0.0937578226338666,
      "grad_norm": 136.0,
      "learning_rate": 1.849526286457366e-05,
      "loss": 1.3911,
      "step": 4120
    },
    {
      "epoch": 0.09398539016453132,
      "grad_norm": 0.00775146484375,
      "learning_rate": 1.8490618614155677e-05,
      "loss": 0.7027,
      "step": 4130
    },
    {
      "epoch": 0.09421295769519605,
      "grad_norm": 0.00384521484375,
      "learning_rate": 1.8485974363737694e-05,
      "loss": 1.0323,
      "step": 4140
    },
    {
      "epoch": 0.09444052522586077,
      "grad_norm": 95.0,
      "learning_rate": 1.848133011331971e-05,
      "loss": 0.403,
      "step": 4150
    },
    {
      "epoch": 0.0946680927565255,
      "grad_norm": 0.00714111328125,
      "learning_rate": 1.8476685862901728e-05,
      "loss": 0.6832,
      "step": 4160
    },
    {
      "epoch": 0.09489566028719022,
      "grad_norm": 0.00115966796875,
      "learning_rate": 1.8472041612483745e-05,
      "loss": 0.7143,
      "step": 4170
    },
    {
      "epoch": 0.09512322781785496,
      "grad_norm": 175.0,
      "learning_rate": 1.8467397362065765e-05,
      "loss": 0.7458,
      "step": 4180
    },
    {
      "epoch": 0.09535079534851967,
      "grad_norm": 47.75,
      "learning_rate": 1.8462753111647782e-05,
      "loss": 0.8852,
      "step": 4190
    },
    {
      "epoch": 0.09557836287918439,
      "grad_norm": 688.0,
      "learning_rate": 1.84581088612298e-05,
      "loss": 1.6918,
      "step": 4200
    },
    {
      "epoch": 0.09580593040984912,
      "grad_norm": 230.0,
      "learning_rate": 1.8453464610811816e-05,
      "loss": 1.1986,
      "step": 4210
    },
    {
      "epoch": 0.09603349794051384,
      "grad_norm": 0.01336669921875,
      "learning_rate": 1.8448820360393834e-05,
      "loss": 1.2436,
      "step": 4220
    },
    {
      "epoch": 0.09626106547117858,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.844417610997585e-05,
      "loss": 0.4942,
      "step": 4230
    },
    {
      "epoch": 0.0964886330018433,
      "grad_norm": 2.4375,
      "learning_rate": 1.8439531859557868e-05,
      "loss": 0.8558,
      "step": 4240
    },
    {
      "epoch": 0.09671620053250803,
      "grad_norm": 256.0,
      "learning_rate": 1.8434887609139888e-05,
      "loss": 2.4483,
      "step": 4250
    },
    {
      "epoch": 0.09694376806317274,
      "grad_norm": 161.0,
      "learning_rate": 1.8430243358721905e-05,
      "loss": 1.2884,
      "step": 4260
    },
    {
      "epoch": 0.09717133559383748,
      "grad_norm": 0.56640625,
      "learning_rate": 1.8425599108303922e-05,
      "loss": 1.0118,
      "step": 4270
    },
    {
      "epoch": 0.0973989031245022,
      "grad_norm": 0.0252685546875,
      "learning_rate": 1.842095485788594e-05,
      "loss": 1.5501,
      "step": 4280
    },
    {
      "epoch": 0.09762647065516693,
      "grad_norm": 168.0,
      "learning_rate": 1.8416310607467956e-05,
      "loss": 0.9357,
      "step": 4290
    },
    {
      "epoch": 0.09785403818583165,
      "grad_norm": 3.71875,
      "learning_rate": 1.8411666357049973e-05,
      "loss": 0.7578,
      "step": 4300
    },
    {
      "epoch": 0.09808160571649636,
      "grad_norm": 113.0,
      "learning_rate": 1.840702210663199e-05,
      "loss": 1.304,
      "step": 4310
    },
    {
      "epoch": 0.0983091732471611,
      "grad_norm": 34.5,
      "learning_rate": 1.8402377856214007e-05,
      "loss": 0.5295,
      "step": 4320
    },
    {
      "epoch": 0.09853674077782582,
      "grad_norm": 528.0,
      "learning_rate": 1.8397733605796028e-05,
      "loss": 1.0567,
      "step": 4330
    },
    {
      "epoch": 0.09876430830849055,
      "grad_norm": 60.25,
      "learning_rate": 1.8393089355378045e-05,
      "loss": 2.4355,
      "step": 4340
    },
    {
      "epoch": 0.09899187583915527,
      "grad_norm": 184.0,
      "learning_rate": 1.838844510496006e-05,
      "loss": 0.9955,
      "step": 4350
    },
    {
      "epoch": 0.09921944336982,
      "grad_norm": 187.0,
      "learning_rate": 1.838380085454208e-05,
      "loss": 0.6422,
      "step": 4360
    },
    {
      "epoch": 0.09944701090048472,
      "grad_norm": 126.0,
      "learning_rate": 1.8379156604124096e-05,
      "loss": 0.964,
      "step": 4370
    },
    {
      "epoch": 0.09967457843114945,
      "grad_norm": 0.00162506103515625,
      "learning_rate": 1.8374512353706113e-05,
      "loss": 1.6056,
      "step": 4380
    },
    {
      "epoch": 0.09990214596181417,
      "grad_norm": 106.5,
      "learning_rate": 1.836986810328813e-05,
      "loss": 1.0443,
      "step": 4390
    },
    {
      "epoch": 0.10012971349247889,
      "grad_norm": 256.0,
      "learning_rate": 1.8365223852870147e-05,
      "loss": 0.9527,
      "step": 4400
    },
    {
      "epoch": 0.10035728102314362,
      "grad_norm": 378.0,
      "learning_rate": 1.8360579602452168e-05,
      "loss": 1.4318,
      "step": 4410
    },
    {
      "epoch": 0.10058484855380834,
      "grad_norm": 490.0,
      "learning_rate": 1.835593535203418e-05,
      "loss": 0.737,
      "step": 4420
    },
    {
      "epoch": 0.10081241608447307,
      "grad_norm": 12.8125,
      "learning_rate": 1.83512911016162e-05,
      "loss": 1.3707,
      "step": 4430
    },
    {
      "epoch": 0.10103998361513779,
      "grad_norm": 0.08984375,
      "learning_rate": 1.834664685119822e-05,
      "loss": 0.5255,
      "step": 4440
    },
    {
      "epoch": 0.10126755114580252,
      "grad_norm": 6.09375,
      "learning_rate": 1.8342002600780236e-05,
      "loss": 0.9958,
      "step": 4450
    },
    {
      "epoch": 0.10149511867646724,
      "grad_norm": 458.0,
      "learning_rate": 1.8337358350362253e-05,
      "loss": 1.8693,
      "step": 4460
    },
    {
      "epoch": 0.10172268620713197,
      "grad_norm": 229.0,
      "learning_rate": 1.833271409994427e-05,
      "loss": 1.1017,
      "step": 4470
    },
    {
      "epoch": 0.10195025373779669,
      "grad_norm": 160.0,
      "learning_rate": 1.8328069849526287e-05,
      "loss": 0.5105,
      "step": 4480
    },
    {
      "epoch": 0.10217782126846142,
      "grad_norm": 236.0,
      "learning_rate": 1.8323425599108304e-05,
      "loss": 0.8901,
      "step": 4490
    },
    {
      "epoch": 0.10240538879912614,
      "grad_norm": 136.0,
      "learning_rate": 1.831878134869032e-05,
      "loss": 1.1035,
      "step": 4500
    },
    {
      "epoch": 0.10263295632979086,
      "grad_norm": 0.00201416015625,
      "learning_rate": 1.8314137098272338e-05,
      "loss": 0.237,
      "step": 4510
    },
    {
      "epoch": 0.10286052386045559,
      "grad_norm": 0.80078125,
      "learning_rate": 1.830949284785436e-05,
      "loss": 0.6616,
      "step": 4520
    },
    {
      "epoch": 0.10308809139112031,
      "grad_norm": 204.0,
      "learning_rate": 1.8304848597436376e-05,
      "loss": 0.4937,
      "step": 4530
    },
    {
      "epoch": 0.10331565892178504,
      "grad_norm": 189.0,
      "learning_rate": 1.8300204347018393e-05,
      "loss": 0.7375,
      "step": 4540
    },
    {
      "epoch": 0.10354322645244976,
      "grad_norm": 161.0,
      "learning_rate": 1.829556009660041e-05,
      "loss": 1.5167,
      "step": 4550
    },
    {
      "epoch": 0.10377079398311449,
      "grad_norm": 102.0,
      "learning_rate": 1.8290915846182427e-05,
      "loss": 1.0837,
      "step": 4560
    },
    {
      "epoch": 0.10399836151377921,
      "grad_norm": 49.25,
      "learning_rate": 1.8286271595764444e-05,
      "loss": 1.2148,
      "step": 4570
    },
    {
      "epoch": 0.10422592904444394,
      "grad_norm": 84.5,
      "learning_rate": 1.828162734534646e-05,
      "loss": 0.9579,
      "step": 4580
    },
    {
      "epoch": 0.10445349657510866,
      "grad_norm": 0.0018768310546875,
      "learning_rate": 1.827698309492848e-05,
      "loss": 1.3927,
      "step": 4590
    },
    {
      "epoch": 0.1046810641057734,
      "grad_norm": 155.0,
      "learning_rate": 1.82723388445105e-05,
      "loss": 0.6643,
      "step": 4600
    },
    {
      "epoch": 0.10490863163643811,
      "grad_norm": 0.009033203125,
      "learning_rate": 1.8267694594092516e-05,
      "loss": 1.0951,
      "step": 4610
    },
    {
      "epoch": 0.10513619916710283,
      "grad_norm": 85.0,
      "learning_rate": 1.8263050343674533e-05,
      "loss": 0.7842,
      "step": 4620
    },
    {
      "epoch": 0.10536376669776756,
      "grad_norm": 8.6875,
      "learning_rate": 1.825840609325655e-05,
      "loss": 0.5471,
      "step": 4630
    },
    {
      "epoch": 0.10559133422843228,
      "grad_norm": 0.000812530517578125,
      "learning_rate": 1.8253761842838567e-05,
      "loss": 0.5135,
      "step": 4640
    },
    {
      "epoch": 0.10581890175909701,
      "grad_norm": 201.0,
      "learning_rate": 1.8249117592420584e-05,
      "loss": 0.7411,
      "step": 4650
    },
    {
      "epoch": 0.10604646928976173,
      "grad_norm": 418.0,
      "learning_rate": 1.82444733420026e-05,
      "loss": 1.7365,
      "step": 4660
    },
    {
      "epoch": 0.10627403682042647,
      "grad_norm": 0.000965118408203125,
      "learning_rate": 1.823982909158462e-05,
      "loss": 0.5488,
      "step": 4670
    },
    {
      "epoch": 0.10650160435109118,
      "grad_norm": 112.5,
      "learning_rate": 1.823518484116664e-05,
      "loss": 0.9816,
      "step": 4680
    },
    {
      "epoch": 0.10672917188175592,
      "grad_norm": 144.0,
      "learning_rate": 1.8230540590748655e-05,
      "loss": 1.6494,
      "step": 4690
    },
    {
      "epoch": 0.10695673941242063,
      "grad_norm": 121.0,
      "learning_rate": 1.8225896340330672e-05,
      "loss": 1.0814,
      "step": 4700
    },
    {
      "epoch": 0.10718430694308537,
      "grad_norm": 88.5,
      "learning_rate": 1.822125208991269e-05,
      "loss": 1.2509,
      "step": 4710
    },
    {
      "epoch": 0.10741187447375009,
      "grad_norm": 199.0,
      "learning_rate": 1.8216607839494707e-05,
      "loss": 1.3065,
      "step": 4720
    },
    {
      "epoch": 0.1076394420044148,
      "grad_norm": 110.5,
      "learning_rate": 1.8211963589076724e-05,
      "loss": 1.008,
      "step": 4730
    },
    {
      "epoch": 0.10786700953507954,
      "grad_norm": 98.0,
      "learning_rate": 1.820731933865874e-05,
      "loss": 0.9274,
      "step": 4740
    },
    {
      "epoch": 0.10809457706574425,
      "grad_norm": 256.0,
      "learning_rate": 1.820267508824076e-05,
      "loss": 1.8326,
      "step": 4750
    },
    {
      "epoch": 0.10832214459640899,
      "grad_norm": 394.0,
      "learning_rate": 1.8198030837822778e-05,
      "loss": 1.1789,
      "step": 4760
    },
    {
      "epoch": 0.1085497121270737,
      "grad_norm": 242.0,
      "learning_rate": 1.8193386587404795e-05,
      "loss": 0.8762,
      "step": 4770
    },
    {
      "epoch": 0.10877727965773844,
      "grad_norm": 201.0,
      "learning_rate": 1.8188742336986812e-05,
      "loss": 1.3885,
      "step": 4780
    },
    {
      "epoch": 0.10900484718840316,
      "grad_norm": 184.0,
      "learning_rate": 1.818409808656883e-05,
      "loss": 1.2246,
      "step": 4790
    },
    {
      "epoch": 0.10923241471906789,
      "grad_norm": 133.0,
      "learning_rate": 1.8179453836150846e-05,
      "loss": 1.7681,
      "step": 4800
    },
    {
      "epoch": 0.1094599822497326,
      "grad_norm": 104.0,
      "learning_rate": 1.8174809585732863e-05,
      "loss": 0.7568,
      "step": 4810
    },
    {
      "epoch": 0.10968754978039734,
      "grad_norm": 56.25,
      "learning_rate": 1.817016533531488e-05,
      "loss": 0.3507,
      "step": 4820
    },
    {
      "epoch": 0.10991511731106206,
      "grad_norm": 0.119140625,
      "learning_rate": 1.81655210848969e-05,
      "loss": 1.3329,
      "step": 4830
    },
    {
      "epoch": 0.11014268484172678,
      "grad_norm": 53.5,
      "learning_rate": 1.8160876834478918e-05,
      "loss": 0.6663,
      "step": 4840
    },
    {
      "epoch": 0.11037025237239151,
      "grad_norm": 19.75,
      "learning_rate": 1.815623258406093e-05,
      "loss": 2.3035,
      "step": 4850
    },
    {
      "epoch": 0.11059781990305623,
      "grad_norm": 0.0159912109375,
      "learning_rate": 1.8151588333642952e-05,
      "loss": 0.5893,
      "step": 4860
    },
    {
      "epoch": 0.11082538743372096,
      "grad_norm": 159.0,
      "learning_rate": 1.814694408322497e-05,
      "loss": 1.0684,
      "step": 4870
    },
    {
      "epoch": 0.11105295496438568,
      "grad_norm": 139.0,
      "learning_rate": 1.8142299832806986e-05,
      "loss": 1.0775,
      "step": 4880
    },
    {
      "epoch": 0.11128052249505041,
      "grad_norm": 108.0,
      "learning_rate": 1.8137655582389003e-05,
      "loss": 0.6041,
      "step": 4890
    },
    {
      "epoch": 0.11150809002571513,
      "grad_norm": 59.5,
      "learning_rate": 1.8133011331971024e-05,
      "loss": 0.4602,
      "step": 4900
    },
    {
      "epoch": 0.11173565755637986,
      "grad_norm": 27.25,
      "learning_rate": 1.812836708155304e-05,
      "loss": 0.623,
      "step": 4910
    },
    {
      "epoch": 0.11196322508704458,
      "grad_norm": 0.2333984375,
      "learning_rate": 1.8123722831135054e-05,
      "loss": 0.5325,
      "step": 4920
    },
    {
      "epoch": 0.11219079261770931,
      "grad_norm": 122.0,
      "learning_rate": 1.8119078580717075e-05,
      "loss": 1.1797,
      "step": 4930
    },
    {
      "epoch": 0.11241836014837403,
      "grad_norm": 106.0,
      "learning_rate": 1.8114434330299092e-05,
      "loss": 1.0768,
      "step": 4940
    },
    {
      "epoch": 0.11264592767903875,
      "grad_norm": 0.0010528564453125,
      "learning_rate": 1.810979007988111e-05,
      "loss": 0.6651,
      "step": 4950
    },
    {
      "epoch": 0.11287349520970348,
      "grad_norm": 0.0005340576171875,
      "learning_rate": 1.8105145829463126e-05,
      "loss": 1.1469,
      "step": 4960
    },
    {
      "epoch": 0.1131010627403682,
      "grad_norm": 17.625,
      "learning_rate": 1.8100501579045143e-05,
      "loss": 1.2118,
      "step": 4970
    },
    {
      "epoch": 0.11332863027103293,
      "grad_norm": 203.0,
      "learning_rate": 1.8095857328627164e-05,
      "loss": 0.7032,
      "step": 4980
    },
    {
      "epoch": 0.11355619780169765,
      "grad_norm": 0.07763671875,
      "learning_rate": 1.8091213078209177e-05,
      "loss": 0.9442,
      "step": 4990
    },
    {
      "epoch": 0.11378376533236238,
      "grad_norm": 120.0,
      "learning_rate": 1.8086568827791194e-05,
      "loss": 1.0119,
      "step": 5000
    },
    {
      "epoch": 0.1140113328630271,
      "grad_norm": 14.8125,
      "learning_rate": 1.8081924577373215e-05,
      "loss": 0.4565,
      "step": 5010
    },
    {
      "epoch": 0.11423890039369183,
      "grad_norm": 113.5,
      "learning_rate": 1.8077280326955232e-05,
      "loss": 1.6709,
      "step": 5020
    },
    {
      "epoch": 0.11446646792435655,
      "grad_norm": 133.0,
      "learning_rate": 1.807263607653725e-05,
      "loss": 1.5785,
      "step": 5030
    },
    {
      "epoch": 0.11469403545502128,
      "grad_norm": 91.5,
      "learning_rate": 1.8067991826119266e-05,
      "loss": 0.6575,
      "step": 5040
    },
    {
      "epoch": 0.114921602985686,
      "grad_norm": 195.0,
      "learning_rate": 1.8063347575701283e-05,
      "loss": 0.9251,
      "step": 5050
    },
    {
      "epoch": 0.11514917051635072,
      "grad_norm": 406.0,
      "learning_rate": 1.80587033252833e-05,
      "loss": 1.4432,
      "step": 5060
    },
    {
      "epoch": 0.11537673804701545,
      "grad_norm": 126.5,
      "learning_rate": 1.8054059074865317e-05,
      "loss": 0.572,
      "step": 5070
    },
    {
      "epoch": 0.11560430557768017,
      "grad_norm": 134.0,
      "learning_rate": 1.8049414824447334e-05,
      "loss": 0.4565,
      "step": 5080
    },
    {
      "epoch": 0.1158318731083449,
      "grad_norm": 131.0,
      "learning_rate": 1.8044770574029355e-05,
      "loss": 0.4836,
      "step": 5090
    },
    {
      "epoch": 0.11605944063900962,
      "grad_norm": 53.75,
      "learning_rate": 1.804012632361137e-05,
      "loss": 1.062,
      "step": 5100
    },
    {
      "epoch": 0.11628700816967436,
      "grad_norm": 76.5,
      "learning_rate": 1.803548207319339e-05,
      "loss": 0.9775,
      "step": 5110
    },
    {
      "epoch": 0.11651457570033907,
      "grad_norm": 79.0,
      "learning_rate": 1.8030837822775406e-05,
      "loss": 1.2615,
      "step": 5120
    },
    {
      "epoch": 0.1167421432310038,
      "grad_norm": 382.0,
      "learning_rate": 1.8026193572357423e-05,
      "loss": 1.5911,
      "step": 5130
    },
    {
      "epoch": 0.11696971076166852,
      "grad_norm": 79.5,
      "learning_rate": 1.802154932193944e-05,
      "loss": 0.7531,
      "step": 5140
    },
    {
      "epoch": 0.11719727829233326,
      "grad_norm": 378.0,
      "learning_rate": 1.8016905071521457e-05,
      "loss": 1.6273,
      "step": 5150
    },
    {
      "epoch": 0.11742484582299798,
      "grad_norm": 16.625,
      "learning_rate": 1.8012260821103474e-05,
      "loss": 0.6442,
      "step": 5160
    },
    {
      "epoch": 0.1176524133536627,
      "grad_norm": 38.25,
      "learning_rate": 1.8007616570685494e-05,
      "loss": 1.073,
      "step": 5170
    },
    {
      "epoch": 0.11787998088432743,
      "grad_norm": 0.00408935546875,
      "learning_rate": 1.800297232026751e-05,
      "loss": 0.5841,
      "step": 5180
    },
    {
      "epoch": 0.11810754841499214,
      "grad_norm": 0.004913330078125,
      "learning_rate": 1.799832806984953e-05,
      "loss": 0.6945,
      "step": 5190
    },
    {
      "epoch": 0.11833511594565688,
      "grad_norm": 0.004791259765625,
      "learning_rate": 1.7993683819431546e-05,
      "loss": 0.314,
      "step": 5200
    },
    {
      "epoch": 0.1185626834763216,
      "grad_norm": 0.00152587890625,
      "learning_rate": 1.7989039569013563e-05,
      "loss": 0.8284,
      "step": 5210
    },
    {
      "epoch": 0.11879025100698633,
      "grad_norm": 22.125,
      "learning_rate": 1.798439531859558e-05,
      "loss": 0.9987,
      "step": 5220
    },
    {
      "epoch": 0.11901781853765105,
      "grad_norm": 272.0,
      "learning_rate": 1.7979751068177597e-05,
      "loss": 0.5607,
      "step": 5230
    },
    {
      "epoch": 0.11924538606831578,
      "grad_norm": 0.07421875,
      "learning_rate": 1.7975106817759617e-05,
      "loss": 1.4606,
      "step": 5240
    },
    {
      "epoch": 0.1194729535989805,
      "grad_norm": 83.0,
      "learning_rate": 1.7970462567341634e-05,
      "loss": 1.2152,
      "step": 5250
    },
    {
      "epoch": 0.11970052112964522,
      "grad_norm": 282.0,
      "learning_rate": 1.796581831692365e-05,
      "loss": 1.4405,
      "step": 5260
    },
    {
      "epoch": 0.11992808866030995,
      "grad_norm": 43.0,
      "learning_rate": 1.796117406650567e-05,
      "loss": 0.6858,
      "step": 5270
    },
    {
      "epoch": 0.12015565619097467,
      "grad_norm": 0.12451171875,
      "learning_rate": 1.7956529816087685e-05,
      "loss": 1.086,
      "step": 5280
    },
    {
      "epoch": 0.1203832237216394,
      "grad_norm": 2720.0,
      "learning_rate": 1.7951885565669702e-05,
      "loss": 1.2338,
      "step": 5290
    },
    {
      "epoch": 0.12061079125230412,
      "grad_norm": 41.5,
      "learning_rate": 1.794724131525172e-05,
      "loss": 0.6785,
      "step": 5300
    },
    {
      "epoch": 0.12083835878296885,
      "grad_norm": 0.9140625,
      "learning_rate": 1.7942597064833737e-05,
      "loss": 0.9054,
      "step": 5310
    },
    {
      "epoch": 0.12106592631363357,
      "grad_norm": 196.0,
      "learning_rate": 1.7937952814415757e-05,
      "loss": 1.6345,
      "step": 5320
    },
    {
      "epoch": 0.1212934938442983,
      "grad_norm": 418.0,
      "learning_rate": 1.7933308563997774e-05,
      "loss": 2.2791,
      "step": 5330
    },
    {
      "epoch": 0.12152106137496302,
      "grad_norm": 0.0220947265625,
      "learning_rate": 1.7928664313579788e-05,
      "loss": 0.8189,
      "step": 5340
    },
    {
      "epoch": 0.12174862890562775,
      "grad_norm": 121.5,
      "learning_rate": 1.7924020063161808e-05,
      "loss": 1.2166,
      "step": 5350
    },
    {
      "epoch": 0.12197619643629247,
      "grad_norm": 48.5,
      "learning_rate": 1.7919375812743825e-05,
      "loss": 0.8443,
      "step": 5360
    },
    {
      "epoch": 0.12220376396695719,
      "grad_norm": 408.0,
      "learning_rate": 1.7914731562325842e-05,
      "loss": 0.8392,
      "step": 5370
    },
    {
      "epoch": 0.12243133149762192,
      "grad_norm": 48.0,
      "learning_rate": 1.791008731190786e-05,
      "loss": 0.2791,
      "step": 5380
    },
    {
      "epoch": 0.12265889902828664,
      "grad_norm": 77.5,
      "learning_rate": 1.7905443061489876e-05,
      "loss": 0.7567,
      "step": 5390
    },
    {
      "epoch": 0.12288646655895137,
      "grad_norm": 170.0,
      "learning_rate": 1.7900798811071897e-05,
      "loss": 0.882,
      "step": 5400
    },
    {
      "epoch": 0.12311403408961609,
      "grad_norm": 408.0,
      "learning_rate": 1.789615456065391e-05,
      "loss": 0.9804,
      "step": 5410
    },
    {
      "epoch": 0.12334160162028082,
      "grad_norm": 94.5,
      "learning_rate": 1.7891510310235928e-05,
      "loss": 0.6625,
      "step": 5420
    },
    {
      "epoch": 0.12356916915094554,
      "grad_norm": 208.0,
      "learning_rate": 1.7886866059817948e-05,
      "loss": 0.7562,
      "step": 5430
    },
    {
      "epoch": 0.12379673668161027,
      "grad_norm": 386.0,
      "learning_rate": 1.7882221809399965e-05,
      "loss": 1.4922,
      "step": 5440
    },
    {
      "epoch": 0.12402430421227499,
      "grad_norm": 169.0,
      "learning_rate": 1.7877577558981982e-05,
      "loss": 0.7264,
      "step": 5450
    },
    {
      "epoch": 0.12425187174293972,
      "grad_norm": 96.5,
      "learning_rate": 1.7872933308564e-05,
      "loss": 0.6276,
      "step": 5460
    },
    {
      "epoch": 0.12447943927360444,
      "grad_norm": 0.08837890625,
      "learning_rate": 1.7868289058146016e-05,
      "loss": 0.4965,
      "step": 5470
    },
    {
      "epoch": 0.12470700680426916,
      "grad_norm": 222.0,
      "learning_rate": 1.7863644807728033e-05,
      "loss": 1.4582,
      "step": 5480
    },
    {
      "epoch": 0.12493457433493389,
      "grad_norm": 211.0,
      "learning_rate": 1.785900055731005e-05,
      "loss": 1.0934,
      "step": 5490
    },
    {
      "epoch": 0.12516214186559863,
      "grad_norm": 282.0,
      "learning_rate": 1.7854356306892067e-05,
      "loss": 1.0216,
      "step": 5500
    },
    {
      "epoch": 0.12538970939626334,
      "grad_norm": 0.11767578125,
      "learning_rate": 1.7849712056474088e-05,
      "loss": 0.8858,
      "step": 5510
    },
    {
      "epoch": 0.12561727692692806,
      "grad_norm": 0.001251220703125,
      "learning_rate": 1.7845067806056105e-05,
      "loss": 0.6206,
      "step": 5520
    },
    {
      "epoch": 0.12584484445759278,
      "grad_norm": 122.0,
      "learning_rate": 1.7840423555638122e-05,
      "loss": 0.6628,
      "step": 5530
    },
    {
      "epoch": 0.12607241198825753,
      "grad_norm": 154.0,
      "learning_rate": 1.783577930522014e-05,
      "loss": 1.1301,
      "step": 5540
    },
    {
      "epoch": 0.12629997951892225,
      "grad_norm": 0.78125,
      "learning_rate": 1.7831135054802156e-05,
      "loss": 0.8139,
      "step": 5550
    },
    {
      "epoch": 0.12652754704958696,
      "grad_norm": 218.0,
      "learning_rate": 1.7826490804384173e-05,
      "loss": 1.2704,
      "step": 5560
    },
    {
      "epoch": 0.12675511458025168,
      "grad_norm": 0.005645751953125,
      "learning_rate": 1.782184655396619e-05,
      "loss": 1.1157,
      "step": 5570
    },
    {
      "epoch": 0.12698268211091643,
      "grad_norm": 3.6875,
      "learning_rate": 1.781720230354821e-05,
      "loss": 1.1606,
      "step": 5580
    },
    {
      "epoch": 0.12721024964158115,
      "grad_norm": 0.0172119140625,
      "learning_rate": 1.7812558053130228e-05,
      "loss": 1.1314,
      "step": 5590
    },
    {
      "epoch": 0.12743781717224587,
      "grad_norm": 91.0,
      "learning_rate": 1.7807913802712245e-05,
      "loss": 0.6956,
      "step": 5600
    },
    {
      "epoch": 0.12766538470291058,
      "grad_norm": 1048.0,
      "learning_rate": 1.7803269552294262e-05,
      "loss": 1.0079,
      "step": 5610
    },
    {
      "epoch": 0.1278929522335753,
      "grad_norm": 39.25,
      "learning_rate": 1.779862530187628e-05,
      "loss": 0.4496,
      "step": 5620
    },
    {
      "epoch": 0.12812051976424005,
      "grad_norm": 71.5,
      "learning_rate": 1.7793981051458296e-05,
      "loss": 0.3356,
      "step": 5630
    },
    {
      "epoch": 0.12834808729490477,
      "grad_norm": 98.5,
      "learning_rate": 1.7789336801040313e-05,
      "loss": 1.2306,
      "step": 5640
    },
    {
      "epoch": 0.12857565482556949,
      "grad_norm": 140.0,
      "learning_rate": 1.778469255062233e-05,
      "loss": 1.479,
      "step": 5650
    },
    {
      "epoch": 0.1288032223562342,
      "grad_norm": 0.0038909912109375,
      "learning_rate": 1.778004830020435e-05,
      "loss": 1.1418,
      "step": 5660
    },
    {
      "epoch": 0.12903078988689895,
      "grad_norm": 112.5,
      "learning_rate": 1.7775404049786367e-05,
      "loss": 1.1269,
      "step": 5670
    },
    {
      "epoch": 0.12925835741756367,
      "grad_norm": 288.0,
      "learning_rate": 1.7770759799368385e-05,
      "loss": 1.2111,
      "step": 5680
    },
    {
      "epoch": 0.1294859249482284,
      "grad_norm": 22.625,
      "learning_rate": 1.77661155489504e-05,
      "loss": 0.6762,
      "step": 5690
    },
    {
      "epoch": 0.1297134924788931,
      "grad_norm": 0.376953125,
      "learning_rate": 1.776147129853242e-05,
      "loss": 0.5571,
      "step": 5700
    },
    {
      "epoch": 0.12994106000955782,
      "grad_norm": 400.0,
      "learning_rate": 1.7756827048114436e-05,
      "loss": 0.4836,
      "step": 5710
    },
    {
      "epoch": 0.13016862754022257,
      "grad_norm": 1.8359375,
      "learning_rate": 1.7752182797696453e-05,
      "loss": 0.4536,
      "step": 5720
    },
    {
      "epoch": 0.1303961950708873,
      "grad_norm": 69.5,
      "learning_rate": 1.774753854727847e-05,
      "loss": 1.1546,
      "step": 5730
    },
    {
      "epoch": 0.130623762601552,
      "grad_norm": 5.65625,
      "learning_rate": 1.774289429686049e-05,
      "loss": 1.4627,
      "step": 5740
    },
    {
      "epoch": 0.13085133013221673,
      "grad_norm": 0.00138092041015625,
      "learning_rate": 1.7738250046442507e-05,
      "loss": 1.1748,
      "step": 5750
    },
    {
      "epoch": 0.13107889766288147,
      "grad_norm": 0.0035400390625,
      "learning_rate": 1.773360579602452e-05,
      "loss": 1.0538,
      "step": 5760
    },
    {
      "epoch": 0.1313064651935462,
      "grad_norm": 540.0,
      "learning_rate": 1.772896154560654e-05,
      "loss": 0.5145,
      "step": 5770
    },
    {
      "epoch": 0.1315340327242109,
      "grad_norm": 21.0,
      "learning_rate": 1.772431729518856e-05,
      "loss": 1.1869,
      "step": 5780
    },
    {
      "epoch": 0.13176160025487563,
      "grad_norm": 241.0,
      "learning_rate": 1.7719673044770576e-05,
      "loss": 1.4321,
      "step": 5790
    },
    {
      "epoch": 0.13198916778554035,
      "grad_norm": 132.0,
      "learning_rate": 1.7715028794352593e-05,
      "loss": 0.9871,
      "step": 5800
    },
    {
      "epoch": 0.1322167353162051,
      "grad_norm": 176.0,
      "learning_rate": 1.771038454393461e-05,
      "loss": 1.4919,
      "step": 5810
    },
    {
      "epoch": 0.1324443028468698,
      "grad_norm": 41.25,
      "learning_rate": 1.770574029351663e-05,
      "loss": 0.2899,
      "step": 5820
    },
    {
      "epoch": 0.13267187037753453,
      "grad_norm": 0.0021514892578125,
      "learning_rate": 1.7701096043098644e-05,
      "loss": 0.7219,
      "step": 5830
    },
    {
      "epoch": 0.13289943790819925,
      "grad_norm": 175.0,
      "learning_rate": 1.769645179268066e-05,
      "loss": 0.7068,
      "step": 5840
    },
    {
      "epoch": 0.133127005438864,
      "grad_norm": 314.0,
      "learning_rate": 1.769180754226268e-05,
      "loss": 1.0009,
      "step": 5850
    },
    {
      "epoch": 0.1333545729695287,
      "grad_norm": 400.0,
      "learning_rate": 1.7687163291844698e-05,
      "loss": 1.0511,
      "step": 5860
    },
    {
      "epoch": 0.13358214050019343,
      "grad_norm": 0.10205078125,
      "learning_rate": 1.7682519041426715e-05,
      "loss": 0.7609,
      "step": 5870
    },
    {
      "epoch": 0.13380970803085815,
      "grad_norm": 0.11181640625,
      "learning_rate": 1.7677874791008732e-05,
      "loss": 0.7456,
      "step": 5880
    },
    {
      "epoch": 0.1340372755615229,
      "grad_norm": 2.609375,
      "learning_rate": 1.7673230540590753e-05,
      "loss": 0.9804,
      "step": 5890
    },
    {
      "epoch": 0.1342648430921876,
      "grad_norm": 1.9921875,
      "learning_rate": 1.7668586290172766e-05,
      "loss": 0.8484,
      "step": 5900
    },
    {
      "epoch": 0.13449241062285233,
      "grad_norm": 37.75,
      "learning_rate": 1.7663942039754784e-05,
      "loss": 0.6471,
      "step": 5910
    },
    {
      "epoch": 0.13471997815351705,
      "grad_norm": 124.5,
      "learning_rate": 1.76592977893368e-05,
      "loss": 0.4274,
      "step": 5920
    },
    {
      "epoch": 0.13494754568418177,
      "grad_norm": 87.0,
      "learning_rate": 1.765465353891882e-05,
      "loss": 1.25,
      "step": 5930
    },
    {
      "epoch": 0.13517511321484652,
      "grad_norm": 126.5,
      "learning_rate": 1.7650009288500838e-05,
      "loss": 0.6935,
      "step": 5940
    },
    {
      "epoch": 0.13540268074551123,
      "grad_norm": 134.0,
      "learning_rate": 1.7645365038082855e-05,
      "loss": 1.1282,
      "step": 5950
    },
    {
      "epoch": 0.13563024827617595,
      "grad_norm": 600.0,
      "learning_rate": 1.7640720787664872e-05,
      "loss": 2.2582,
      "step": 5960
    },
    {
      "epoch": 0.13585781580684067,
      "grad_norm": 63.0,
      "learning_rate": 1.763607653724689e-05,
      "loss": 1.2863,
      "step": 5970
    },
    {
      "epoch": 0.13608538333750542,
      "grad_norm": 0.2392578125,
      "learning_rate": 1.7631432286828906e-05,
      "loss": 1.4813,
      "step": 5980
    },
    {
      "epoch": 0.13631295086817014,
      "grad_norm": 237.0,
      "learning_rate": 1.7626788036410923e-05,
      "loss": 1.4184,
      "step": 5990
    },
    {
      "epoch": 0.13654051839883485,
      "grad_norm": 252.0,
      "learning_rate": 1.7622143785992944e-05,
      "loss": 0.9143,
      "step": 6000
    },
    {
      "epoch": 0.13676808592949957,
      "grad_norm": 92.0,
      "learning_rate": 1.761749953557496e-05,
      "loss": 1.3036,
      "step": 6010
    },
    {
      "epoch": 0.1369956534601643,
      "grad_norm": 266.0,
      "learning_rate": 1.7612855285156978e-05,
      "loss": 1.0776,
      "step": 6020
    },
    {
      "epoch": 0.13722322099082904,
      "grad_norm": 133.0,
      "learning_rate": 1.7608211034738995e-05,
      "loss": 1.4665,
      "step": 6030
    },
    {
      "epoch": 0.13745078852149376,
      "grad_norm": 0.02099609375,
      "learning_rate": 1.7603566784321012e-05,
      "loss": 0.7046,
      "step": 6040
    },
    {
      "epoch": 0.13767835605215847,
      "grad_norm": 0.03369140625,
      "learning_rate": 1.759892253390303e-05,
      "loss": 0.7061,
      "step": 6050
    },
    {
      "epoch": 0.1379059235828232,
      "grad_norm": 219.0,
      "learning_rate": 1.7594278283485046e-05,
      "loss": 0.645,
      "step": 6060
    },
    {
      "epoch": 0.13813349111348794,
      "grad_norm": 194.0,
      "learning_rate": 1.7589634033067063e-05,
      "loss": 1.5072,
      "step": 6070
    },
    {
      "epoch": 0.13836105864415266,
      "grad_norm": 94.5,
      "learning_rate": 1.7584989782649084e-05,
      "loss": 1.1925,
      "step": 6080
    },
    {
      "epoch": 0.13858862617481738,
      "grad_norm": 296.0,
      "learning_rate": 1.75803455322311e-05,
      "loss": 0.4612,
      "step": 6090
    },
    {
      "epoch": 0.1388161937054821,
      "grad_norm": 0.06201171875,
      "learning_rate": 1.7575701281813118e-05,
      "loss": 0.6033,
      "step": 6100
    },
    {
      "epoch": 0.13904376123614684,
      "grad_norm": 215.0,
      "learning_rate": 1.7571057031395135e-05,
      "loss": 0.491,
      "step": 6110
    },
    {
      "epoch": 0.13927132876681156,
      "grad_norm": 268.0,
      "learning_rate": 1.7566412780977152e-05,
      "loss": 2.2093,
      "step": 6120
    },
    {
      "epoch": 0.13949889629747628,
      "grad_norm": 83.5,
      "learning_rate": 1.756176853055917e-05,
      "loss": 1.8203,
      "step": 6130
    },
    {
      "epoch": 0.139726463828141,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.7557124280141186e-05,
      "loss": 0.3302,
      "step": 6140
    },
    {
      "epoch": 0.13995403135880571,
      "grad_norm": 0.953125,
      "learning_rate": 1.7552480029723203e-05,
      "loss": 1.2541,
      "step": 6150
    },
    {
      "epoch": 0.14018159888947046,
      "grad_norm": 7.28125,
      "learning_rate": 1.7547835779305223e-05,
      "loss": 0.7118,
      "step": 6160
    },
    {
      "epoch": 0.14040916642013518,
      "grad_norm": 0.43359375,
      "learning_rate": 1.754319152888724e-05,
      "loss": 0.8478,
      "step": 6170
    },
    {
      "epoch": 0.1406367339507999,
      "grad_norm": 122.5,
      "learning_rate": 1.7538547278469254e-05,
      "loss": 1.0271,
      "step": 6180
    },
    {
      "epoch": 0.14086430148146462,
      "grad_norm": 0.40625,
      "learning_rate": 1.7533903028051275e-05,
      "loss": 0.6457,
      "step": 6190
    },
    {
      "epoch": 0.14109186901212936,
      "grad_norm": 12.5,
      "learning_rate": 1.7529258777633292e-05,
      "loss": 1.44,
      "step": 6200
    },
    {
      "epoch": 0.14131943654279408,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.752461452721531e-05,
      "loss": 2.0494,
      "step": 6210
    },
    {
      "epoch": 0.1415470040734588,
      "grad_norm": 76.0,
      "learning_rate": 1.7519970276797326e-05,
      "loss": 1.0752,
      "step": 6220
    },
    {
      "epoch": 0.14177457160412352,
      "grad_norm": 143.0,
      "learning_rate": 1.7515326026379343e-05,
      "loss": 1.5948,
      "step": 6230
    },
    {
      "epoch": 0.14200213913478824,
      "grad_norm": 60.0,
      "learning_rate": 1.7510681775961363e-05,
      "loss": 0.8077,
      "step": 6240
    },
    {
      "epoch": 0.14222970666545298,
      "grad_norm": 0.3515625,
      "learning_rate": 1.7506037525543377e-05,
      "loss": 0.4638,
      "step": 6250
    },
    {
      "epoch": 0.1424572741961177,
      "grad_norm": 176.0,
      "learning_rate": 1.7501393275125394e-05,
      "loss": 0.9143,
      "step": 6260
    },
    {
      "epoch": 0.14268484172678242,
      "grad_norm": 137.0,
      "learning_rate": 1.7496749024707414e-05,
      "loss": 0.6196,
      "step": 6270
    },
    {
      "epoch": 0.14291240925744714,
      "grad_norm": 0.283203125,
      "learning_rate": 1.749210477428943e-05,
      "loss": 0.7072,
      "step": 6280
    },
    {
      "epoch": 0.14313997678811188,
      "grad_norm": 102.5,
      "learning_rate": 1.748746052387145e-05,
      "loss": 1.1552,
      "step": 6290
    },
    {
      "epoch": 0.1433675443187766,
      "grad_norm": 0.015625,
      "learning_rate": 1.7482816273453466e-05,
      "loss": 0.1142,
      "step": 6300
    },
    {
      "epoch": 0.14359511184944132,
      "grad_norm": 30.875,
      "learning_rate": 1.7478172023035486e-05,
      "loss": 0.802,
      "step": 6310
    },
    {
      "epoch": 0.14382267938010604,
      "grad_norm": 169.0,
      "learning_rate": 1.74735277726175e-05,
      "loss": 0.5857,
      "step": 6320
    },
    {
      "epoch": 0.14405024691077079,
      "grad_norm": 102.5,
      "learning_rate": 1.7468883522199517e-05,
      "loss": 1.1703,
      "step": 6330
    },
    {
      "epoch": 0.1442778144414355,
      "grad_norm": 14.375,
      "learning_rate": 1.7464239271781537e-05,
      "loss": 0.1739,
      "step": 6340
    },
    {
      "epoch": 0.14450538197210022,
      "grad_norm": 4.875,
      "learning_rate": 1.7459595021363554e-05,
      "loss": 2.0655,
      "step": 6350
    },
    {
      "epoch": 0.14473294950276494,
      "grad_norm": 60.5,
      "learning_rate": 1.745495077094557e-05,
      "loss": 0.4238,
      "step": 6360
    },
    {
      "epoch": 0.14496051703342966,
      "grad_norm": 152.0,
      "learning_rate": 1.745030652052759e-05,
      "loss": 1.2654,
      "step": 6370
    },
    {
      "epoch": 0.1451880845640944,
      "grad_norm": 21.75,
      "learning_rate": 1.7445662270109605e-05,
      "loss": 1.1451,
      "step": 6380
    },
    {
      "epoch": 0.14541565209475912,
      "grad_norm": 111.5,
      "learning_rate": 1.7441018019691623e-05,
      "loss": 0.5242,
      "step": 6390
    },
    {
      "epoch": 0.14564321962542384,
      "grad_norm": 0.8828125,
      "learning_rate": 1.743637376927364e-05,
      "loss": 1.1951,
      "step": 6400
    },
    {
      "epoch": 0.14587078715608856,
      "grad_norm": 169.0,
      "learning_rate": 1.7431729518855657e-05,
      "loss": 1.0785,
      "step": 6410
    },
    {
      "epoch": 0.1460983546867533,
      "grad_norm": 168.0,
      "learning_rate": 1.7427085268437677e-05,
      "loss": 1.0101,
      "step": 6420
    },
    {
      "epoch": 0.14632592221741803,
      "grad_norm": 0.0029754638671875,
      "learning_rate": 1.7422441018019694e-05,
      "loss": 0.5591,
      "step": 6430
    },
    {
      "epoch": 0.14655348974808274,
      "grad_norm": 0.07763671875,
      "learning_rate": 1.741779676760171e-05,
      "loss": 1.203,
      "step": 6440
    },
    {
      "epoch": 0.14678105727874746,
      "grad_norm": 63.0,
      "learning_rate": 1.7413152517183728e-05,
      "loss": 0.6398,
      "step": 6450
    },
    {
      "epoch": 0.14700862480941218,
      "grad_norm": 63.0,
      "learning_rate": 1.7408508266765745e-05,
      "loss": 0.7362,
      "step": 6460
    },
    {
      "epoch": 0.14723619234007693,
      "grad_norm": 756.0,
      "learning_rate": 1.7403864016347762e-05,
      "loss": 1.0576,
      "step": 6470
    },
    {
      "epoch": 0.14746375987074165,
      "grad_norm": 122.0,
      "learning_rate": 1.739921976592978e-05,
      "loss": 0.4032,
      "step": 6480
    },
    {
      "epoch": 0.14769132740140636,
      "grad_norm": 150.0,
      "learning_rate": 1.7394575515511796e-05,
      "loss": 0.8499,
      "step": 6490
    },
    {
      "epoch": 0.14791889493207108,
      "grad_norm": 0.0185546875,
      "learning_rate": 1.7389931265093817e-05,
      "loss": 1.0421,
      "step": 6500
    },
    {
      "epoch": 0.14814646246273583,
      "grad_norm": 185.0,
      "learning_rate": 1.7385287014675834e-05,
      "loss": 1.1967,
      "step": 6510
    },
    {
      "epoch": 0.14837402999340055,
      "grad_norm": 21.25,
      "learning_rate": 1.738064276425785e-05,
      "loss": 0.6081,
      "step": 6520
    },
    {
      "epoch": 0.14860159752406527,
      "grad_norm": 166.0,
      "learning_rate": 1.7375998513839868e-05,
      "loss": 1.1182,
      "step": 6530
    },
    {
      "epoch": 0.14882916505472998,
      "grad_norm": 160.0,
      "learning_rate": 1.7371354263421885e-05,
      "loss": 0.5833,
      "step": 6540
    },
    {
      "epoch": 0.1490567325853947,
      "grad_norm": 229.0,
      "learning_rate": 1.7366710013003902e-05,
      "loss": 0.6007,
      "step": 6550
    },
    {
      "epoch": 0.14928430011605945,
      "grad_norm": 0.0034942626953125,
      "learning_rate": 1.736206576258592e-05,
      "loss": 1.3705,
      "step": 6560
    },
    {
      "epoch": 0.14951186764672417,
      "grad_norm": 276.0,
      "learning_rate": 1.7357421512167936e-05,
      "loss": 1.8626,
      "step": 6570
    },
    {
      "epoch": 0.14973943517738889,
      "grad_norm": 332.0,
      "learning_rate": 1.7352777261749957e-05,
      "loss": 1.2415,
      "step": 6580
    },
    {
      "epoch": 0.1499670027080536,
      "grad_norm": 203.0,
      "learning_rate": 1.7348133011331974e-05,
      "loss": 1.7854,
      "step": 6590
    },
    {
      "epoch": 0.15019457023871835,
      "grad_norm": 126.5,
      "learning_rate": 1.7343488760913987e-05,
      "loss": 0.8788,
      "step": 6600
    },
    {
      "epoch": 0.15042213776938307,
      "grad_norm": 125.5,
      "learning_rate": 1.7338844510496008e-05,
      "loss": 0.5434,
      "step": 6610
    },
    {
      "epoch": 0.1506497053000478,
      "grad_norm": 213.0,
      "learning_rate": 1.7334200260078025e-05,
      "loss": 1.3607,
      "step": 6620
    },
    {
      "epoch": 0.1508772728307125,
      "grad_norm": 177.0,
      "learning_rate": 1.7329556009660042e-05,
      "loss": 1.2645,
      "step": 6630
    },
    {
      "epoch": 0.15110484036137725,
      "grad_norm": 0.016845703125,
      "learning_rate": 1.732491175924206e-05,
      "loss": 0.8235,
      "step": 6640
    },
    {
      "epoch": 0.15133240789204197,
      "grad_norm": 0.451171875,
      "learning_rate": 1.732026750882408e-05,
      "loss": 0.9221,
      "step": 6650
    },
    {
      "epoch": 0.1515599754227067,
      "grad_norm": 0.004669189453125,
      "learning_rate": 1.7315623258406097e-05,
      "loss": 0.6191,
      "step": 6660
    },
    {
      "epoch": 0.1517875429533714,
      "grad_norm": 128.0,
      "learning_rate": 1.731097900798811e-05,
      "loss": 0.5876,
      "step": 6670
    },
    {
      "epoch": 0.15201511048403613,
      "grad_norm": 245.0,
      "learning_rate": 1.730633475757013e-05,
      "loss": 1.759,
      "step": 6680
    },
    {
      "epoch": 0.15224267801470087,
      "grad_norm": 5.46875,
      "learning_rate": 1.7301690507152148e-05,
      "loss": 1.3202,
      "step": 6690
    },
    {
      "epoch": 0.1524702455453656,
      "grad_norm": 8.25,
      "learning_rate": 1.7297046256734165e-05,
      "loss": 0.9397,
      "step": 6700
    },
    {
      "epoch": 0.1526978130760303,
      "grad_norm": 116.0,
      "learning_rate": 1.7292402006316182e-05,
      "loss": 1.0977,
      "step": 6710
    },
    {
      "epoch": 0.15292538060669503,
      "grad_norm": 0.1708984375,
      "learning_rate": 1.72877577558982e-05,
      "loss": 0.5098,
      "step": 6720
    },
    {
      "epoch": 0.15315294813735977,
      "grad_norm": 194.0,
      "learning_rate": 1.728311350548022e-05,
      "loss": 0.6471,
      "step": 6730
    },
    {
      "epoch": 0.1533805156680245,
      "grad_norm": 150.0,
      "learning_rate": 1.7278469255062233e-05,
      "loss": 0.6037,
      "step": 6740
    },
    {
      "epoch": 0.1536080831986892,
      "grad_norm": 133.0,
      "learning_rate": 1.727382500464425e-05,
      "loss": 0.3354,
      "step": 6750
    },
    {
      "epoch": 0.15383565072935393,
      "grad_norm": 6.0625,
      "learning_rate": 1.726918075422627e-05,
      "loss": 1.2476,
      "step": 6760
    },
    {
      "epoch": 0.15406321826001865,
      "grad_norm": 2.296875,
      "learning_rate": 1.7264536503808288e-05,
      "loss": 0.7941,
      "step": 6770
    },
    {
      "epoch": 0.1542907857906834,
      "grad_norm": 0.004180908203125,
      "learning_rate": 1.7259892253390305e-05,
      "loss": 0.7447,
      "step": 6780
    },
    {
      "epoch": 0.1545183533213481,
      "grad_norm": 126.5,
      "learning_rate": 1.725524800297232e-05,
      "loss": 0.386,
      "step": 6790
    },
    {
      "epoch": 0.15474592085201283,
      "grad_norm": 0.30078125,
      "learning_rate": 1.725060375255434e-05,
      "loss": 0.5434,
      "step": 6800
    },
    {
      "epoch": 0.15497348838267755,
      "grad_norm": 153.0,
      "learning_rate": 1.7245959502136356e-05,
      "loss": 0.4657,
      "step": 6810
    },
    {
      "epoch": 0.1552010559133423,
      "grad_norm": 0.625,
      "learning_rate": 1.7241315251718373e-05,
      "loss": 1.4304,
      "step": 6820
    },
    {
      "epoch": 0.155428623444007,
      "grad_norm": 2.28125,
      "learning_rate": 1.723667100130039e-05,
      "loss": 0.7146,
      "step": 6830
    },
    {
      "epoch": 0.15565619097467173,
      "grad_norm": 184.0,
      "learning_rate": 1.723202675088241e-05,
      "loss": 0.5056,
      "step": 6840
    },
    {
      "epoch": 0.15588375850533645,
      "grad_norm": 0.416015625,
      "learning_rate": 1.7227382500464427e-05,
      "loss": 0.4158,
      "step": 6850
    },
    {
      "epoch": 0.1561113260360012,
      "grad_norm": 0.00019931793212890625,
      "learning_rate": 1.7222738250046444e-05,
      "loss": 1.3381,
      "step": 6860
    },
    {
      "epoch": 0.15633889356666592,
      "grad_norm": 100.5,
      "learning_rate": 1.721809399962846e-05,
      "loss": 1.2948,
      "step": 6870
    },
    {
      "epoch": 0.15656646109733063,
      "grad_norm": 0.0002899169921875,
      "learning_rate": 1.721344974921048e-05,
      "loss": 0.4134,
      "step": 6880
    },
    {
      "epoch": 0.15679402862799535,
      "grad_norm": 0.18359375,
      "learning_rate": 1.7208805498792496e-05,
      "loss": 0.514,
      "step": 6890
    },
    {
      "epoch": 0.15702159615866007,
      "grad_norm": 129.0,
      "learning_rate": 1.7204161248374513e-05,
      "loss": 1.7467,
      "step": 6900
    },
    {
      "epoch": 0.15724916368932482,
      "grad_norm": 136.0,
      "learning_rate": 1.719951699795653e-05,
      "loss": 0.0679,
      "step": 6910
    },
    {
      "epoch": 0.15747673121998954,
      "grad_norm": 115.5,
      "learning_rate": 1.719487274753855e-05,
      "loss": 0.8541,
      "step": 6920
    },
    {
      "epoch": 0.15770429875065425,
      "grad_norm": 0.1416015625,
      "learning_rate": 1.7190228497120567e-05,
      "loss": 0.5283,
      "step": 6930
    },
    {
      "epoch": 0.15793186628131897,
      "grad_norm": 163.0,
      "learning_rate": 1.7185584246702584e-05,
      "loss": 2.3244,
      "step": 6940
    },
    {
      "epoch": 0.15815943381198372,
      "grad_norm": 274.0,
      "learning_rate": 1.71809399962846e-05,
      "loss": 1.0178,
      "step": 6950
    },
    {
      "epoch": 0.15838700134264844,
      "grad_norm": 39.0,
      "learning_rate": 1.717629574586662e-05,
      "loss": 1.1244,
      "step": 6960
    },
    {
      "epoch": 0.15861456887331316,
      "grad_norm": 292.0,
      "learning_rate": 1.7171651495448635e-05,
      "loss": 0.7638,
      "step": 6970
    },
    {
      "epoch": 0.15884213640397787,
      "grad_norm": 0.00244140625,
      "learning_rate": 1.7167007245030652e-05,
      "loss": 1.0443,
      "step": 6980
    },
    {
      "epoch": 0.1590697039346426,
      "grad_norm": 60.25,
      "learning_rate": 1.7162362994612673e-05,
      "loss": 0.9193,
      "step": 6990
    },
    {
      "epoch": 0.15929727146530734,
      "grad_norm": 0.0015716552734375,
      "learning_rate": 1.715771874419469e-05,
      "loss": 1.0885,
      "step": 7000
    },
    {
      "epoch": 0.15952483899597206,
      "grad_norm": 122.5,
      "learning_rate": 1.7153074493776707e-05,
      "loss": 0.759,
      "step": 7010
    },
    {
      "epoch": 0.15975240652663678,
      "grad_norm": 402.0,
      "learning_rate": 1.7148430243358724e-05,
      "loss": 1.4976,
      "step": 7020
    },
    {
      "epoch": 0.1599799740573015,
      "grad_norm": 93.0,
      "learning_rate": 1.714378599294074e-05,
      "loss": 0.6965,
      "step": 7030
    },
    {
      "epoch": 0.16020754158796624,
      "grad_norm": 143.0,
      "learning_rate": 1.7139141742522758e-05,
      "loss": 0.5035,
      "step": 7040
    },
    {
      "epoch": 0.16043510911863096,
      "grad_norm": 197.0,
      "learning_rate": 1.7134497492104775e-05,
      "loss": 1.0245,
      "step": 7050
    },
    {
      "epoch": 0.16066267664929568,
      "grad_norm": 22.625,
      "learning_rate": 1.7129853241686792e-05,
      "loss": 1.0365,
      "step": 7060
    },
    {
      "epoch": 0.1608902441799604,
      "grad_norm": 8.9375,
      "learning_rate": 1.7125208991268813e-05,
      "loss": 0.7281,
      "step": 7070
    },
    {
      "epoch": 0.16111781171062514,
      "grad_norm": 0.00128936767578125,
      "learning_rate": 1.712056474085083e-05,
      "loss": 0.2253,
      "step": 7080
    },
    {
      "epoch": 0.16134537924128986,
      "grad_norm": 0.01544189453125,
      "learning_rate": 1.7115920490432843e-05,
      "loss": 0.7205,
      "step": 7090
    },
    {
      "epoch": 0.16157294677195458,
      "grad_norm": 540.0,
      "learning_rate": 1.7111276240014864e-05,
      "loss": 1.5571,
      "step": 7100
    },
    {
      "epoch": 0.1618005143026193,
      "grad_norm": 0.00506591796875,
      "learning_rate": 1.710663198959688e-05,
      "loss": 0.7409,
      "step": 7110
    },
    {
      "epoch": 0.16202808183328402,
      "grad_norm": 292.0,
      "learning_rate": 1.7101987739178898e-05,
      "loss": 1.1149,
      "step": 7120
    },
    {
      "epoch": 0.16225564936394876,
      "grad_norm": 33.25,
      "learning_rate": 1.7097343488760915e-05,
      "loss": 0.8775,
      "step": 7130
    },
    {
      "epoch": 0.16248321689461348,
      "grad_norm": 9.5625,
      "learning_rate": 1.7092699238342932e-05,
      "loss": 1.2986,
      "step": 7140
    },
    {
      "epoch": 0.1627107844252782,
      "grad_norm": 100.5,
      "learning_rate": 1.7088054987924953e-05,
      "loss": 0.2365,
      "step": 7150
    },
    {
      "epoch": 0.16293835195594292,
      "grad_norm": 0.1005859375,
      "learning_rate": 1.7083410737506966e-05,
      "loss": 0.447,
      "step": 7160
    },
    {
      "epoch": 0.16316591948660766,
      "grad_norm": 80.5,
      "learning_rate": 1.7078766487088983e-05,
      "loss": 1.1457,
      "step": 7170
    },
    {
      "epoch": 0.16339348701727238,
      "grad_norm": 67.5,
      "learning_rate": 1.7074122236671004e-05,
      "loss": 1.4269,
      "step": 7180
    },
    {
      "epoch": 0.1636210545479371,
      "grad_norm": 106.0,
      "learning_rate": 1.706947798625302e-05,
      "loss": 0.7253,
      "step": 7190
    },
    {
      "epoch": 0.16384862207860182,
      "grad_norm": 83.0,
      "learning_rate": 1.7064833735835038e-05,
      "loss": 0.8678,
      "step": 7200
    },
    {
      "epoch": 0.16407618960926654,
      "grad_norm": 312.0,
      "learning_rate": 1.7060189485417055e-05,
      "loss": 1.5495,
      "step": 7210
    },
    {
      "epoch": 0.16430375713993128,
      "grad_norm": 374.0,
      "learning_rate": 1.7055545234999072e-05,
      "loss": 0.9757,
      "step": 7220
    },
    {
      "epoch": 0.164531324670596,
      "grad_norm": 169.0,
      "learning_rate": 1.705090098458109e-05,
      "loss": 1.7204,
      "step": 7230
    },
    {
      "epoch": 0.16475889220126072,
      "grad_norm": 302.0,
      "learning_rate": 1.7046256734163106e-05,
      "loss": 1.2606,
      "step": 7240
    },
    {
      "epoch": 0.16498645973192544,
      "grad_norm": 92.5,
      "learning_rate": 1.7041612483745123e-05,
      "loss": 1.2893,
      "step": 7250
    },
    {
      "epoch": 0.16521402726259019,
      "grad_norm": 6.21875,
      "learning_rate": 1.7036968233327144e-05,
      "loss": 0.4885,
      "step": 7260
    },
    {
      "epoch": 0.1654415947932549,
      "grad_norm": 231.0,
      "learning_rate": 1.703232398290916e-05,
      "loss": 0.8655,
      "step": 7270
    },
    {
      "epoch": 0.16566916232391962,
      "grad_norm": 43.0,
      "learning_rate": 1.7027679732491178e-05,
      "loss": 0.7372,
      "step": 7280
    },
    {
      "epoch": 0.16589672985458434,
      "grad_norm": 115.0,
      "learning_rate": 1.7023035482073195e-05,
      "loss": 1.7704,
      "step": 7290
    },
    {
      "epoch": 0.16612429738524906,
      "grad_norm": 35.75,
      "learning_rate": 1.7018391231655212e-05,
      "loss": 0.96,
      "step": 7300
    },
    {
      "epoch": 0.1663518649159138,
      "grad_norm": 252.0,
      "learning_rate": 1.701374698123723e-05,
      "loss": 0.9163,
      "step": 7310
    },
    {
      "epoch": 0.16657943244657852,
      "grad_norm": 360.0,
      "learning_rate": 1.7009102730819246e-05,
      "loss": 1.3474,
      "step": 7320
    },
    {
      "epoch": 0.16680699997724324,
      "grad_norm": 145.0,
      "learning_rate": 1.7004458480401266e-05,
      "loss": 1.0022,
      "step": 7330
    },
    {
      "epoch": 0.16703456750790796,
      "grad_norm": 235.0,
      "learning_rate": 1.6999814229983283e-05,
      "loss": 1.0826,
      "step": 7340
    },
    {
      "epoch": 0.1672621350385727,
      "grad_norm": 0.0213623046875,
      "learning_rate": 1.69951699795653e-05,
      "loss": 0.9586,
      "step": 7350
    },
    {
      "epoch": 0.16748970256923743,
      "grad_norm": 358.0,
      "learning_rate": 1.6990525729147318e-05,
      "loss": 1.6082,
      "step": 7360
    },
    {
      "epoch": 0.16771727009990214,
      "grad_norm": 0.05322265625,
      "learning_rate": 1.6985881478729335e-05,
      "loss": 0.6437,
      "step": 7370
    },
    {
      "epoch": 0.16794483763056686,
      "grad_norm": 25.125,
      "learning_rate": 1.698123722831135e-05,
      "loss": 0.575,
      "step": 7380
    },
    {
      "epoch": 0.1681724051612316,
      "grad_norm": 0.0004558563232421875,
      "learning_rate": 1.697659297789337e-05,
      "loss": 0.375,
      "step": 7390
    },
    {
      "epoch": 0.16839997269189633,
      "grad_norm": 294.0,
      "learning_rate": 1.6971948727475386e-05,
      "loss": 1.0458,
      "step": 7400
    },
    {
      "epoch": 0.16862754022256105,
      "grad_norm": 0.00057220458984375,
      "learning_rate": 1.6967304477057406e-05,
      "loss": 1.5702,
      "step": 7410
    },
    {
      "epoch": 0.16885510775322576,
      "grad_norm": 89.0,
      "learning_rate": 1.6962660226639423e-05,
      "loss": 1.0285,
      "step": 7420
    },
    {
      "epoch": 0.16908267528389048,
      "grad_norm": 0.06494140625,
      "learning_rate": 1.695801597622144e-05,
      "loss": 0.8773,
      "step": 7430
    },
    {
      "epoch": 0.16931024281455523,
      "grad_norm": 82.5,
      "learning_rate": 1.6953371725803457e-05,
      "loss": 1.412,
      "step": 7440
    },
    {
      "epoch": 0.16953781034521995,
      "grad_norm": 58.25,
      "learning_rate": 1.6948727475385474e-05,
      "loss": 1.0848,
      "step": 7450
    },
    {
      "epoch": 0.16976537787588467,
      "grad_norm": 103.0,
      "learning_rate": 1.694408322496749e-05,
      "loss": 0.4552,
      "step": 7460
    },
    {
      "epoch": 0.16999294540654938,
      "grad_norm": 166.0,
      "learning_rate": 1.693943897454951e-05,
      "loss": 0.7059,
      "step": 7470
    },
    {
      "epoch": 0.17022051293721413,
      "grad_norm": 160.0,
      "learning_rate": 1.6934794724131526e-05,
      "loss": 1.3265,
      "step": 7480
    },
    {
      "epoch": 0.17044808046787885,
      "grad_norm": 92.0,
      "learning_rate": 1.6930150473713546e-05,
      "loss": 1.2586,
      "step": 7490
    },
    {
      "epoch": 0.17067564799854357,
      "grad_norm": 0.04541015625,
      "learning_rate": 1.6925506223295563e-05,
      "loss": 1.5763,
      "step": 7500
    },
    {
      "epoch": 0.17090321552920829,
      "grad_norm": 13.1875,
      "learning_rate": 1.6920861972877577e-05,
      "loss": 1.3828,
      "step": 7510
    },
    {
      "epoch": 0.171130783059873,
      "grad_norm": 36.0,
      "learning_rate": 1.6916217722459597e-05,
      "loss": 1.8073,
      "step": 7520
    },
    {
      "epoch": 0.17135835059053775,
      "grad_norm": 270.0,
      "learning_rate": 1.6911573472041614e-05,
      "loss": 1.1041,
      "step": 7530
    },
    {
      "epoch": 0.17158591812120247,
      "grad_norm": 332.0,
      "learning_rate": 1.690692922162363e-05,
      "loss": 1.2955,
      "step": 7540
    },
    {
      "epoch": 0.1718134856518672,
      "grad_norm": 342.0,
      "learning_rate": 1.690228497120565e-05,
      "loss": 1.2396,
      "step": 7550
    },
    {
      "epoch": 0.1720410531825319,
      "grad_norm": 376.0,
      "learning_rate": 1.6897640720787665e-05,
      "loss": 0.9224,
      "step": 7560
    },
    {
      "epoch": 0.17226862071319665,
      "grad_norm": 133.0,
      "learning_rate": 1.6892996470369686e-05,
      "loss": 1.2762,
      "step": 7570
    },
    {
      "epoch": 0.17249618824386137,
      "grad_norm": 145.0,
      "learning_rate": 1.68883522199517e-05,
      "loss": 1.0612,
      "step": 7580
    },
    {
      "epoch": 0.1727237557745261,
      "grad_norm": 177.0,
      "learning_rate": 1.6883707969533717e-05,
      "loss": 0.762,
      "step": 7590
    },
    {
      "epoch": 0.1729513233051908,
      "grad_norm": 3.75,
      "learning_rate": 1.6879063719115737e-05,
      "loss": 0.7186,
      "step": 7600
    },
    {
      "epoch": 0.17317889083585555,
      "grad_norm": 0.0247802734375,
      "learning_rate": 1.6874419468697754e-05,
      "loss": 1.5222,
      "step": 7610
    },
    {
      "epoch": 0.17340645836652027,
      "grad_norm": 176.0,
      "learning_rate": 1.686977521827977e-05,
      "loss": 0.8191,
      "step": 7620
    },
    {
      "epoch": 0.173634025897185,
      "grad_norm": 77.5,
      "learning_rate": 1.6865130967861788e-05,
      "loss": 0.4421,
      "step": 7630
    },
    {
      "epoch": 0.1738615934278497,
      "grad_norm": 104.5,
      "learning_rate": 1.686048671744381e-05,
      "loss": 0.7015,
      "step": 7640
    },
    {
      "epoch": 0.17408916095851443,
      "grad_norm": 57.5,
      "learning_rate": 1.6855842467025822e-05,
      "loss": 0.7043,
      "step": 7650
    },
    {
      "epoch": 0.17431672848917917,
      "grad_norm": 88.5,
      "learning_rate": 1.685119821660784e-05,
      "loss": 0.4255,
      "step": 7660
    },
    {
      "epoch": 0.1745442960198439,
      "grad_norm": 11.125,
      "learning_rate": 1.6846553966189856e-05,
      "loss": 1.2705,
      "step": 7670
    },
    {
      "epoch": 0.1747718635505086,
      "grad_norm": 15.8125,
      "learning_rate": 1.6841909715771877e-05,
      "loss": 0.3141,
      "step": 7680
    },
    {
      "epoch": 0.17499943108117333,
      "grad_norm": 195.0,
      "learning_rate": 1.6837265465353894e-05,
      "loss": 1.3025,
      "step": 7690
    },
    {
      "epoch": 0.17522699861183808,
      "grad_norm": 0.00070953369140625,
      "learning_rate": 1.683262121493591e-05,
      "loss": 0.194,
      "step": 7700
    },
    {
      "epoch": 0.1754545661425028,
      "grad_norm": 2.328125,
      "learning_rate": 1.6827976964517928e-05,
      "loss": 1.2467,
      "step": 7710
    },
    {
      "epoch": 0.1756821336731675,
      "grad_norm": 17.375,
      "learning_rate": 1.6823332714099945e-05,
      "loss": 0.8494,
      "step": 7720
    },
    {
      "epoch": 0.17590970120383223,
      "grad_norm": 30.875,
      "learning_rate": 1.6818688463681962e-05,
      "loss": 0.8112,
      "step": 7730
    },
    {
      "epoch": 0.17613726873449695,
      "grad_norm": 312.0,
      "learning_rate": 1.681404421326398e-05,
      "loss": 1.2374,
      "step": 7740
    },
    {
      "epoch": 0.1763648362651617,
      "grad_norm": 0.11767578125,
      "learning_rate": 1.6809399962846e-05,
      "loss": 0.6607,
      "step": 7750
    },
    {
      "epoch": 0.1765924037958264,
      "grad_norm": 93.5,
      "learning_rate": 1.6804755712428017e-05,
      "loss": 1.3001,
      "step": 7760
    },
    {
      "epoch": 0.17681997132649113,
      "grad_norm": 86.5,
      "learning_rate": 1.6800111462010034e-05,
      "loss": 1.0569,
      "step": 7770
    },
    {
      "epoch": 0.17704753885715585,
      "grad_norm": 0.00579833984375,
      "learning_rate": 1.679546721159205e-05,
      "loss": 0.9398,
      "step": 7780
    },
    {
      "epoch": 0.1772751063878206,
      "grad_norm": 91.0,
      "learning_rate": 1.6790822961174068e-05,
      "loss": 0.9225,
      "step": 7790
    },
    {
      "epoch": 0.17750267391848532,
      "grad_norm": 0.058349609375,
      "learning_rate": 1.6786178710756085e-05,
      "loss": 1.1901,
      "step": 7800
    },
    {
      "epoch": 0.17773024144915003,
      "grad_norm": 56.0,
      "learning_rate": 1.6781534460338102e-05,
      "loss": 0.8603,
      "step": 7810
    },
    {
      "epoch": 0.17795780897981475,
      "grad_norm": 262.0,
      "learning_rate": 1.677689020992012e-05,
      "loss": 1.119,
      "step": 7820
    },
    {
      "epoch": 0.1781853765104795,
      "grad_norm": 0.0272216796875,
      "learning_rate": 1.677224595950214e-05,
      "loss": 1.3608,
      "step": 7830
    },
    {
      "epoch": 0.17841294404114422,
      "grad_norm": 158.0,
      "learning_rate": 1.6767601709084156e-05,
      "loss": 0.6991,
      "step": 7840
    },
    {
      "epoch": 0.17864051157180894,
      "grad_norm": 120.0,
      "learning_rate": 1.6762957458666174e-05,
      "loss": 0.5036,
      "step": 7850
    },
    {
      "epoch": 0.17886807910247365,
      "grad_norm": 0.62890625,
      "learning_rate": 1.675831320824819e-05,
      "loss": 1.0562,
      "step": 7860
    },
    {
      "epoch": 0.17909564663313837,
      "grad_norm": 0.035888671875,
      "learning_rate": 1.6753668957830208e-05,
      "loss": 0.5813,
      "step": 7870
    },
    {
      "epoch": 0.17932321416380312,
      "grad_norm": 0.00185394287109375,
      "learning_rate": 1.6749024707412225e-05,
      "loss": 0.2205,
      "step": 7880
    },
    {
      "epoch": 0.17955078169446784,
      "grad_norm": 48.25,
      "learning_rate": 1.6744380456994242e-05,
      "loss": 0.5209,
      "step": 7890
    },
    {
      "epoch": 0.17977834922513256,
      "grad_norm": 62.25,
      "learning_rate": 1.673973620657626e-05,
      "loss": 1.117,
      "step": 7900
    },
    {
      "epoch": 0.18000591675579727,
      "grad_norm": 412.0,
      "learning_rate": 1.673509195615828e-05,
      "loss": 2.0653,
      "step": 7910
    },
    {
      "epoch": 0.18023348428646202,
      "grad_norm": 125.0,
      "learning_rate": 1.6730447705740296e-05,
      "loss": 0.9558,
      "step": 7920
    },
    {
      "epoch": 0.18046105181712674,
      "grad_norm": 149.0,
      "learning_rate": 1.672580345532231e-05,
      "loss": 0.6131,
      "step": 7930
    },
    {
      "epoch": 0.18068861934779146,
      "grad_norm": 0.03564453125,
      "learning_rate": 1.672115920490433e-05,
      "loss": 0.6854,
      "step": 7940
    },
    {
      "epoch": 0.18091618687845618,
      "grad_norm": 1.234375,
      "learning_rate": 1.6716514954486347e-05,
      "loss": 1.5561,
      "step": 7950
    },
    {
      "epoch": 0.1811437544091209,
      "grad_norm": 0.07666015625,
      "learning_rate": 1.6711870704068365e-05,
      "loss": 0.7332,
      "step": 7960
    },
    {
      "epoch": 0.18137132193978564,
      "grad_norm": 228.0,
      "learning_rate": 1.670722645365038e-05,
      "loss": 1.0131,
      "step": 7970
    },
    {
      "epoch": 0.18159888947045036,
      "grad_norm": 3.21875,
      "learning_rate": 1.67025822032324e-05,
      "loss": 1.1754,
      "step": 7980
    },
    {
      "epoch": 0.18182645700111508,
      "grad_norm": 204.0,
      "learning_rate": 1.669793795281442e-05,
      "loss": 1.5442,
      "step": 7990
    },
    {
      "epoch": 0.1820540245317798,
      "grad_norm": 9.0625,
      "learning_rate": 1.6693293702396433e-05,
      "loss": 1.109,
      "step": 8000
    },
    {
      "epoch": 0.18228159206244454,
      "grad_norm": 185.0,
      "learning_rate": 1.668864945197845e-05,
      "loss": 0.7482,
      "step": 8010
    },
    {
      "epoch": 0.18250915959310926,
      "grad_norm": 0.0634765625,
      "learning_rate": 1.668400520156047e-05,
      "loss": 0.7846,
      "step": 8020
    },
    {
      "epoch": 0.18273672712377398,
      "grad_norm": 204.0,
      "learning_rate": 1.6679360951142487e-05,
      "loss": 2.0096,
      "step": 8030
    },
    {
      "epoch": 0.1829642946544387,
      "grad_norm": 0.859375,
      "learning_rate": 1.6674716700724504e-05,
      "loss": 0.8699,
      "step": 8040
    },
    {
      "epoch": 0.18319186218510342,
      "grad_norm": 191.0,
      "learning_rate": 1.667007245030652e-05,
      "loss": 1.8181,
      "step": 8050
    },
    {
      "epoch": 0.18341942971576816,
      "grad_norm": 42.5,
      "learning_rate": 1.6665428199888542e-05,
      "loss": 0.6636,
      "step": 8060
    },
    {
      "epoch": 0.18364699724643288,
      "grad_norm": 0.0072021484375,
      "learning_rate": 1.6660783949470556e-05,
      "loss": 0.5124,
      "step": 8070
    },
    {
      "epoch": 0.1838745647770976,
      "grad_norm": 8.3125,
      "learning_rate": 1.6656139699052573e-05,
      "loss": 0.3554,
      "step": 8080
    },
    {
      "epoch": 0.18410213230776232,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.6651495448634593e-05,
      "loss": 0.9731,
      "step": 8090
    },
    {
      "epoch": 0.18432969983842706,
      "grad_norm": 1.7421875,
      "learning_rate": 1.664685119821661e-05,
      "loss": 0.0524,
      "step": 8100
    },
    {
      "epoch": 0.18455726736909178,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.6642206947798627e-05,
      "loss": 1.8723,
      "step": 8110
    },
    {
      "epoch": 0.1847848348997565,
      "grad_norm": 21.25,
      "learning_rate": 1.6637562697380644e-05,
      "loss": 0.5502,
      "step": 8120
    },
    {
      "epoch": 0.18501240243042122,
      "grad_norm": 93.5,
      "learning_rate": 1.663291844696266e-05,
      "loss": 1.0929,
      "step": 8130
    },
    {
      "epoch": 0.18523996996108597,
      "grad_norm": 124.0,
      "learning_rate": 1.6628274196544678e-05,
      "loss": 1.3083,
      "step": 8140
    },
    {
      "epoch": 0.18546753749175068,
      "grad_norm": 205.0,
      "learning_rate": 1.6623629946126695e-05,
      "loss": 0.9293,
      "step": 8150
    },
    {
      "epoch": 0.1856951050224154,
      "grad_norm": 189.0,
      "learning_rate": 1.6618985695708712e-05,
      "loss": 0.4026,
      "step": 8160
    },
    {
      "epoch": 0.18592267255308012,
      "grad_norm": 8.6875,
      "learning_rate": 1.6614341445290733e-05,
      "loss": 1.103,
      "step": 8170
    },
    {
      "epoch": 0.18615024008374484,
      "grad_norm": 0.00024318695068359375,
      "learning_rate": 1.660969719487275e-05,
      "loss": 1.4376,
      "step": 8180
    },
    {
      "epoch": 0.18637780761440959,
      "grad_norm": 35.75,
      "learning_rate": 1.6605052944454767e-05,
      "loss": 1.0469,
      "step": 8190
    },
    {
      "epoch": 0.1866053751450743,
      "grad_norm": 272.0,
      "learning_rate": 1.6600408694036784e-05,
      "loss": 0.958,
      "step": 8200
    },
    {
      "epoch": 0.18683294267573902,
      "grad_norm": 77.0,
      "learning_rate": 1.65957644436188e-05,
      "loss": 0.5895,
      "step": 8210
    },
    {
      "epoch": 0.18706051020640374,
      "grad_norm": 54.0,
      "learning_rate": 1.6591120193200818e-05,
      "loss": 0.8982,
      "step": 8220
    },
    {
      "epoch": 0.1872880777370685,
      "grad_norm": 278.0,
      "learning_rate": 1.6586475942782835e-05,
      "loss": 0.7707,
      "step": 8230
    },
    {
      "epoch": 0.1875156452677332,
      "grad_norm": 0.034912109375,
      "learning_rate": 1.6581831692364852e-05,
      "loss": 0.5619,
      "step": 8240
    },
    {
      "epoch": 0.18774321279839792,
      "grad_norm": 0.375,
      "learning_rate": 1.6577187441946873e-05,
      "loss": 0.1071,
      "step": 8250
    },
    {
      "epoch": 0.18797078032906264,
      "grad_norm": 67.5,
      "learning_rate": 1.657254319152889e-05,
      "loss": 1.6867,
      "step": 8260
    },
    {
      "epoch": 0.18819834785972736,
      "grad_norm": 0.046630859375,
      "learning_rate": 1.6567898941110907e-05,
      "loss": 0.9262,
      "step": 8270
    },
    {
      "epoch": 0.1884259153903921,
      "grad_norm": 0.0033111572265625,
      "learning_rate": 1.6563254690692924e-05,
      "loss": 0.8618,
      "step": 8280
    },
    {
      "epoch": 0.18865348292105683,
      "grad_norm": 2.140625,
      "learning_rate": 1.655861044027494e-05,
      "loss": 0.97,
      "step": 8290
    },
    {
      "epoch": 0.18888105045172154,
      "grad_norm": 226.0,
      "learning_rate": 1.6553966189856958e-05,
      "loss": 0.3739,
      "step": 8300
    },
    {
      "epoch": 0.18910861798238626,
      "grad_norm": 0.000640869140625,
      "learning_rate": 1.6549321939438975e-05,
      "loss": 0.6889,
      "step": 8310
    },
    {
      "epoch": 0.189336185513051,
      "grad_norm": 14.4375,
      "learning_rate": 1.6544677689020992e-05,
      "loss": 0.4425,
      "step": 8320
    },
    {
      "epoch": 0.18956375304371573,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.6540033438603012e-05,
      "loss": 0.724,
      "step": 8330
    },
    {
      "epoch": 0.18979132057438045,
      "grad_norm": 13.4375,
      "learning_rate": 1.653538918818503e-05,
      "loss": 0.4163,
      "step": 8340
    },
    {
      "epoch": 0.19001888810504516,
      "grad_norm": 0.0031890869140625,
      "learning_rate": 1.6530744937767043e-05,
      "loss": 1.1748,
      "step": 8350
    },
    {
      "epoch": 0.1902464556357099,
      "grad_norm": 106.5,
      "learning_rate": 1.6526100687349064e-05,
      "loss": 0.6386,
      "step": 8360
    },
    {
      "epoch": 0.19047402316637463,
      "grad_norm": 29.25,
      "learning_rate": 1.652145643693108e-05,
      "loss": 1.3416,
      "step": 8370
    },
    {
      "epoch": 0.19070159069703935,
      "grad_norm": 4.78125,
      "learning_rate": 1.6516812186513098e-05,
      "loss": 0.8009,
      "step": 8380
    },
    {
      "epoch": 0.19092915822770407,
      "grad_norm": 0.01519775390625,
      "learning_rate": 1.6512167936095115e-05,
      "loss": 0.496,
      "step": 8390
    },
    {
      "epoch": 0.19115672575836878,
      "grad_norm": 53.75,
      "learning_rate": 1.6507523685677135e-05,
      "loss": 0.1985,
      "step": 8400
    },
    {
      "epoch": 0.19138429328903353,
      "grad_norm": 9.125,
      "learning_rate": 1.6502879435259152e-05,
      "loss": 0.4983,
      "step": 8410
    },
    {
      "epoch": 0.19161186081969825,
      "grad_norm": 360.0,
      "learning_rate": 1.6498235184841166e-05,
      "loss": 0.8691,
      "step": 8420
    },
    {
      "epoch": 0.19183942835036297,
      "grad_norm": 148.0,
      "learning_rate": 1.6493590934423186e-05,
      "loss": 1.0504,
      "step": 8430
    },
    {
      "epoch": 0.19206699588102769,
      "grad_norm": 0.0283203125,
      "learning_rate": 1.6488946684005203e-05,
      "loss": 1.3706,
      "step": 8440
    },
    {
      "epoch": 0.19229456341169243,
      "grad_norm": 77.5,
      "learning_rate": 1.648430243358722e-05,
      "loss": 0.9144,
      "step": 8450
    },
    {
      "epoch": 0.19252213094235715,
      "grad_norm": 92.5,
      "learning_rate": 1.6479658183169238e-05,
      "loss": 1.2851,
      "step": 8460
    },
    {
      "epoch": 0.19274969847302187,
      "grad_norm": 0.0050048828125,
      "learning_rate": 1.6475013932751255e-05,
      "loss": 0.7395,
      "step": 8470
    },
    {
      "epoch": 0.1929772660036866,
      "grad_norm": 171.0,
      "learning_rate": 1.6470369682333275e-05,
      "loss": 0.8143,
      "step": 8480
    },
    {
      "epoch": 0.1932048335343513,
      "grad_norm": 83.0,
      "learning_rate": 1.646572543191529e-05,
      "loss": 0.8697,
      "step": 8490
    },
    {
      "epoch": 0.19343240106501605,
      "grad_norm": 187.0,
      "learning_rate": 1.6461081181497306e-05,
      "loss": 0.9257,
      "step": 8500
    },
    {
      "epoch": 0.19365996859568077,
      "grad_norm": 260.0,
      "learning_rate": 1.6456436931079326e-05,
      "loss": 0.9098,
      "step": 8510
    },
    {
      "epoch": 0.1938875361263455,
      "grad_norm": 6.09375,
      "learning_rate": 1.6451792680661343e-05,
      "loss": 1.1601,
      "step": 8520
    },
    {
      "epoch": 0.1941151036570102,
      "grad_norm": 136.0,
      "learning_rate": 1.644714843024336e-05,
      "loss": 1.0097,
      "step": 8530
    },
    {
      "epoch": 0.19434267118767495,
      "grad_norm": 3.90625,
      "learning_rate": 1.6442504179825377e-05,
      "loss": 0.9386,
      "step": 8540
    },
    {
      "epoch": 0.19457023871833967,
      "grad_norm": 0.349609375,
      "learning_rate": 1.6437859929407394e-05,
      "loss": 1.0312,
      "step": 8550
    },
    {
      "epoch": 0.1947978062490044,
      "grad_norm": 0.1025390625,
      "learning_rate": 1.643321567898941e-05,
      "loss": 0.3504,
      "step": 8560
    },
    {
      "epoch": 0.1950253737796691,
      "grad_norm": 154.0,
      "learning_rate": 1.642857142857143e-05,
      "loss": 0.6983,
      "step": 8570
    },
    {
      "epoch": 0.19525294131033386,
      "grad_norm": 2.84375,
      "learning_rate": 1.6423927178153446e-05,
      "loss": 0.276,
      "step": 8580
    },
    {
      "epoch": 0.19548050884099857,
      "grad_norm": 74.0,
      "learning_rate": 1.6419282927735466e-05,
      "loss": 0.3187,
      "step": 8590
    },
    {
      "epoch": 0.1957080763716633,
      "grad_norm": 242.0,
      "learning_rate": 1.6414638677317483e-05,
      "loss": 0.971,
      "step": 8600
    },
    {
      "epoch": 0.195935643902328,
      "grad_norm": 140.0,
      "learning_rate": 1.64099944268995e-05,
      "loss": 0.7978,
      "step": 8610
    },
    {
      "epoch": 0.19616321143299273,
      "grad_norm": 0.0201416015625,
      "learning_rate": 1.6405350176481517e-05,
      "loss": 0.8708,
      "step": 8620
    },
    {
      "epoch": 0.19639077896365748,
      "grad_norm": 24.625,
      "learning_rate": 1.6400705926063534e-05,
      "loss": 0.2596,
      "step": 8630
    },
    {
      "epoch": 0.1966183464943222,
      "grad_norm": 0.0004138946533203125,
      "learning_rate": 1.639606167564555e-05,
      "loss": 1.9555,
      "step": 8640
    },
    {
      "epoch": 0.1968459140249869,
      "grad_norm": 0.466796875,
      "learning_rate": 1.639141742522757e-05,
      "loss": 0.9995,
      "step": 8650
    },
    {
      "epoch": 0.19707348155565163,
      "grad_norm": 1.6328125,
      "learning_rate": 1.6386773174809585e-05,
      "loss": 0.8859,
      "step": 8660
    },
    {
      "epoch": 0.19730104908631638,
      "grad_norm": 57.5,
      "learning_rate": 1.6382128924391606e-05,
      "loss": 1.0278,
      "step": 8670
    },
    {
      "epoch": 0.1975286166169811,
      "grad_norm": 4.25,
      "learning_rate": 1.6377484673973623e-05,
      "loss": 1.0874,
      "step": 8680
    },
    {
      "epoch": 0.1977561841476458,
      "grad_norm": 0.00125885009765625,
      "learning_rate": 1.637284042355564e-05,
      "loss": 1.6119,
      "step": 8690
    },
    {
      "epoch": 0.19798375167831053,
      "grad_norm": 22.875,
      "learning_rate": 1.6368196173137657e-05,
      "loss": 0.6031,
      "step": 8700
    },
    {
      "epoch": 0.19821131920897525,
      "grad_norm": 0.0023345947265625,
      "learning_rate": 1.6363551922719674e-05,
      "loss": 0.6826,
      "step": 8710
    },
    {
      "epoch": 0.19843888673964,
      "grad_norm": 0.76171875,
      "learning_rate": 1.635890767230169e-05,
      "loss": 0.9791,
      "step": 8720
    },
    {
      "epoch": 0.19866645427030472,
      "grad_norm": 81.5,
      "learning_rate": 1.6354263421883708e-05,
      "loss": 0.411,
      "step": 8730
    },
    {
      "epoch": 0.19889402180096943,
      "grad_norm": 0.0023956298828125,
      "learning_rate": 1.634961917146573e-05,
      "loss": 0.8363,
      "step": 8740
    },
    {
      "epoch": 0.19912158933163415,
      "grad_norm": 203.0,
      "learning_rate": 1.6344974921047746e-05,
      "loss": 1.0741,
      "step": 8750
    },
    {
      "epoch": 0.1993491568622989,
      "grad_norm": 69.0,
      "learning_rate": 1.6340330670629763e-05,
      "loss": 0.4443,
      "step": 8760
    },
    {
      "epoch": 0.19957672439296362,
      "grad_norm": 0.125,
      "learning_rate": 1.633568642021178e-05,
      "loss": 0.7039,
      "step": 8770
    },
    {
      "epoch": 0.19980429192362834,
      "grad_norm": 0.0390625,
      "learning_rate": 1.6331042169793797e-05,
      "loss": 0.2728,
      "step": 8780
    },
    {
      "epoch": 0.20003185945429305,
      "grad_norm": 85.0,
      "learning_rate": 1.6326397919375814e-05,
      "loss": 0.7663,
      "step": 8790
    },
    {
      "epoch": 0.20025942698495777,
      "grad_norm": 60.75,
      "learning_rate": 1.632175366895783e-05,
      "loss": 0.997,
      "step": 8800
    },
    {
      "epoch": 0.20048699451562252,
      "grad_norm": 103.5,
      "learning_rate": 1.6317109418539848e-05,
      "loss": 0.8035,
      "step": 8810
    },
    {
      "epoch": 0.20071456204628724,
      "grad_norm": 0.71875,
      "learning_rate": 1.631246516812187e-05,
      "loss": 1.3411,
      "step": 8820
    },
    {
      "epoch": 0.20094212957695196,
      "grad_norm": 0.0030059814453125,
      "learning_rate": 1.6307820917703886e-05,
      "loss": 1.5105,
      "step": 8830
    },
    {
      "epoch": 0.20116969710761667,
      "grad_norm": 23.5,
      "learning_rate": 1.63031766672859e-05,
      "loss": 1.1315,
      "step": 8840
    },
    {
      "epoch": 0.20139726463828142,
      "grad_norm": 79.0,
      "learning_rate": 1.629853241686792e-05,
      "loss": 0.5367,
      "step": 8850
    },
    {
      "epoch": 0.20162483216894614,
      "grad_norm": 348.0,
      "learning_rate": 1.6293888166449937e-05,
      "loss": 0.6799,
      "step": 8860
    },
    {
      "epoch": 0.20185239969961086,
      "grad_norm": 230.0,
      "learning_rate": 1.6289243916031954e-05,
      "loss": 0.7966,
      "step": 8870
    },
    {
      "epoch": 0.20207996723027558,
      "grad_norm": 118.0,
      "learning_rate": 1.628459966561397e-05,
      "loss": 0.3831,
      "step": 8880
    },
    {
      "epoch": 0.20230753476094032,
      "grad_norm": 118.0,
      "learning_rate": 1.6279955415195988e-05,
      "loss": 0.4284,
      "step": 8890
    },
    {
      "epoch": 0.20253510229160504,
      "grad_norm": 245.0,
      "learning_rate": 1.627531116477801e-05,
      "loss": 0.8387,
      "step": 8900
    },
    {
      "epoch": 0.20276266982226976,
      "grad_norm": 904.0,
      "learning_rate": 1.6270666914360022e-05,
      "loss": 1.3238,
      "step": 8910
    },
    {
      "epoch": 0.20299023735293448,
      "grad_norm": 198.0,
      "learning_rate": 1.626602266394204e-05,
      "loss": 0.7154,
      "step": 8920
    },
    {
      "epoch": 0.2032178048835992,
      "grad_norm": 23.375,
      "learning_rate": 1.626137841352406e-05,
      "loss": 0.9791,
      "step": 8930
    },
    {
      "epoch": 0.20344537241426394,
      "grad_norm": 320.0,
      "learning_rate": 1.6256734163106077e-05,
      "loss": 0.8169,
      "step": 8940
    },
    {
      "epoch": 0.20367293994492866,
      "grad_norm": 712.0,
      "learning_rate": 1.6252089912688094e-05,
      "loss": 1.1222,
      "step": 8950
    },
    {
      "epoch": 0.20390050747559338,
      "grad_norm": 276.0,
      "learning_rate": 1.624744566227011e-05,
      "loss": 1.2758,
      "step": 8960
    },
    {
      "epoch": 0.2041280750062581,
      "grad_norm": 286.0,
      "learning_rate": 1.6242801411852128e-05,
      "loss": 0.5614,
      "step": 8970
    },
    {
      "epoch": 0.20435564253692284,
      "grad_norm": 494.0,
      "learning_rate": 1.6238157161434145e-05,
      "loss": 1.2251,
      "step": 8980
    },
    {
      "epoch": 0.20458321006758756,
      "grad_norm": 7.875,
      "learning_rate": 1.6233512911016162e-05,
      "loss": 1.6947,
      "step": 8990
    },
    {
      "epoch": 0.20481077759825228,
      "grad_norm": 1.1328125,
      "learning_rate": 1.622886866059818e-05,
      "loss": 1.4567,
      "step": 9000
    },
    {
      "epoch": 0.205038345128917,
      "grad_norm": 56.75,
      "learning_rate": 1.62242244101802e-05,
      "loss": 0.6464,
      "step": 9010
    },
    {
      "epoch": 0.20526591265958172,
      "grad_norm": 314.0,
      "learning_rate": 1.6219580159762216e-05,
      "loss": 0.6796,
      "step": 9020
    },
    {
      "epoch": 0.20549348019024646,
      "grad_norm": 0.031005859375,
      "learning_rate": 1.6214935909344233e-05,
      "loss": 0.5361,
      "step": 9030
    },
    {
      "epoch": 0.20572104772091118,
      "grad_norm": 0.033447265625,
      "learning_rate": 1.621029165892625e-05,
      "loss": 0.9321,
      "step": 9040
    },
    {
      "epoch": 0.2059486152515759,
      "grad_norm": 60.5,
      "learning_rate": 1.6205647408508268e-05,
      "loss": 1.2415,
      "step": 9050
    },
    {
      "epoch": 0.20617618278224062,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.6201003158090285e-05,
      "loss": 0.298,
      "step": 9060
    },
    {
      "epoch": 0.20640375031290537,
      "grad_norm": 1688.0,
      "learning_rate": 1.61963589076723e-05,
      "loss": 1.1027,
      "step": 9070
    },
    {
      "epoch": 0.20663131784357008,
      "grad_norm": 0.84375,
      "learning_rate": 1.6191714657254322e-05,
      "loss": 0.5566,
      "step": 9080
    },
    {
      "epoch": 0.2068588853742348,
      "grad_norm": 472.0,
      "learning_rate": 1.618707040683634e-05,
      "loss": 1.0176,
      "step": 9090
    },
    {
      "epoch": 0.20708645290489952,
      "grad_norm": 97.0,
      "learning_rate": 1.6182426156418356e-05,
      "loss": 1.3551,
      "step": 9100
    },
    {
      "epoch": 0.20731402043556427,
      "grad_norm": 4.21875,
      "learning_rate": 1.6177781906000373e-05,
      "loss": 0.6326,
      "step": 9110
    },
    {
      "epoch": 0.20754158796622899,
      "grad_norm": 10.875,
      "learning_rate": 1.617313765558239e-05,
      "loss": 1.3762,
      "step": 9120
    },
    {
      "epoch": 0.2077691554968937,
      "grad_norm": 41.25,
      "learning_rate": 1.6168493405164407e-05,
      "loss": 1.0091,
      "step": 9130
    },
    {
      "epoch": 0.20799672302755842,
      "grad_norm": 110.5,
      "learning_rate": 1.6163849154746424e-05,
      "loss": 1.857,
      "step": 9140
    },
    {
      "epoch": 0.20822429055822314,
      "grad_norm": 648.0,
      "learning_rate": 1.615920490432844e-05,
      "loss": 1.0509,
      "step": 9150
    },
    {
      "epoch": 0.2084518580888879,
      "grad_norm": 10.75,
      "learning_rate": 1.6154560653910462e-05,
      "loss": 1.3496,
      "step": 9160
    },
    {
      "epoch": 0.2086794256195526,
      "grad_norm": 0.0033721923828125,
      "learning_rate": 1.614991640349248e-05,
      "loss": 1.3452,
      "step": 9170
    },
    {
      "epoch": 0.20890699315021732,
      "grad_norm": 0.035888671875,
      "learning_rate": 1.6145272153074496e-05,
      "loss": 1.2428,
      "step": 9180
    },
    {
      "epoch": 0.20913456068088204,
      "grad_norm": 133.0,
      "learning_rate": 1.6140627902656513e-05,
      "loss": 0.5211,
      "step": 9190
    },
    {
      "epoch": 0.2093621282115468,
      "grad_norm": 89.5,
      "learning_rate": 1.613598365223853e-05,
      "loss": 0.8605,
      "step": 9200
    },
    {
      "epoch": 0.2095896957422115,
      "grad_norm": 11.3125,
      "learning_rate": 1.6131339401820547e-05,
      "loss": 0.5389,
      "step": 9210
    },
    {
      "epoch": 0.20981726327287623,
      "grad_norm": 99.0,
      "learning_rate": 1.6126695151402564e-05,
      "loss": 0.8267,
      "step": 9220
    },
    {
      "epoch": 0.21004483080354094,
      "grad_norm": 0.546875,
      "learning_rate": 1.612205090098458e-05,
      "loss": 0.6988,
      "step": 9230
    },
    {
      "epoch": 0.21027239833420566,
      "grad_norm": 0.033935546875,
      "learning_rate": 1.6117406650566602e-05,
      "loss": 0.8314,
      "step": 9240
    },
    {
      "epoch": 0.2104999658648704,
      "grad_norm": 0.0322265625,
      "learning_rate": 1.611276240014862e-05,
      "loss": 0.4952,
      "step": 9250
    },
    {
      "epoch": 0.21072753339553513,
      "grad_norm": 113.5,
      "learning_rate": 1.6108118149730632e-05,
      "loss": 0.3416,
      "step": 9260
    },
    {
      "epoch": 0.21095510092619985,
      "grad_norm": 69.5,
      "learning_rate": 1.6103473899312653e-05,
      "loss": 1.0139,
      "step": 9270
    },
    {
      "epoch": 0.21118266845686456,
      "grad_norm": 0.0024566650390625,
      "learning_rate": 1.609882964889467e-05,
      "loss": 0.3243,
      "step": 9280
    },
    {
      "epoch": 0.2114102359875293,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6094185398476687e-05,
      "loss": 0.7845,
      "step": 9290
    },
    {
      "epoch": 0.21163780351819403,
      "grad_norm": 101.5,
      "learning_rate": 1.6089541148058704e-05,
      "loss": 1.0799,
      "step": 9300
    },
    {
      "epoch": 0.21186537104885875,
      "grad_norm": 148.0,
      "learning_rate": 1.608489689764072e-05,
      "loss": 0.8988,
      "step": 9310
    },
    {
      "epoch": 0.21209293857952347,
      "grad_norm": 189.0,
      "learning_rate": 1.608025264722274e-05,
      "loss": 1.0435,
      "step": 9320
    },
    {
      "epoch": 0.2123205061101882,
      "grad_norm": 0.3828125,
      "learning_rate": 1.6075608396804755e-05,
      "loss": 0.9045,
      "step": 9330
    },
    {
      "epoch": 0.21254807364085293,
      "grad_norm": 5.0,
      "learning_rate": 1.6070964146386772e-05,
      "loss": 0.3257,
      "step": 9340
    },
    {
      "epoch": 0.21277564117151765,
      "grad_norm": 148.0,
      "learning_rate": 1.6066319895968793e-05,
      "loss": 1.8471,
      "step": 9350
    },
    {
      "epoch": 0.21300320870218237,
      "grad_norm": 202.0,
      "learning_rate": 1.606167564555081e-05,
      "loss": 1.1115,
      "step": 9360
    },
    {
      "epoch": 0.21323077623284709,
      "grad_norm": 124.5,
      "learning_rate": 1.6057031395132827e-05,
      "loss": 1.1629,
      "step": 9370
    },
    {
      "epoch": 0.21345834376351183,
      "grad_norm": 588.0,
      "learning_rate": 1.6052387144714844e-05,
      "loss": 1.0027,
      "step": 9380
    },
    {
      "epoch": 0.21368591129417655,
      "grad_norm": 62.25,
      "learning_rate": 1.6047742894296864e-05,
      "loss": 0.9935,
      "step": 9390
    },
    {
      "epoch": 0.21391347882484127,
      "grad_norm": 87.5,
      "learning_rate": 1.6043098643878878e-05,
      "loss": 1.2058,
      "step": 9400
    },
    {
      "epoch": 0.214141046355506,
      "grad_norm": 422.0,
      "learning_rate": 1.6038454393460895e-05,
      "loss": 0.9817,
      "step": 9410
    },
    {
      "epoch": 0.21436861388617073,
      "grad_norm": 6.3125,
      "learning_rate": 1.6033810143042916e-05,
      "loss": 1.3332,
      "step": 9420
    },
    {
      "epoch": 0.21459618141683545,
      "grad_norm": 0.005340576171875,
      "learning_rate": 1.6029165892624933e-05,
      "loss": 0.4914,
      "step": 9430
    },
    {
      "epoch": 0.21482374894750017,
      "grad_norm": 6.84375,
      "learning_rate": 1.602452164220695e-05,
      "loss": 0.9394,
      "step": 9440
    },
    {
      "epoch": 0.2150513164781649,
      "grad_norm": 0.0255126953125,
      "learning_rate": 1.6019877391788967e-05,
      "loss": 0.455,
      "step": 9450
    },
    {
      "epoch": 0.2152788840088296,
      "grad_norm": 106.0,
      "learning_rate": 1.6015233141370984e-05,
      "loss": 0.3398,
      "step": 9460
    },
    {
      "epoch": 0.21550645153949435,
      "grad_norm": 476.0,
      "learning_rate": 1.6010588890953e-05,
      "loss": 1.4834,
      "step": 9470
    },
    {
      "epoch": 0.21573401907015907,
      "grad_norm": 0.083984375,
      "learning_rate": 1.6005944640535018e-05,
      "loss": 1.0208,
      "step": 9480
    },
    {
      "epoch": 0.2159615866008238,
      "grad_norm": 182.0,
      "learning_rate": 1.6001300390117035e-05,
      "loss": 1.2132,
      "step": 9490
    },
    {
      "epoch": 0.2161891541314885,
      "grad_norm": 7.5,
      "learning_rate": 1.5996656139699055e-05,
      "loss": 0.8619,
      "step": 9500
    },
    {
      "epoch": 0.21641672166215326,
      "grad_norm": 0.000774383544921875,
      "learning_rate": 1.5992011889281072e-05,
      "loss": 0.7331,
      "step": 9510
    },
    {
      "epoch": 0.21664428919281797,
      "grad_norm": 65.5,
      "learning_rate": 1.598736763886309e-05,
      "loss": 0.7816,
      "step": 9520
    },
    {
      "epoch": 0.2168718567234827,
      "grad_norm": 6.125,
      "learning_rate": 1.5982723388445107e-05,
      "loss": 0.4981,
      "step": 9530
    },
    {
      "epoch": 0.2170994242541474,
      "grad_norm": 31.0,
      "learning_rate": 1.5978079138027124e-05,
      "loss": 0.3798,
      "step": 9540
    },
    {
      "epoch": 0.21732699178481216,
      "grad_norm": 0.0390625,
      "learning_rate": 1.597343488760914e-05,
      "loss": 1.3404,
      "step": 9550
    },
    {
      "epoch": 0.21755455931547688,
      "grad_norm": 147.0,
      "learning_rate": 1.5968790637191158e-05,
      "loss": 0.8008,
      "step": 9560
    },
    {
      "epoch": 0.2177821268461416,
      "grad_norm": 87.5,
      "learning_rate": 1.5964146386773175e-05,
      "loss": 1.0854,
      "step": 9570
    },
    {
      "epoch": 0.2180096943768063,
      "grad_norm": 0.019287109375,
      "learning_rate": 1.5959502136355195e-05,
      "loss": 0.4928,
      "step": 9580
    },
    {
      "epoch": 0.21823726190747103,
      "grad_norm": 286.0,
      "learning_rate": 1.5954857885937212e-05,
      "loss": 2.1381,
      "step": 9590
    },
    {
      "epoch": 0.21846482943813578,
      "grad_norm": 187.0,
      "learning_rate": 1.595021363551923e-05,
      "loss": 0.2718,
      "step": 9600
    },
    {
      "epoch": 0.2186923969688005,
      "grad_norm": 0.0037689208984375,
      "learning_rate": 1.5945569385101246e-05,
      "loss": 1.0599,
      "step": 9610
    },
    {
      "epoch": 0.2189199644994652,
      "grad_norm": 180.0,
      "learning_rate": 1.5940925134683263e-05,
      "loss": 0.2528,
      "step": 9620
    },
    {
      "epoch": 0.21914753203012993,
      "grad_norm": 0.0162353515625,
      "learning_rate": 1.593628088426528e-05,
      "loss": 0.5866,
      "step": 9630
    },
    {
      "epoch": 0.21937509956079468,
      "grad_norm": 20.625,
      "learning_rate": 1.5931636633847298e-05,
      "loss": 0.4141,
      "step": 9640
    },
    {
      "epoch": 0.2196026670914594,
      "grad_norm": 165.0,
      "learning_rate": 1.5926992383429315e-05,
      "loss": 0.3983,
      "step": 9650
    },
    {
      "epoch": 0.21983023462212412,
      "grad_norm": 0.302734375,
      "learning_rate": 1.5922348133011335e-05,
      "loss": 0.9579,
      "step": 9660
    },
    {
      "epoch": 0.22005780215278883,
      "grad_norm": 85.0,
      "learning_rate": 1.5917703882593352e-05,
      "loss": 1.6846,
      "step": 9670
    },
    {
      "epoch": 0.22028536968345355,
      "grad_norm": 47.0,
      "learning_rate": 1.5913059632175366e-05,
      "loss": 0.6325,
      "step": 9680
    },
    {
      "epoch": 0.2205129372141183,
      "grad_norm": 3.171875,
      "learning_rate": 1.5908415381757386e-05,
      "loss": 1.1963,
      "step": 9690
    },
    {
      "epoch": 0.22074050474478302,
      "grad_norm": 294.0,
      "learning_rate": 1.5903771131339403e-05,
      "loss": 1.2549,
      "step": 9700
    },
    {
      "epoch": 0.22096807227544774,
      "grad_norm": 0.0233154296875,
      "learning_rate": 1.589912688092142e-05,
      "loss": 0.396,
      "step": 9710
    },
    {
      "epoch": 0.22119563980611245,
      "grad_norm": 480.0,
      "learning_rate": 1.5894482630503437e-05,
      "loss": 0.9566,
      "step": 9720
    },
    {
      "epoch": 0.2214232073367772,
      "grad_norm": 87.5,
      "learning_rate": 1.5889838380085458e-05,
      "loss": 0.6978,
      "step": 9730
    },
    {
      "epoch": 0.22165077486744192,
      "grad_norm": 74.5,
      "learning_rate": 1.5885194129667475e-05,
      "loss": 1.0224,
      "step": 9740
    },
    {
      "epoch": 0.22187834239810664,
      "grad_norm": 108.5,
      "learning_rate": 1.588054987924949e-05,
      "loss": 1.4368,
      "step": 9750
    },
    {
      "epoch": 0.22210590992877136,
      "grad_norm": 69.0,
      "learning_rate": 1.5875905628831506e-05,
      "loss": 0.7704,
      "step": 9760
    },
    {
      "epoch": 0.22233347745943607,
      "grad_norm": 370.0,
      "learning_rate": 1.5871261378413526e-05,
      "loss": 1.8142,
      "step": 9770
    },
    {
      "epoch": 0.22256104499010082,
      "grad_norm": 32.25,
      "learning_rate": 1.5866617127995543e-05,
      "loss": 0.7583,
      "step": 9780
    },
    {
      "epoch": 0.22278861252076554,
      "grad_norm": 203.0,
      "learning_rate": 1.586197287757756e-05,
      "loss": 0.9393,
      "step": 9790
    },
    {
      "epoch": 0.22301618005143026,
      "grad_norm": 438.0,
      "learning_rate": 1.5857328627159577e-05,
      "loss": 1.1138,
      "step": 9800
    },
    {
      "epoch": 0.22324374758209498,
      "grad_norm": 79.0,
      "learning_rate": 1.5852684376741598e-05,
      "loss": 0.5253,
      "step": 9810
    },
    {
      "epoch": 0.22347131511275972,
      "grad_norm": 46.25,
      "learning_rate": 1.584804012632361e-05,
      "loss": 1.057,
      "step": 9820
    },
    {
      "epoch": 0.22369888264342444,
      "grad_norm": 33.25,
      "learning_rate": 1.584339587590563e-05,
      "loss": 0.8265,
      "step": 9830
    },
    {
      "epoch": 0.22392645017408916,
      "grad_norm": 80.5,
      "learning_rate": 1.583875162548765e-05,
      "loss": 0.8869,
      "step": 9840
    },
    {
      "epoch": 0.22415401770475388,
      "grad_norm": 334.0,
      "learning_rate": 1.5834107375069666e-05,
      "loss": 1.4824,
      "step": 9850
    },
    {
      "epoch": 0.22438158523541862,
      "grad_norm": 346.0,
      "learning_rate": 1.5829463124651683e-05,
      "loss": 0.4892,
      "step": 9860
    },
    {
      "epoch": 0.22460915276608334,
      "grad_norm": 89.5,
      "learning_rate": 1.58248188742337e-05,
      "loss": 1.0218,
      "step": 9870
    },
    {
      "epoch": 0.22483672029674806,
      "grad_norm": 0.0869140625,
      "learning_rate": 1.5820174623815717e-05,
      "loss": 0.4897,
      "step": 9880
    },
    {
      "epoch": 0.22506428782741278,
      "grad_norm": 106.5,
      "learning_rate": 1.5815530373397734e-05,
      "loss": 1.1623,
      "step": 9890
    },
    {
      "epoch": 0.2252918553580775,
      "grad_norm": 71.5,
      "learning_rate": 1.581088612297975e-05,
      "loss": 0.3532,
      "step": 9900
    },
    {
      "epoch": 0.22551942288874224,
      "grad_norm": 75.5,
      "learning_rate": 1.5806241872561768e-05,
      "loss": 1.0138,
      "step": 9910
    },
    {
      "epoch": 0.22574699041940696,
      "grad_norm": 5.40625,
      "learning_rate": 1.580159762214379e-05,
      "loss": 1.3322,
      "step": 9920
    },
    {
      "epoch": 0.22597455795007168,
      "grad_norm": 0.000904083251953125,
      "learning_rate": 1.5796953371725806e-05,
      "loss": 0.4156,
      "step": 9930
    },
    {
      "epoch": 0.2262021254807364,
      "grad_norm": 0.1435546875,
      "learning_rate": 1.5792309121307823e-05,
      "loss": 0.457,
      "step": 9940
    },
    {
      "epoch": 0.22642969301140115,
      "grad_norm": 0.003265380859375,
      "learning_rate": 1.578766487088984e-05,
      "loss": 0.3782,
      "step": 9950
    },
    {
      "epoch": 0.22665726054206586,
      "grad_norm": 476.0,
      "learning_rate": 1.5783020620471857e-05,
      "loss": 0.7299,
      "step": 9960
    },
    {
      "epoch": 0.22688482807273058,
      "grad_norm": 0.1044921875,
      "learning_rate": 1.5778376370053874e-05,
      "loss": 1.0067,
      "step": 9970
    },
    {
      "epoch": 0.2271123956033953,
      "grad_norm": 3.640625,
      "learning_rate": 1.577373211963589e-05,
      "loss": 0.5289,
      "step": 9980
    },
    {
      "epoch": 0.22733996313406002,
      "grad_norm": 105.0,
      "learning_rate": 1.5769087869217908e-05,
      "loss": 0.6029,
      "step": 9990
    },
    {
      "epoch": 0.22756753066472477,
      "grad_norm": 178.0,
      "learning_rate": 1.576444361879993e-05,
      "loss": 0.7951,
      "step": 10000
    },
    {
      "epoch": 0.22779509819538948,
      "grad_norm": 27.125,
      "learning_rate": 1.5759799368381945e-05,
      "loss": 0.228,
      "step": 10010
    },
    {
      "epoch": 0.2280226657260542,
      "grad_norm": 0.0272216796875,
      "learning_rate": 1.5755155117963963e-05,
      "loss": 1.0469,
      "step": 10020
    },
    {
      "epoch": 0.22825023325671892,
      "grad_norm": 215.0,
      "learning_rate": 1.575051086754598e-05,
      "loss": 1.8567,
      "step": 10030
    },
    {
      "epoch": 0.22847780078738367,
      "grad_norm": 70.5,
      "learning_rate": 1.5745866617127997e-05,
      "loss": 0.7132,
      "step": 10040
    },
    {
      "epoch": 0.22870536831804839,
      "grad_norm": 201.0,
      "learning_rate": 1.5741222366710014e-05,
      "loss": 0.8474,
      "step": 10050
    },
    {
      "epoch": 0.2289329358487131,
      "grad_norm": 0.0233154296875,
      "learning_rate": 1.573657811629203e-05,
      "loss": 0.4694,
      "step": 10060
    },
    {
      "epoch": 0.22916050337937782,
      "grad_norm": 180.0,
      "learning_rate": 1.5731933865874048e-05,
      "loss": 0.7782,
      "step": 10070
    },
    {
      "epoch": 0.22938807091004257,
      "grad_norm": 151.0,
      "learning_rate": 1.5727289615456068e-05,
      "loss": 0.375,
      "step": 10080
    },
    {
      "epoch": 0.2296156384407073,
      "grad_norm": 34.25,
      "learning_rate": 1.5722645365038085e-05,
      "loss": 0.9511,
      "step": 10090
    },
    {
      "epoch": 0.229843205971372,
      "grad_norm": 0.08935546875,
      "learning_rate": 1.57180011146201e-05,
      "loss": 0.128,
      "step": 10100
    },
    {
      "epoch": 0.23007077350203672,
      "grad_norm": 45.5,
      "learning_rate": 1.571335686420212e-05,
      "loss": 0.7432,
      "step": 10110
    },
    {
      "epoch": 0.23029834103270144,
      "grad_norm": 114.0,
      "learning_rate": 1.5708712613784136e-05,
      "loss": 1.0422,
      "step": 10120
    },
    {
      "epoch": 0.2305259085633662,
      "grad_norm": 274.0,
      "learning_rate": 1.5704068363366154e-05,
      "loss": 0.8241,
      "step": 10130
    },
    {
      "epoch": 0.2307534760940309,
      "grad_norm": 149.0,
      "learning_rate": 1.569942411294817e-05,
      "loss": 1.9021,
      "step": 10140
    },
    {
      "epoch": 0.23098104362469563,
      "grad_norm": 76.5,
      "learning_rate": 1.569477986253019e-05,
      "loss": 0.6726,
      "step": 10150
    },
    {
      "epoch": 0.23120861115536034,
      "grad_norm": 0.98046875,
      "learning_rate": 1.5690135612112208e-05,
      "loss": 0.9766,
      "step": 10160
    },
    {
      "epoch": 0.2314361786860251,
      "grad_norm": 2.734375,
      "learning_rate": 1.5685491361694222e-05,
      "loss": 0.422,
      "step": 10170
    },
    {
      "epoch": 0.2316637462166898,
      "grad_norm": 151.0,
      "learning_rate": 1.5680847111276242e-05,
      "loss": 1.6101,
      "step": 10180
    },
    {
      "epoch": 0.23189131374735453,
      "grad_norm": 136.0,
      "learning_rate": 1.567620286085826e-05,
      "loss": 1.1702,
      "step": 10190
    },
    {
      "epoch": 0.23211888127801925,
      "grad_norm": 104.0,
      "learning_rate": 1.5671558610440276e-05,
      "loss": 0.4103,
      "step": 10200
    },
    {
      "epoch": 0.23234644880868396,
      "grad_norm": 239.0,
      "learning_rate": 1.5666914360022293e-05,
      "loss": 1.6953,
      "step": 10210
    },
    {
      "epoch": 0.2325740163393487,
      "grad_norm": 302.0,
      "learning_rate": 1.566227010960431e-05,
      "loss": 1.4496,
      "step": 10220
    },
    {
      "epoch": 0.23280158387001343,
      "grad_norm": 0.00121307373046875,
      "learning_rate": 1.565762585918633e-05,
      "loss": 1.166,
      "step": 10230
    },
    {
      "epoch": 0.23302915140067815,
      "grad_norm": 131.0,
      "learning_rate": 1.5652981608768345e-05,
      "loss": 0.5258,
      "step": 10240
    },
    {
      "epoch": 0.23325671893134287,
      "grad_norm": 0.000659942626953125,
      "learning_rate": 1.564833735835036e-05,
      "loss": 0.6239,
      "step": 10250
    },
    {
      "epoch": 0.2334842864620076,
      "grad_norm": 1.625,
      "learning_rate": 1.5643693107932382e-05,
      "loss": 0.1668,
      "step": 10260
    },
    {
      "epoch": 0.23371185399267233,
      "grad_norm": 0.0145263671875,
      "learning_rate": 1.56390488575144e-05,
      "loss": 1.8953,
      "step": 10270
    },
    {
      "epoch": 0.23393942152333705,
      "grad_norm": 0.00186920166015625,
      "learning_rate": 1.5634404607096416e-05,
      "loss": 1.2062,
      "step": 10280
    },
    {
      "epoch": 0.23416698905400177,
      "grad_norm": 228.0,
      "learning_rate": 1.5629760356678433e-05,
      "loss": 1.2318,
      "step": 10290
    },
    {
      "epoch": 0.2343945565846665,
      "grad_norm": 117.5,
      "learning_rate": 1.562511610626045e-05,
      "loss": 0.753,
      "step": 10300
    },
    {
      "epoch": 0.23462212411533123,
      "grad_norm": 102.0,
      "learning_rate": 1.5620471855842467e-05,
      "loss": 0.9414,
      "step": 10310
    },
    {
      "epoch": 0.23484969164599595,
      "grad_norm": 258.0,
      "learning_rate": 1.5615827605424484e-05,
      "loss": 0.4186,
      "step": 10320
    },
    {
      "epoch": 0.23507725917666067,
      "grad_norm": 430.0,
      "learning_rate": 1.56111833550065e-05,
      "loss": 1.0574,
      "step": 10330
    },
    {
      "epoch": 0.2353048267073254,
      "grad_norm": 33.75,
      "learning_rate": 1.5606539104588522e-05,
      "loss": 0.3594,
      "step": 10340
    },
    {
      "epoch": 0.23553239423799013,
      "grad_norm": 1.1640625,
      "learning_rate": 1.560189485417054e-05,
      "loss": 1.1954,
      "step": 10350
    },
    {
      "epoch": 0.23575996176865485,
      "grad_norm": 70.0,
      "learning_rate": 1.5597250603752556e-05,
      "loss": 0.8519,
      "step": 10360
    },
    {
      "epoch": 0.23598752929931957,
      "grad_norm": 119.5,
      "learning_rate": 1.5592606353334573e-05,
      "loss": 0.4198,
      "step": 10370
    },
    {
      "epoch": 0.2362150968299843,
      "grad_norm": 113.5,
      "learning_rate": 1.558796210291659e-05,
      "loss": 0.7912,
      "step": 10380
    },
    {
      "epoch": 0.23644266436064904,
      "grad_norm": 0.041748046875,
      "learning_rate": 1.5583317852498607e-05,
      "loss": 0.7607,
      "step": 10390
    },
    {
      "epoch": 0.23667023189131375,
      "grad_norm": 127.0,
      "learning_rate": 1.5578673602080624e-05,
      "loss": 1.0994,
      "step": 10400
    },
    {
      "epoch": 0.23689779942197847,
      "grad_norm": 102.5,
      "learning_rate": 1.557402935166264e-05,
      "loss": 0.9864,
      "step": 10410
    },
    {
      "epoch": 0.2371253669526432,
      "grad_norm": 164.0,
      "learning_rate": 1.556938510124466e-05,
      "loss": 0.8547,
      "step": 10420
    },
    {
      "epoch": 0.2373529344833079,
      "grad_norm": 3.6875,
      "learning_rate": 1.556474085082668e-05,
      "loss": 0.3589,
      "step": 10430
    },
    {
      "epoch": 0.23758050201397266,
      "grad_norm": 13.5,
      "learning_rate": 1.5560096600408696e-05,
      "loss": 0.5556,
      "step": 10440
    },
    {
      "epoch": 0.23780806954463737,
      "grad_norm": 410.0,
      "learning_rate": 1.5555452349990713e-05,
      "loss": 0.9017,
      "step": 10450
    },
    {
      "epoch": 0.2380356370753021,
      "grad_norm": 0.051513671875,
      "learning_rate": 1.555080809957273e-05,
      "loss": 1.1998,
      "step": 10460
    },
    {
      "epoch": 0.2382632046059668,
      "grad_norm": 326.0,
      "learning_rate": 1.5546163849154747e-05,
      "loss": 1.9923,
      "step": 10470
    },
    {
      "epoch": 0.23849077213663156,
      "grad_norm": 61.5,
      "learning_rate": 1.5541519598736764e-05,
      "loss": 1.11,
      "step": 10480
    },
    {
      "epoch": 0.23871833966729628,
      "grad_norm": 21.125,
      "learning_rate": 1.5536875348318784e-05,
      "loss": 0.4778,
      "step": 10490
    },
    {
      "epoch": 0.238945907197961,
      "grad_norm": 306.0,
      "learning_rate": 1.55322310979008e-05,
      "loss": 1.3553,
      "step": 10500
    },
    {
      "epoch": 0.2391734747286257,
      "grad_norm": 42.5,
      "learning_rate": 1.552758684748282e-05,
      "loss": 0.9198,
      "step": 10510
    },
    {
      "epoch": 0.23940104225929043,
      "grad_norm": 87.5,
      "learning_rate": 1.5522942597064836e-05,
      "loss": 0.1433,
      "step": 10520
    },
    {
      "epoch": 0.23962860978995518,
      "grad_norm": 0.62890625,
      "learning_rate": 1.5518298346646853e-05,
      "loss": 0.7744,
      "step": 10530
    },
    {
      "epoch": 0.2398561773206199,
      "grad_norm": 39.5,
      "learning_rate": 1.551365409622887e-05,
      "loss": 1.0875,
      "step": 10540
    },
    {
      "epoch": 0.24008374485128461,
      "grad_norm": 0.296875,
      "learning_rate": 1.5509009845810887e-05,
      "loss": 0.4929,
      "step": 10550
    },
    {
      "epoch": 0.24031131238194933,
      "grad_norm": 28.75,
      "learning_rate": 1.5504365595392904e-05,
      "loss": 0.8594,
      "step": 10560
    },
    {
      "epoch": 0.24053887991261408,
      "grad_norm": 5.375,
      "learning_rate": 1.5499721344974924e-05,
      "loss": 0.6428,
      "step": 10570
    },
    {
      "epoch": 0.2407664474432788,
      "grad_norm": 61.25,
      "learning_rate": 1.549507709455694e-05,
      "loss": 1.1389,
      "step": 10580
    },
    {
      "epoch": 0.24099401497394352,
      "grad_norm": 191.0,
      "learning_rate": 1.5490432844138955e-05,
      "loss": 1.119,
      "step": 10590
    },
    {
      "epoch": 0.24122158250460823,
      "grad_norm": 141.0,
      "learning_rate": 1.5485788593720975e-05,
      "loss": 0.6695,
      "step": 10600
    },
    {
      "epoch": 0.24144915003527298,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.5481144343302992e-05,
      "loss": 0.4979,
      "step": 10610
    },
    {
      "epoch": 0.2416767175659377,
      "grad_norm": 76.0,
      "learning_rate": 1.547650009288501e-05,
      "loss": 0.9168,
      "step": 10620
    },
    {
      "epoch": 0.24190428509660242,
      "grad_norm": 110.5,
      "learning_rate": 1.5471855842467027e-05,
      "loss": 1.0271,
      "step": 10630
    },
    {
      "epoch": 0.24213185262726714,
      "grad_norm": 87.5,
      "learning_rate": 1.5467211592049044e-05,
      "loss": 0.5534,
      "step": 10640
    },
    {
      "epoch": 0.24235942015793185,
      "grad_norm": 199.0,
      "learning_rate": 1.5462567341631064e-05,
      "loss": 1.3976,
      "step": 10650
    },
    {
      "epoch": 0.2425869876885966,
      "grad_norm": 0.02099609375,
      "learning_rate": 1.5457923091213078e-05,
      "loss": 0.9919,
      "step": 10660
    },
    {
      "epoch": 0.24281455521926132,
      "grad_norm": 102.0,
      "learning_rate": 1.5453278840795095e-05,
      "loss": 0.6294,
      "step": 10670
    },
    {
      "epoch": 0.24304212274992604,
      "grad_norm": 28.5,
      "learning_rate": 1.5448634590377115e-05,
      "loss": 0.808,
      "step": 10680
    },
    {
      "epoch": 0.24326969028059076,
      "grad_norm": 73.5,
      "learning_rate": 1.5443990339959132e-05,
      "loss": 0.5964,
      "step": 10690
    },
    {
      "epoch": 0.2434972578112555,
      "grad_norm": 55.0,
      "learning_rate": 1.543934608954115e-05,
      "loss": 0.9213,
      "step": 10700
    },
    {
      "epoch": 0.24372482534192022,
      "grad_norm": 105.5,
      "learning_rate": 1.5434701839123166e-05,
      "loss": 1.5111,
      "step": 10710
    },
    {
      "epoch": 0.24395239287258494,
      "grad_norm": 364.0,
      "learning_rate": 1.5430057588705183e-05,
      "loss": 1.5324,
      "step": 10720
    },
    {
      "epoch": 0.24417996040324966,
      "grad_norm": 276.0,
      "learning_rate": 1.54254133382872e-05,
      "loss": 1.1884,
      "step": 10730
    },
    {
      "epoch": 0.24440752793391438,
      "grad_norm": 282.0,
      "learning_rate": 1.5420769087869218e-05,
      "loss": 0.736,
      "step": 10740
    },
    {
      "epoch": 0.24463509546457912,
      "grad_norm": 636.0,
      "learning_rate": 1.5416124837451235e-05,
      "loss": 0.5855,
      "step": 10750
    },
    {
      "epoch": 0.24486266299524384,
      "grad_norm": 44.5,
      "learning_rate": 1.5411480587033255e-05,
      "loss": 0.5255,
      "step": 10760
    },
    {
      "epoch": 0.24509023052590856,
      "grad_norm": 150.0,
      "learning_rate": 1.5406836336615272e-05,
      "loss": 0.5835,
      "step": 10770
    },
    {
      "epoch": 0.24531779805657328,
      "grad_norm": 109.0,
      "learning_rate": 1.540219208619729e-05,
      "loss": 0.6366,
      "step": 10780
    },
    {
      "epoch": 0.24554536558723802,
      "grad_norm": 0.000881195068359375,
      "learning_rate": 1.5397547835779306e-05,
      "loss": 1.2602,
      "step": 10790
    },
    {
      "epoch": 0.24577293311790274,
      "grad_norm": 1.953125,
      "learning_rate": 1.5392903585361323e-05,
      "loss": 0.5185,
      "step": 10800
    },
    {
      "epoch": 0.24600050064856746,
      "grad_norm": 6.03125,
      "learning_rate": 1.538825933494334e-05,
      "loss": 1.3522,
      "step": 10810
    },
    {
      "epoch": 0.24622806817923218,
      "grad_norm": 125.5,
      "learning_rate": 1.5383615084525357e-05,
      "loss": 1.024,
      "step": 10820
    },
    {
      "epoch": 0.24645563570989693,
      "grad_norm": 2.078125,
      "learning_rate": 1.5378970834107378e-05,
      "loss": 0.2957,
      "step": 10830
    },
    {
      "epoch": 0.24668320324056164,
      "grad_norm": 288.0,
      "learning_rate": 1.5374326583689395e-05,
      "loss": 0.3122,
      "step": 10840
    },
    {
      "epoch": 0.24691077077122636,
      "grad_norm": 194.0,
      "learning_rate": 1.5369682333271412e-05,
      "loss": 0.7117,
      "step": 10850
    },
    {
      "epoch": 0.24713833830189108,
      "grad_norm": 0.076171875,
      "learning_rate": 1.536503808285343e-05,
      "loss": 0.9811,
      "step": 10860
    },
    {
      "epoch": 0.2473659058325558,
      "grad_norm": 92.5,
      "learning_rate": 1.5360393832435446e-05,
      "loss": 0.6828,
      "step": 10870
    },
    {
      "epoch": 0.24759347336322055,
      "grad_norm": 0.0001659393310546875,
      "learning_rate": 1.5355749582017463e-05,
      "loss": 0.479,
      "step": 10880
    },
    {
      "epoch": 0.24782104089388526,
      "grad_norm": 274.0,
      "learning_rate": 1.535110533159948e-05,
      "loss": 1.662,
      "step": 10890
    },
    {
      "epoch": 0.24804860842454998,
      "grad_norm": 14.875,
      "learning_rate": 1.5346461081181497e-05,
      "loss": 0.6964,
      "step": 10900
    },
    {
      "epoch": 0.2482761759552147,
      "grad_norm": 67.0,
      "learning_rate": 1.5341816830763518e-05,
      "loss": 0.5646,
      "step": 10910
    },
    {
      "epoch": 0.24850374348587945,
      "grad_norm": 6.9375,
      "learning_rate": 1.5337172580345535e-05,
      "loss": 0.2894,
      "step": 10920
    },
    {
      "epoch": 0.24873131101654417,
      "grad_norm": 121.5,
      "learning_rate": 1.5332528329927552e-05,
      "loss": 1.223,
      "step": 10930
    },
    {
      "epoch": 0.24895887854720888,
      "grad_norm": 0.0012969970703125,
      "learning_rate": 1.532788407950957e-05,
      "loss": 0.366,
      "step": 10940
    },
    {
      "epoch": 0.2491864460778736,
      "grad_norm": 137.0,
      "learning_rate": 1.5323239829091586e-05,
      "loss": 0.7015,
      "step": 10950
    },
    {
      "epoch": 0.24941401360853832,
      "grad_norm": 220.0,
      "learning_rate": 1.5318595578673603e-05,
      "loss": 1.6721,
      "step": 10960
    },
    {
      "epoch": 0.24964158113920307,
      "grad_norm": 78.5,
      "learning_rate": 1.531395132825562e-05,
      "loss": 0.9245,
      "step": 10970
    },
    {
      "epoch": 0.24986914866986779,
      "grad_norm": 0.34375,
      "learning_rate": 1.5309307077837637e-05,
      "loss": 0.4509,
      "step": 10980
    },
    {
      "epoch": 0.25009671620053253,
      "grad_norm": 6.96875,
      "learning_rate": 1.5304662827419658e-05,
      "loss": 0.8195,
      "step": 10990
    },
    {
      "epoch": 0.25032428373119725,
      "grad_norm": 0.000629425048828125,
      "learning_rate": 1.5300018577001675e-05,
      "loss": 0.9289,
      "step": 11000
    },
    {
      "epoch": 0.25055185126186197,
      "grad_norm": 138.0,
      "learning_rate": 1.5295374326583688e-05,
      "loss": 1.4042,
      "step": 11010
    },
    {
      "epoch": 0.2507794187925267,
      "grad_norm": 151.0,
      "learning_rate": 1.529073007616571e-05,
      "loss": 0.5587,
      "step": 11020
    },
    {
      "epoch": 0.2510069863231914,
      "grad_norm": 13.8125,
      "learning_rate": 1.5286085825747726e-05,
      "loss": 0.7141,
      "step": 11030
    },
    {
      "epoch": 0.2512345538538561,
      "grad_norm": 248.0,
      "learning_rate": 1.5281441575329743e-05,
      "loss": 0.9725,
      "step": 11040
    },
    {
      "epoch": 0.25146212138452084,
      "grad_norm": 1.1640625,
      "learning_rate": 1.527679732491176e-05,
      "loss": 1.3071,
      "step": 11050
    },
    {
      "epoch": 0.25168968891518556,
      "grad_norm": 368.0,
      "learning_rate": 1.5272153074493777e-05,
      "loss": 1.1907,
      "step": 11060
    },
    {
      "epoch": 0.2519172564458503,
      "grad_norm": 218.0,
      "learning_rate": 1.5267508824075797e-05,
      "loss": 0.5203,
      "step": 11070
    },
    {
      "epoch": 0.25214482397651505,
      "grad_norm": 0.00118255615234375,
      "learning_rate": 1.526286457365781e-05,
      "loss": 0.6737,
      "step": 11080
    },
    {
      "epoch": 0.25237239150717977,
      "grad_norm": 0.0255126953125,
      "learning_rate": 1.525822032323983e-05,
      "loss": 0.822,
      "step": 11090
    },
    {
      "epoch": 0.2525999590378445,
      "grad_norm": 0.0174560546875,
      "learning_rate": 1.5253576072821847e-05,
      "loss": 1.1703,
      "step": 11100
    },
    {
      "epoch": 0.2528275265685092,
      "grad_norm": 113.5,
      "learning_rate": 1.5248931822403866e-05,
      "loss": 0.4113,
      "step": 11110
    },
    {
      "epoch": 0.2530550940991739,
      "grad_norm": 512.0,
      "learning_rate": 1.5244287571985883e-05,
      "loss": 0.5001,
      "step": 11120
    },
    {
      "epoch": 0.25328266162983865,
      "grad_norm": 0.00408935546875,
      "learning_rate": 1.5239643321567901e-05,
      "loss": 0.9808,
      "step": 11130
    },
    {
      "epoch": 0.25351022916050336,
      "grad_norm": 45.0,
      "learning_rate": 1.5234999071149918e-05,
      "loss": 1.0447,
      "step": 11140
    },
    {
      "epoch": 0.2537377966911681,
      "grad_norm": 136.0,
      "learning_rate": 1.5230354820731934e-05,
      "loss": 0.8825,
      "step": 11150
    },
    {
      "epoch": 0.25396536422183286,
      "grad_norm": 153.0,
      "learning_rate": 1.5225710570313953e-05,
      "loss": 0.6856,
      "step": 11160
    },
    {
      "epoch": 0.2541929317524976,
      "grad_norm": 130.0,
      "learning_rate": 1.522106631989597e-05,
      "loss": 0.9714,
      "step": 11170
    },
    {
      "epoch": 0.2544204992831623,
      "grad_norm": 99.0,
      "learning_rate": 1.5216422069477987e-05,
      "loss": 1.7929,
      "step": 11180
    },
    {
      "epoch": 0.254648066813827,
      "grad_norm": 197.0,
      "learning_rate": 1.5211777819060005e-05,
      "loss": 0.5649,
      "step": 11190
    },
    {
      "epoch": 0.25487563434449173,
      "grad_norm": 53.5,
      "learning_rate": 1.5207133568642022e-05,
      "loss": 0.4676,
      "step": 11200
    },
    {
      "epoch": 0.25510320187515645,
      "grad_norm": 115.5,
      "learning_rate": 1.5202489318224041e-05,
      "loss": 0.3084,
      "step": 11210
    },
    {
      "epoch": 0.25533076940582117,
      "grad_norm": 114.0,
      "learning_rate": 1.5197845067806057e-05,
      "loss": 1.4257,
      "step": 11220
    },
    {
      "epoch": 0.2555583369364859,
      "grad_norm": 0.046875,
      "learning_rate": 1.5193200817388074e-05,
      "loss": 1.0898,
      "step": 11230
    },
    {
      "epoch": 0.2557859044671506,
      "grad_norm": 2.25,
      "learning_rate": 1.5188556566970092e-05,
      "loss": 0.2494,
      "step": 11240
    },
    {
      "epoch": 0.2560134719978154,
      "grad_norm": 152.0,
      "learning_rate": 1.518391231655211e-05,
      "loss": 0.7775,
      "step": 11250
    },
    {
      "epoch": 0.2562410395284801,
      "grad_norm": 374.0,
      "learning_rate": 1.5179268066134128e-05,
      "loss": 1.4239,
      "step": 11260
    },
    {
      "epoch": 0.2564686070591448,
      "grad_norm": 0.0034332275390625,
      "learning_rate": 1.5174623815716145e-05,
      "loss": 0.1608,
      "step": 11270
    },
    {
      "epoch": 0.25669617458980953,
      "grad_norm": 118.5,
      "learning_rate": 1.5169979565298162e-05,
      "loss": 0.898,
      "step": 11280
    },
    {
      "epoch": 0.25692374212047425,
      "grad_norm": 160.0,
      "learning_rate": 1.516533531488018e-05,
      "loss": 0.5473,
      "step": 11290
    },
    {
      "epoch": 0.25715130965113897,
      "grad_norm": 98.0,
      "learning_rate": 1.5160691064462196e-05,
      "loss": 0.6661,
      "step": 11300
    },
    {
      "epoch": 0.2573788771818037,
      "grad_norm": 516.0,
      "learning_rate": 1.5156046814044213e-05,
      "loss": 1.9695,
      "step": 11310
    },
    {
      "epoch": 0.2576064447124684,
      "grad_norm": 312.0,
      "learning_rate": 1.5151402563626232e-05,
      "loss": 1.2844,
      "step": 11320
    },
    {
      "epoch": 0.2578340122431331,
      "grad_norm": 32.75,
      "learning_rate": 1.514675831320825e-05,
      "loss": 1.5731,
      "step": 11330
    },
    {
      "epoch": 0.2580615797737979,
      "grad_norm": 0.0260009765625,
      "learning_rate": 1.5142114062790268e-05,
      "loss": 0.1098,
      "step": 11340
    },
    {
      "epoch": 0.2582891473044626,
      "grad_norm": 23.75,
      "learning_rate": 1.5137469812372285e-05,
      "loss": 1.3285,
      "step": 11350
    },
    {
      "epoch": 0.25851671483512734,
      "grad_norm": 0.41796875,
      "learning_rate": 1.51328255619543e-05,
      "loss": 0.7926,
      "step": 11360
    },
    {
      "epoch": 0.25874428236579206,
      "grad_norm": 0.0047607421875,
      "learning_rate": 1.512818131153632e-05,
      "loss": 0.3374,
      "step": 11370
    },
    {
      "epoch": 0.2589718498964568,
      "grad_norm": 117.5,
      "learning_rate": 1.5123537061118336e-05,
      "loss": 1.4112,
      "step": 11380
    },
    {
      "epoch": 0.2591994174271215,
      "grad_norm": 304.0,
      "learning_rate": 1.5118892810700353e-05,
      "loss": 0.8933,
      "step": 11390
    },
    {
      "epoch": 0.2594269849577862,
      "grad_norm": 86.0,
      "learning_rate": 1.5114248560282372e-05,
      "loss": 1.1187,
      "step": 11400
    },
    {
      "epoch": 0.25965455248845093,
      "grad_norm": 64.0,
      "learning_rate": 1.5109604309864389e-05,
      "loss": 0.9441,
      "step": 11410
    },
    {
      "epoch": 0.25988212001911565,
      "grad_norm": 238.0,
      "learning_rate": 1.5104960059446408e-05,
      "loss": 1.0595,
      "step": 11420
    },
    {
      "epoch": 0.2601096875497804,
      "grad_norm": 7.625,
      "learning_rate": 1.5100315809028423e-05,
      "loss": 0.9034,
      "step": 11430
    },
    {
      "epoch": 0.26033725508044514,
      "grad_norm": 1.3125,
      "learning_rate": 1.509567155861044e-05,
      "loss": 0.4803,
      "step": 11440
    },
    {
      "epoch": 0.26056482261110986,
      "grad_norm": 5.8125,
      "learning_rate": 1.5091027308192459e-05,
      "loss": 0.3417,
      "step": 11450
    },
    {
      "epoch": 0.2607923901417746,
      "grad_norm": 0.00150299072265625,
      "learning_rate": 1.5086383057774476e-05,
      "loss": 0.3545,
      "step": 11460
    },
    {
      "epoch": 0.2610199576724393,
      "grad_norm": 119.0,
      "learning_rate": 1.5081738807356495e-05,
      "loss": 0.9882,
      "step": 11470
    },
    {
      "epoch": 0.261247525203104,
      "grad_norm": 0.00109100341796875,
      "learning_rate": 1.5077094556938512e-05,
      "loss": 0.9615,
      "step": 11480
    },
    {
      "epoch": 0.26147509273376873,
      "grad_norm": 12.1875,
      "learning_rate": 1.5072450306520529e-05,
      "loss": 0.8557,
      "step": 11490
    },
    {
      "epoch": 0.26170266026443345,
      "grad_norm": 0.00188446044921875,
      "learning_rate": 1.5067806056102546e-05,
      "loss": 0.9699,
      "step": 11500
    },
    {
      "epoch": 0.26193022779509817,
      "grad_norm": 146.0,
      "learning_rate": 1.5063161805684563e-05,
      "loss": 0.6161,
      "step": 11510
    },
    {
      "epoch": 0.26215779532576294,
      "grad_norm": 52.0,
      "learning_rate": 1.505851755526658e-05,
      "loss": 0.4844,
      "step": 11520
    },
    {
      "epoch": 0.26238536285642766,
      "grad_norm": 66.5,
      "learning_rate": 1.5053873304848599e-05,
      "loss": 0.5849,
      "step": 11530
    },
    {
      "epoch": 0.2626129303870924,
      "grad_norm": 198.0,
      "learning_rate": 1.5049229054430616e-05,
      "loss": 0.8534,
      "step": 11540
    },
    {
      "epoch": 0.2628404979177571,
      "grad_norm": 60.75,
      "learning_rate": 1.5044584804012635e-05,
      "loss": 0.5026,
      "step": 11550
    },
    {
      "epoch": 0.2630680654484218,
      "grad_norm": 0.11181640625,
      "learning_rate": 1.5039940553594652e-05,
      "loss": 0.3563,
      "step": 11560
    },
    {
      "epoch": 0.26329563297908654,
      "grad_norm": 19.25,
      "learning_rate": 1.5035296303176667e-05,
      "loss": 0.6289,
      "step": 11570
    },
    {
      "epoch": 0.26352320050975125,
      "grad_norm": 88.0,
      "learning_rate": 1.5030652052758686e-05,
      "loss": 0.57,
      "step": 11580
    },
    {
      "epoch": 0.263750768040416,
      "grad_norm": 2.625,
      "learning_rate": 1.5026007802340703e-05,
      "loss": 0.5479,
      "step": 11590
    },
    {
      "epoch": 0.2639783355710807,
      "grad_norm": 0.001739501953125,
      "learning_rate": 1.5021363551922722e-05,
      "loss": 0.6788,
      "step": 11600
    },
    {
      "epoch": 0.26420590310174547,
      "grad_norm": 0.01470947265625,
      "learning_rate": 1.5016719301504739e-05,
      "loss": 0.5358,
      "step": 11610
    },
    {
      "epoch": 0.2644334706324102,
      "grad_norm": 0.000743865966796875,
      "learning_rate": 1.5012075051086756e-05,
      "loss": 0.8938,
      "step": 11620
    },
    {
      "epoch": 0.2646610381630749,
      "grad_norm": 0.006561279296875,
      "learning_rate": 1.5007430800668774e-05,
      "loss": 0.0581,
      "step": 11630
    },
    {
      "epoch": 0.2648886056937396,
      "grad_norm": 60.25,
      "learning_rate": 1.500278655025079e-05,
      "loss": 0.4747,
      "step": 11640
    },
    {
      "epoch": 0.26511617322440434,
      "grad_norm": 163.0,
      "learning_rate": 1.4998142299832807e-05,
      "loss": 0.402,
      "step": 11650
    },
    {
      "epoch": 0.26534374075506906,
      "grad_norm": 147.0,
      "learning_rate": 1.4993498049414826e-05,
      "loss": 0.6031,
      "step": 11660
    },
    {
      "epoch": 0.2655713082857338,
      "grad_norm": 0.12451171875,
      "learning_rate": 1.4988853798996843e-05,
      "loss": 0.3656,
      "step": 11670
    },
    {
      "epoch": 0.2657988758163985,
      "grad_norm": 0.000514984130859375,
      "learning_rate": 1.4984209548578861e-05,
      "loss": 0.4934,
      "step": 11680
    },
    {
      "epoch": 0.26602644334706327,
      "grad_norm": 296.0,
      "learning_rate": 1.4979565298160878e-05,
      "loss": 0.776,
      "step": 11690
    },
    {
      "epoch": 0.266254010877728,
      "grad_norm": 173.0,
      "learning_rate": 1.4974921047742896e-05,
      "loss": 0.5252,
      "step": 11700
    },
    {
      "epoch": 0.2664815784083927,
      "grad_norm": 101.0,
      "learning_rate": 1.4970276797324913e-05,
      "loss": 0.9683,
      "step": 11710
    },
    {
      "epoch": 0.2667091459390574,
      "grad_norm": 21.875,
      "learning_rate": 1.496563254690693e-05,
      "loss": 0.5718,
      "step": 11720
    },
    {
      "epoch": 0.26693671346972214,
      "grad_norm": 0.0002231597900390625,
      "learning_rate": 1.4960988296488947e-05,
      "loss": 0.3878,
      "step": 11730
    },
    {
      "epoch": 0.26716428100038686,
      "grad_norm": 179.0,
      "learning_rate": 1.4956344046070965e-05,
      "loss": 1.224,
      "step": 11740
    },
    {
      "epoch": 0.2673918485310516,
      "grad_norm": 162.0,
      "learning_rate": 1.4951699795652982e-05,
      "loss": 0.1164,
      "step": 11750
    },
    {
      "epoch": 0.2676194160617163,
      "grad_norm": 312.0,
      "learning_rate": 1.4947055545235001e-05,
      "loss": 0.9184,
      "step": 11760
    },
    {
      "epoch": 0.267846983592381,
      "grad_norm": 192.0,
      "learning_rate": 1.4942411294817018e-05,
      "loss": 1.1383,
      "step": 11770
    },
    {
      "epoch": 0.2680745511230458,
      "grad_norm": 76.0,
      "learning_rate": 1.4937767044399034e-05,
      "loss": 0.3553,
      "step": 11780
    },
    {
      "epoch": 0.2683021186537105,
      "grad_norm": 272.0,
      "learning_rate": 1.4933122793981052e-05,
      "loss": 0.5806,
      "step": 11790
    },
    {
      "epoch": 0.2685296861843752,
      "grad_norm": 239.0,
      "learning_rate": 1.492847854356307e-05,
      "loss": 1.0596,
      "step": 11800
    },
    {
      "epoch": 0.26875725371503995,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4923834293145088e-05,
      "loss": 2.0093,
      "step": 11810
    },
    {
      "epoch": 0.26898482124570466,
      "grad_norm": 97.0,
      "learning_rate": 1.4919190042727105e-05,
      "loss": 0.6592,
      "step": 11820
    },
    {
      "epoch": 0.2692123887763694,
      "grad_norm": 0.000171661376953125,
      "learning_rate": 1.4914545792309122e-05,
      "loss": 0.7887,
      "step": 11830
    },
    {
      "epoch": 0.2694399563070341,
      "grad_norm": 55.0,
      "learning_rate": 1.4909901541891141e-05,
      "loss": 0.6898,
      "step": 11840
    },
    {
      "epoch": 0.2696675238376988,
      "grad_norm": 0.8359375,
      "learning_rate": 1.4905257291473156e-05,
      "loss": 0.2239,
      "step": 11850
    },
    {
      "epoch": 0.26989509136836354,
      "grad_norm": 0.0002956390380859375,
      "learning_rate": 1.4900613041055173e-05,
      "loss": 0.4225,
      "step": 11860
    },
    {
      "epoch": 0.2701226588990283,
      "grad_norm": 0.0302734375,
      "learning_rate": 1.4895968790637192e-05,
      "loss": 0.7709,
      "step": 11870
    },
    {
      "epoch": 0.27035022642969303,
      "grad_norm": 202.0,
      "learning_rate": 1.489132454021921e-05,
      "loss": 1.5388,
      "step": 11880
    },
    {
      "epoch": 0.27057779396035775,
      "grad_norm": 190.0,
      "learning_rate": 1.4886680289801228e-05,
      "loss": 0.8734,
      "step": 11890
    },
    {
      "epoch": 0.27080536149102247,
      "grad_norm": 66.0,
      "learning_rate": 1.4882036039383245e-05,
      "loss": 0.9225,
      "step": 11900
    },
    {
      "epoch": 0.2710329290216872,
      "grad_norm": 150.0,
      "learning_rate": 1.4877391788965264e-05,
      "loss": 1.9771,
      "step": 11910
    },
    {
      "epoch": 0.2712604965523519,
      "grad_norm": 15.75,
      "learning_rate": 1.487274753854728e-05,
      "loss": 0.6869,
      "step": 11920
    },
    {
      "epoch": 0.2714880640830166,
      "grad_norm": 0.002655029296875,
      "learning_rate": 1.4868103288129296e-05,
      "loss": 0.5966,
      "step": 11930
    },
    {
      "epoch": 0.27171563161368134,
      "grad_norm": 0.026123046875,
      "learning_rate": 1.4863459037711313e-05,
      "loss": 0.7905,
      "step": 11940
    },
    {
      "epoch": 0.27194319914434606,
      "grad_norm": 46.0,
      "learning_rate": 1.4858814787293332e-05,
      "loss": 1.7054,
      "step": 11950
    },
    {
      "epoch": 0.27217076667501083,
      "grad_norm": 120.5,
      "learning_rate": 1.4854170536875349e-05,
      "loss": 0.7836,
      "step": 11960
    },
    {
      "epoch": 0.27239833420567555,
      "grad_norm": 0.2236328125,
      "learning_rate": 1.4849526286457368e-05,
      "loss": 0.6396,
      "step": 11970
    },
    {
      "epoch": 0.27262590173634027,
      "grad_norm": 0.609375,
      "learning_rate": 1.4844882036039385e-05,
      "loss": 0.4424,
      "step": 11980
    },
    {
      "epoch": 0.272853469267005,
      "grad_norm": 238.0,
      "learning_rate": 1.48402377856214e-05,
      "loss": 0.4038,
      "step": 11990
    },
    {
      "epoch": 0.2730810367976697,
      "grad_norm": 0.00139617919921875,
      "learning_rate": 1.4835593535203419e-05,
      "loss": 1.0529,
      "step": 12000
    },
    {
      "epoch": 0.2733086043283344,
      "grad_norm": 148.0,
      "learning_rate": 1.4830949284785436e-05,
      "loss": 0.801,
      "step": 12010
    },
    {
      "epoch": 0.27353617185899914,
      "grad_norm": 166.0,
      "learning_rate": 1.4826305034367455e-05,
      "loss": 0.7908,
      "step": 12020
    },
    {
      "epoch": 0.27376373938966386,
      "grad_norm": 44.0,
      "learning_rate": 1.4821660783949472e-05,
      "loss": 0.9472,
      "step": 12030
    },
    {
      "epoch": 0.2739913069203286,
      "grad_norm": 344.0,
      "learning_rate": 1.4817016533531489e-05,
      "loss": 1.2067,
      "step": 12040
    },
    {
      "epoch": 0.27421887445099336,
      "grad_norm": 2.21875,
      "learning_rate": 1.4812372283113508e-05,
      "loss": 0.7768,
      "step": 12050
    },
    {
      "epoch": 0.2744464419816581,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.4807728032695523e-05,
      "loss": 1.1369,
      "step": 12060
    },
    {
      "epoch": 0.2746740095123228,
      "grad_norm": 241.0,
      "learning_rate": 1.480308378227754e-05,
      "loss": 0.7804,
      "step": 12070
    },
    {
      "epoch": 0.2749015770429875,
      "grad_norm": 187.0,
      "learning_rate": 1.4798439531859559e-05,
      "loss": 1.5419,
      "step": 12080
    },
    {
      "epoch": 0.27512914457365223,
      "grad_norm": 169.0,
      "learning_rate": 1.4793795281441576e-05,
      "loss": 0.6534,
      "step": 12090
    },
    {
      "epoch": 0.27535671210431695,
      "grad_norm": 196.0,
      "learning_rate": 1.4789151031023595e-05,
      "loss": 1.6335,
      "step": 12100
    },
    {
      "epoch": 0.27558427963498167,
      "grad_norm": 43.0,
      "learning_rate": 1.4784506780605612e-05,
      "loss": 0.3383,
      "step": 12110
    },
    {
      "epoch": 0.2758118471656464,
      "grad_norm": 2.8125,
      "learning_rate": 1.477986253018763e-05,
      "loss": 0.5496,
      "step": 12120
    },
    {
      "epoch": 0.2760394146963111,
      "grad_norm": 123.0,
      "learning_rate": 1.4775218279769646e-05,
      "loss": 0.9482,
      "step": 12130
    },
    {
      "epoch": 0.2762669822269759,
      "grad_norm": 0.037109375,
      "learning_rate": 1.4770574029351663e-05,
      "loss": 0.5666,
      "step": 12140
    },
    {
      "epoch": 0.2764945497576406,
      "grad_norm": 0.07275390625,
      "learning_rate": 1.4765929778933682e-05,
      "loss": 0.689,
      "step": 12150
    },
    {
      "epoch": 0.2767221172883053,
      "grad_norm": 181.0,
      "learning_rate": 1.4761285528515699e-05,
      "loss": 0.7099,
      "step": 12160
    },
    {
      "epoch": 0.27694968481897003,
      "grad_norm": 83.5,
      "learning_rate": 1.4756641278097716e-05,
      "loss": 0.7678,
      "step": 12170
    },
    {
      "epoch": 0.27717725234963475,
      "grad_norm": 5.03125,
      "learning_rate": 1.4751997027679734e-05,
      "loss": 0.9393,
      "step": 12180
    },
    {
      "epoch": 0.27740481988029947,
      "grad_norm": 0.03662109375,
      "learning_rate": 1.4747352777261752e-05,
      "loss": 1.3429,
      "step": 12190
    },
    {
      "epoch": 0.2776323874109642,
      "grad_norm": 29.875,
      "learning_rate": 1.4742708526843767e-05,
      "loss": 1.5743,
      "step": 12200
    },
    {
      "epoch": 0.2778599549416289,
      "grad_norm": 0.029052734375,
      "learning_rate": 1.4738064276425786e-05,
      "loss": 0.4431,
      "step": 12210
    },
    {
      "epoch": 0.2780875224722937,
      "grad_norm": 0.035888671875,
      "learning_rate": 1.4733420026007803e-05,
      "loss": 0.608,
      "step": 12220
    },
    {
      "epoch": 0.2783150900029584,
      "grad_norm": 2.484375,
      "learning_rate": 1.4728775775589821e-05,
      "loss": 0.5203,
      "step": 12230
    },
    {
      "epoch": 0.2785426575336231,
      "grad_norm": 206.0,
      "learning_rate": 1.4724131525171839e-05,
      "loss": 1.1412,
      "step": 12240
    },
    {
      "epoch": 0.27877022506428784,
      "grad_norm": 111.0,
      "learning_rate": 1.4719487274753856e-05,
      "loss": 0.4841,
      "step": 12250
    },
    {
      "epoch": 0.27899779259495255,
      "grad_norm": 226.0,
      "learning_rate": 1.4714843024335874e-05,
      "loss": 1.2142,
      "step": 12260
    },
    {
      "epoch": 0.2792253601256173,
      "grad_norm": 59.0,
      "learning_rate": 1.471019877391789e-05,
      "loss": 1.4668,
      "step": 12270
    },
    {
      "epoch": 0.279452927656282,
      "grad_norm": 2.21875,
      "learning_rate": 1.4705554523499907e-05,
      "loss": 0.6751,
      "step": 12280
    },
    {
      "epoch": 0.2796804951869467,
      "grad_norm": 150.0,
      "learning_rate": 1.4700910273081925e-05,
      "loss": 0.8443,
      "step": 12290
    },
    {
      "epoch": 0.27990806271761143,
      "grad_norm": 111.0,
      "learning_rate": 1.4696266022663943e-05,
      "loss": 0.3593,
      "step": 12300
    },
    {
      "epoch": 0.2801356302482762,
      "grad_norm": 334.0,
      "learning_rate": 1.4691621772245961e-05,
      "loss": 0.529,
      "step": 12310
    },
    {
      "epoch": 0.2803631977789409,
      "grad_norm": 340.0,
      "learning_rate": 1.4686977521827978e-05,
      "loss": 0.5697,
      "step": 12320
    },
    {
      "epoch": 0.28059076530960564,
      "grad_norm": 76.5,
      "learning_rate": 1.4682333271409997e-05,
      "loss": 0.7044,
      "step": 12330
    },
    {
      "epoch": 0.28081833284027036,
      "grad_norm": 44.5,
      "learning_rate": 1.4677689020992012e-05,
      "loss": 0.8766,
      "step": 12340
    },
    {
      "epoch": 0.2810459003709351,
      "grad_norm": 30.375,
      "learning_rate": 1.467304477057403e-05,
      "loss": 0.4032,
      "step": 12350
    },
    {
      "epoch": 0.2812734679015998,
      "grad_norm": 131.0,
      "learning_rate": 1.4668400520156048e-05,
      "loss": 0.7512,
      "step": 12360
    },
    {
      "epoch": 0.2815010354322645,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.4663756269738065e-05,
      "loss": 0.8798,
      "step": 12370
    },
    {
      "epoch": 0.28172860296292923,
      "grad_norm": 68.5,
      "learning_rate": 1.4659112019320082e-05,
      "loss": 0.3226,
      "step": 12380
    },
    {
      "epoch": 0.28195617049359395,
      "grad_norm": 15.75,
      "learning_rate": 1.4654467768902101e-05,
      "loss": 1.0542,
      "step": 12390
    },
    {
      "epoch": 0.2821837380242587,
      "grad_norm": 0.007415771484375,
      "learning_rate": 1.4649823518484118e-05,
      "loss": 0.7764,
      "step": 12400
    },
    {
      "epoch": 0.28241130555492344,
      "grad_norm": 0.05859375,
      "learning_rate": 1.4645179268066134e-05,
      "loss": 0.3243,
      "step": 12410
    },
    {
      "epoch": 0.28263887308558816,
      "grad_norm": 286.0,
      "learning_rate": 1.4640535017648152e-05,
      "loss": 1.1503,
      "step": 12420
    },
    {
      "epoch": 0.2828664406162529,
      "grad_norm": 59.0,
      "learning_rate": 1.463589076723017e-05,
      "loss": 0.4305,
      "step": 12430
    },
    {
      "epoch": 0.2830940081469176,
      "grad_norm": 150.0,
      "learning_rate": 1.4631246516812188e-05,
      "loss": 1.2739,
      "step": 12440
    },
    {
      "epoch": 0.2833215756775823,
      "grad_norm": 43.0,
      "learning_rate": 1.4626602266394205e-05,
      "loss": 0.3754,
      "step": 12450
    },
    {
      "epoch": 0.28354914320824703,
      "grad_norm": 236.0,
      "learning_rate": 1.4621958015976224e-05,
      "loss": 1.2023,
      "step": 12460
    },
    {
      "epoch": 0.28377671073891175,
      "grad_norm": 0.021240234375,
      "learning_rate": 1.4617313765558241e-05,
      "loss": 0.7067,
      "step": 12470
    },
    {
      "epoch": 0.28400427826957647,
      "grad_norm": 75.0,
      "learning_rate": 1.4612669515140258e-05,
      "loss": 1.188,
      "step": 12480
    },
    {
      "epoch": 0.28423184580024125,
      "grad_norm": 7.375,
      "learning_rate": 1.4608025264722275e-05,
      "loss": 0.6633,
      "step": 12490
    },
    {
      "epoch": 0.28445941333090596,
      "grad_norm": 536.0,
      "learning_rate": 1.4603381014304292e-05,
      "loss": 0.5964,
      "step": 12500
    },
    {
      "epoch": 0.2846869808615707,
      "grad_norm": 0.030029296875,
      "learning_rate": 1.459873676388631e-05,
      "loss": 0.4,
      "step": 12510
    },
    {
      "epoch": 0.2849145483922354,
      "grad_norm": 110.5,
      "learning_rate": 1.4594092513468328e-05,
      "loss": 0.4239,
      "step": 12520
    },
    {
      "epoch": 0.2851421159229001,
      "grad_norm": 84.0,
      "learning_rate": 1.4589448263050345e-05,
      "loss": 0.9811,
      "step": 12530
    },
    {
      "epoch": 0.28536968345356484,
      "grad_norm": 53.75,
      "learning_rate": 1.4584804012632364e-05,
      "loss": 0.9984,
      "step": 12540
    },
    {
      "epoch": 0.28559725098422956,
      "grad_norm": 177.0,
      "learning_rate": 1.458015976221438e-05,
      "loss": 0.5117,
      "step": 12550
    },
    {
      "epoch": 0.2858248185148943,
      "grad_norm": 18.0,
      "learning_rate": 1.4575515511796396e-05,
      "loss": 0.356,
      "step": 12560
    },
    {
      "epoch": 0.286052386045559,
      "grad_norm": 0.047607421875,
      "learning_rate": 1.4570871261378415e-05,
      "loss": 0.4773,
      "step": 12570
    },
    {
      "epoch": 0.28627995357622377,
      "grad_norm": 0.0030975341796875,
      "learning_rate": 1.4566227010960432e-05,
      "loss": 0.8755,
      "step": 12580
    },
    {
      "epoch": 0.2865075211068885,
      "grad_norm": 39.0,
      "learning_rate": 1.4561582760542449e-05,
      "loss": 0.6901,
      "step": 12590
    },
    {
      "epoch": 0.2867350886375532,
      "grad_norm": 85.0,
      "learning_rate": 1.4556938510124468e-05,
      "loss": 1.5524,
      "step": 12600
    },
    {
      "epoch": 0.2869626561682179,
      "grad_norm": 0.0002593994140625,
      "learning_rate": 1.4552294259706485e-05,
      "loss": 1.4922,
      "step": 12610
    },
    {
      "epoch": 0.28719022369888264,
      "grad_norm": 35.25,
      "learning_rate": 1.4547650009288504e-05,
      "loss": 0.418,
      "step": 12620
    },
    {
      "epoch": 0.28741779122954736,
      "grad_norm": 0.384765625,
      "learning_rate": 1.4543005758870519e-05,
      "loss": 1.2868,
      "step": 12630
    },
    {
      "epoch": 0.2876453587602121,
      "grad_norm": 418.0,
      "learning_rate": 1.4538361508452536e-05,
      "loss": 0.6271,
      "step": 12640
    },
    {
      "epoch": 0.2878729262908768,
      "grad_norm": 262.0,
      "learning_rate": 1.4533717258034555e-05,
      "loss": 1.1223,
      "step": 12650
    },
    {
      "epoch": 0.28810049382154157,
      "grad_norm": 74.0,
      "learning_rate": 1.4529073007616572e-05,
      "loss": 0.9126,
      "step": 12660
    },
    {
      "epoch": 0.2883280613522063,
      "grad_norm": 152.0,
      "learning_rate": 1.452442875719859e-05,
      "loss": 0.3271,
      "step": 12670
    },
    {
      "epoch": 0.288555628882871,
      "grad_norm": 8.4375,
      "learning_rate": 1.4519784506780608e-05,
      "loss": 0.3825,
      "step": 12680
    },
    {
      "epoch": 0.2887831964135357,
      "grad_norm": 136.0,
      "learning_rate": 1.4515140256362625e-05,
      "loss": 1.1227,
      "step": 12690
    },
    {
      "epoch": 0.28901076394420044,
      "grad_norm": 92.0,
      "learning_rate": 1.4510496005944642e-05,
      "loss": 0.6567,
      "step": 12700
    },
    {
      "epoch": 0.28923833147486516,
      "grad_norm": 105.5,
      "learning_rate": 1.4505851755526659e-05,
      "loss": 0.6709,
      "step": 12710
    },
    {
      "epoch": 0.2894658990055299,
      "grad_norm": 247.0,
      "learning_rate": 1.4501207505108676e-05,
      "loss": 0.8598,
      "step": 12720
    },
    {
      "epoch": 0.2896934665361946,
      "grad_norm": 364.0,
      "learning_rate": 1.4496563254690695e-05,
      "loss": 0.8281,
      "step": 12730
    },
    {
      "epoch": 0.2899210340668593,
      "grad_norm": 37.75,
      "learning_rate": 1.4491919004272712e-05,
      "loss": 0.6952,
      "step": 12740
    },
    {
      "epoch": 0.2901486015975241,
      "grad_norm": 210.0,
      "learning_rate": 1.448727475385473e-05,
      "loss": 0.6784,
      "step": 12750
    },
    {
      "epoch": 0.2903761691281888,
      "grad_norm": 95.5,
      "learning_rate": 1.4482630503436747e-05,
      "loss": 0.3614,
      "step": 12760
    },
    {
      "epoch": 0.29060373665885353,
      "grad_norm": 77.5,
      "learning_rate": 1.4477986253018763e-05,
      "loss": 0.7702,
      "step": 12770
    },
    {
      "epoch": 0.29083130418951825,
      "grad_norm": 163.0,
      "learning_rate": 1.4473342002600782e-05,
      "loss": 1.3064,
      "step": 12780
    },
    {
      "epoch": 0.29105887172018297,
      "grad_norm": 0.25,
      "learning_rate": 1.4468697752182799e-05,
      "loss": 1.1545,
      "step": 12790
    },
    {
      "epoch": 0.2912864392508477,
      "grad_norm": 146.0,
      "learning_rate": 1.4464053501764817e-05,
      "loss": 0.5964,
      "step": 12800
    },
    {
      "epoch": 0.2915140067815124,
      "grad_norm": 388.0,
      "learning_rate": 1.4459409251346834e-05,
      "loss": 0.8254,
      "step": 12810
    },
    {
      "epoch": 0.2917415743121771,
      "grad_norm": 184.0,
      "learning_rate": 1.4454765000928851e-05,
      "loss": 1.1512,
      "step": 12820
    },
    {
      "epoch": 0.29196914184284184,
      "grad_norm": 143.0,
      "learning_rate": 1.445012075051087e-05,
      "loss": 0.5881,
      "step": 12830
    },
    {
      "epoch": 0.2921967093735066,
      "grad_norm": 55.0,
      "learning_rate": 1.4445476500092886e-05,
      "loss": 1.2003,
      "step": 12840
    },
    {
      "epoch": 0.29242427690417133,
      "grad_norm": 0.0230712890625,
      "learning_rate": 1.4440832249674903e-05,
      "loss": 1.1012,
      "step": 12850
    },
    {
      "epoch": 0.29265184443483605,
      "grad_norm": 0.0019989013671875,
      "learning_rate": 1.4436187999256921e-05,
      "loss": 0.5693,
      "step": 12860
    },
    {
      "epoch": 0.29287941196550077,
      "grad_norm": 193.0,
      "learning_rate": 1.4431543748838938e-05,
      "loss": 0.9226,
      "step": 12870
    },
    {
      "epoch": 0.2931069794961655,
      "grad_norm": 127.0,
      "learning_rate": 1.4426899498420957e-05,
      "loss": 1.1348,
      "step": 12880
    },
    {
      "epoch": 0.2933345470268302,
      "grad_norm": 76.0,
      "learning_rate": 1.4422255248002974e-05,
      "loss": 0.3556,
      "step": 12890
    },
    {
      "epoch": 0.2935621145574949,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.4417610997584991e-05,
      "loss": 1.1984,
      "step": 12900
    },
    {
      "epoch": 0.29378968208815964,
      "grad_norm": 54.75,
      "learning_rate": 1.4412966747167008e-05,
      "loss": 0.7554,
      "step": 12910
    },
    {
      "epoch": 0.29401724961882436,
      "grad_norm": 6.96875,
      "learning_rate": 1.4408322496749025e-05,
      "loss": 0.8067,
      "step": 12920
    },
    {
      "epoch": 0.29424481714948914,
      "grad_norm": 370.0,
      "learning_rate": 1.4403678246331042e-05,
      "loss": 1.626,
      "step": 12930
    },
    {
      "epoch": 0.29447238468015385,
      "grad_norm": 1.1171875,
      "learning_rate": 1.4399033995913061e-05,
      "loss": 0.5988,
      "step": 12940
    },
    {
      "epoch": 0.29469995221081857,
      "grad_norm": 139.0,
      "learning_rate": 1.4394389745495078e-05,
      "loss": 1.0072,
      "step": 12950
    },
    {
      "epoch": 0.2949275197414833,
      "grad_norm": 199.0,
      "learning_rate": 1.4389745495077097e-05,
      "loss": 1.3956,
      "step": 12960
    },
    {
      "epoch": 0.295155087272148,
      "grad_norm": 62.75,
      "learning_rate": 1.4385101244659114e-05,
      "loss": 1.3001,
      "step": 12970
    },
    {
      "epoch": 0.2953826548028127,
      "grad_norm": 0.1748046875,
      "learning_rate": 1.438045699424113e-05,
      "loss": 0.5668,
      "step": 12980
    },
    {
      "epoch": 0.29561022233347745,
      "grad_norm": 207.0,
      "learning_rate": 1.4375812743823148e-05,
      "loss": 1.1521,
      "step": 12990
    },
    {
      "epoch": 0.29583778986414216,
      "grad_norm": 0.045654296875,
      "learning_rate": 1.4371168493405165e-05,
      "loss": 0.8196,
      "step": 13000
    },
    {
      "epoch": 0.2960653573948069,
      "grad_norm": 157.0,
      "learning_rate": 1.4366524242987184e-05,
      "loss": 1.2031,
      "step": 13010
    },
    {
      "epoch": 0.29629292492547166,
      "grad_norm": 74.5,
      "learning_rate": 1.4361879992569201e-05,
      "loss": 0.8377,
      "step": 13020
    },
    {
      "epoch": 0.2965204924561364,
      "grad_norm": 116.5,
      "learning_rate": 1.4357235742151218e-05,
      "loss": 0.8797,
      "step": 13030
    },
    {
      "epoch": 0.2967480599868011,
      "grad_norm": 258.0,
      "learning_rate": 1.4352591491733237e-05,
      "loss": 0.7258,
      "step": 13040
    },
    {
      "epoch": 0.2969756275174658,
      "grad_norm": 2.140625,
      "learning_rate": 1.4347947241315252e-05,
      "loss": 0.7343,
      "step": 13050
    },
    {
      "epoch": 0.29720319504813053,
      "grad_norm": 1.21875,
      "learning_rate": 1.434330299089727e-05,
      "loss": 0.7734,
      "step": 13060
    },
    {
      "epoch": 0.29743076257879525,
      "grad_norm": 14.9375,
      "learning_rate": 1.4338658740479288e-05,
      "loss": 1.6472,
      "step": 13070
    },
    {
      "epoch": 0.29765833010945997,
      "grad_norm": 9.0625,
      "learning_rate": 1.4334014490061305e-05,
      "loss": 0.4777,
      "step": 13080
    },
    {
      "epoch": 0.2978858976401247,
      "grad_norm": 0.001556396484375,
      "learning_rate": 1.4329370239643324e-05,
      "loss": 0.5379,
      "step": 13090
    },
    {
      "epoch": 0.2981134651707894,
      "grad_norm": 171.0,
      "learning_rate": 1.432472598922534e-05,
      "loss": 1.0923,
      "step": 13100
    },
    {
      "epoch": 0.2983410327014542,
      "grad_norm": 34.75,
      "learning_rate": 1.432008173880736e-05,
      "loss": 1.0435,
      "step": 13110
    },
    {
      "epoch": 0.2985686002321189,
      "grad_norm": 0.00732421875,
      "learning_rate": 1.4315437488389375e-05,
      "loss": 0.1765,
      "step": 13120
    },
    {
      "epoch": 0.2987961677627836,
      "grad_norm": 0.0003910064697265625,
      "learning_rate": 1.4310793237971392e-05,
      "loss": 1.7982,
      "step": 13130
    },
    {
      "epoch": 0.29902373529344833,
      "grad_norm": 0.05859375,
      "learning_rate": 1.4306148987553409e-05,
      "loss": 1.1083,
      "step": 13140
    },
    {
      "epoch": 0.29925130282411305,
      "grad_norm": 0.53515625,
      "learning_rate": 1.4301504737135428e-05,
      "loss": 1.2239,
      "step": 13150
    },
    {
      "epoch": 0.29947887035477777,
      "grad_norm": 0.0096435546875,
      "learning_rate": 1.4296860486717445e-05,
      "loss": 0.3278,
      "step": 13160
    },
    {
      "epoch": 0.2997064378854425,
      "grad_norm": 276.0,
      "learning_rate": 1.4292216236299464e-05,
      "loss": 0.9661,
      "step": 13170
    },
    {
      "epoch": 0.2999340054161072,
      "grad_norm": 284.0,
      "learning_rate": 1.428757198588148e-05,
      "loss": 0.7225,
      "step": 13180
    },
    {
      "epoch": 0.300161572946772,
      "grad_norm": 0.005279541015625,
      "learning_rate": 1.4282927735463496e-05,
      "loss": 1.3146,
      "step": 13190
    },
    {
      "epoch": 0.3003891404774367,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.4278283485045515e-05,
      "loss": 0.4293,
      "step": 13200
    },
    {
      "epoch": 0.3006167080081014,
      "grad_norm": 68.0,
      "learning_rate": 1.4273639234627532e-05,
      "loss": 0.6278,
      "step": 13210
    },
    {
      "epoch": 0.30084427553876614,
      "grad_norm": 100.5,
      "learning_rate": 1.426899498420955e-05,
      "loss": 0.6124,
      "step": 13220
    },
    {
      "epoch": 0.30107184306943086,
      "grad_norm": 247.0,
      "learning_rate": 1.4264350733791568e-05,
      "loss": 0.7771,
      "step": 13230
    },
    {
      "epoch": 0.3012994106000956,
      "grad_norm": 0.296875,
      "learning_rate": 1.4259706483373585e-05,
      "loss": 0.6646,
      "step": 13240
    },
    {
      "epoch": 0.3015269781307603,
      "grad_norm": 0.01275634765625,
      "learning_rate": 1.4255062232955603e-05,
      "loss": 1.1415,
      "step": 13250
    },
    {
      "epoch": 0.301754545661425,
      "grad_norm": 0.0203857421875,
      "learning_rate": 1.4250417982537619e-05,
      "loss": 1.4797,
      "step": 13260
    },
    {
      "epoch": 0.30198211319208973,
      "grad_norm": 0.000583648681640625,
      "learning_rate": 1.4245773732119636e-05,
      "loss": 0.2479,
      "step": 13270
    },
    {
      "epoch": 0.3022096807227545,
      "grad_norm": 146.0,
      "learning_rate": 1.4241129481701655e-05,
      "loss": 1.4586,
      "step": 13280
    },
    {
      "epoch": 0.3024372482534192,
      "grad_norm": 77.5,
      "learning_rate": 1.4236485231283672e-05,
      "loss": 0.5281,
      "step": 13290
    },
    {
      "epoch": 0.30266481578408394,
      "grad_norm": 308.0,
      "learning_rate": 1.423184098086569e-05,
      "loss": 1.3738,
      "step": 13300
    },
    {
      "epoch": 0.30289238331474866,
      "grad_norm": 400.0,
      "learning_rate": 1.4227196730447707e-05,
      "loss": 1.9557,
      "step": 13310
    },
    {
      "epoch": 0.3031199508454134,
      "grad_norm": 90.0,
      "learning_rate": 1.4222552480029726e-05,
      "loss": 0.5008,
      "step": 13320
    },
    {
      "epoch": 0.3033475183760781,
      "grad_norm": 0.0020751953125,
      "learning_rate": 1.4217908229611742e-05,
      "loss": 0.7592,
      "step": 13330
    },
    {
      "epoch": 0.3035750859067428,
      "grad_norm": 181.0,
      "learning_rate": 1.4213263979193759e-05,
      "loss": 0.7766,
      "step": 13340
    },
    {
      "epoch": 0.30380265343740753,
      "grad_norm": 140.0,
      "learning_rate": 1.4208619728775777e-05,
      "loss": 1.2421,
      "step": 13350
    },
    {
      "epoch": 0.30403022096807225,
      "grad_norm": 19.75,
      "learning_rate": 1.4203975478357794e-05,
      "loss": 0.4028,
      "step": 13360
    },
    {
      "epoch": 0.304257788498737,
      "grad_norm": 0.2080078125,
      "learning_rate": 1.4199331227939811e-05,
      "loss": 0.9126,
      "step": 13370
    },
    {
      "epoch": 0.30448535602940174,
      "grad_norm": 95.5,
      "learning_rate": 1.419468697752183e-05,
      "loss": 0.6843,
      "step": 13380
    },
    {
      "epoch": 0.30471292356006646,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.4190042727103847e-05,
      "loss": 1.6616,
      "step": 13390
    },
    {
      "epoch": 0.3049404910907312,
      "grad_norm": 28.625,
      "learning_rate": 1.4185398476685863e-05,
      "loss": 0.4716,
      "step": 13400
    },
    {
      "epoch": 0.3051680586213959,
      "grad_norm": 90.5,
      "learning_rate": 1.4180754226267881e-05,
      "loss": 0.6886,
      "step": 13410
    },
    {
      "epoch": 0.3053956261520606,
      "grad_norm": 45.25,
      "learning_rate": 1.4176109975849898e-05,
      "loss": 0.5729,
      "step": 13420
    },
    {
      "epoch": 0.30562319368272534,
      "grad_norm": 50.75,
      "learning_rate": 1.4171465725431917e-05,
      "loss": 0.2861,
      "step": 13430
    },
    {
      "epoch": 0.30585076121339005,
      "grad_norm": 0.000698089599609375,
      "learning_rate": 1.4166821475013934e-05,
      "loss": 0.8428,
      "step": 13440
    },
    {
      "epoch": 0.3060783287440548,
      "grad_norm": 216.0,
      "learning_rate": 1.4162177224595951e-05,
      "loss": 1.6182,
      "step": 13450
    },
    {
      "epoch": 0.30630589627471955,
      "grad_norm": 310.0,
      "learning_rate": 1.415753297417797e-05,
      "loss": 1.1133,
      "step": 13460
    },
    {
      "epoch": 0.30653346380538427,
      "grad_norm": 37.25,
      "learning_rate": 1.4152888723759985e-05,
      "loss": 0.493,
      "step": 13470
    },
    {
      "epoch": 0.306761031336049,
      "grad_norm": 85.5,
      "learning_rate": 1.4148244473342002e-05,
      "loss": 1.0833,
      "step": 13480
    },
    {
      "epoch": 0.3069885988667137,
      "grad_norm": 1.1484375,
      "learning_rate": 1.4143600222924021e-05,
      "loss": 0.9036,
      "step": 13490
    },
    {
      "epoch": 0.3072161663973784,
      "grad_norm": 0.0031585693359375,
      "learning_rate": 1.4138955972506038e-05,
      "loss": 0.6097,
      "step": 13500
    },
    {
      "epoch": 0.30744373392804314,
      "grad_norm": 91.5,
      "learning_rate": 1.4134311722088057e-05,
      "loss": 1.5184,
      "step": 13510
    },
    {
      "epoch": 0.30767130145870786,
      "grad_norm": 0.004730224609375,
      "learning_rate": 1.4129667471670074e-05,
      "loss": 0.5837,
      "step": 13520
    },
    {
      "epoch": 0.3078988689893726,
      "grad_norm": 169.0,
      "learning_rate": 1.4125023221252093e-05,
      "loss": 1.4995,
      "step": 13530
    },
    {
      "epoch": 0.3081264365200373,
      "grad_norm": 101.5,
      "learning_rate": 1.4120378970834108e-05,
      "loss": 0.7159,
      "step": 13540
    },
    {
      "epoch": 0.30835400405070207,
      "grad_norm": 0.068359375,
      "learning_rate": 1.4115734720416125e-05,
      "loss": 0.5835,
      "step": 13550
    },
    {
      "epoch": 0.3085815715813668,
      "grad_norm": 0.15234375,
      "learning_rate": 1.4111090469998144e-05,
      "loss": 0.6386,
      "step": 13560
    },
    {
      "epoch": 0.3088091391120315,
      "grad_norm": 0.09765625,
      "learning_rate": 1.4106446219580161e-05,
      "loss": 0.0486,
      "step": 13570
    },
    {
      "epoch": 0.3090367066426962,
      "grad_norm": 113.5,
      "learning_rate": 1.4101801969162178e-05,
      "loss": 0.8683,
      "step": 13580
    },
    {
      "epoch": 0.30926427417336094,
      "grad_norm": 0.796875,
      "learning_rate": 1.4097157718744197e-05,
      "loss": 0.8838,
      "step": 13590
    },
    {
      "epoch": 0.30949184170402566,
      "grad_norm": 17.125,
      "learning_rate": 1.4092513468326214e-05,
      "loss": 1.3741,
      "step": 13600
    },
    {
      "epoch": 0.3097194092346904,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.408786921790823e-05,
      "loss": 0.9884,
      "step": 13610
    },
    {
      "epoch": 0.3099469767653551,
      "grad_norm": 172.0,
      "learning_rate": 1.4083224967490248e-05,
      "loss": 0.7601,
      "step": 13620
    },
    {
      "epoch": 0.3101745442960198,
      "grad_norm": 0.0390625,
      "learning_rate": 1.4078580717072265e-05,
      "loss": 1.0375,
      "step": 13630
    },
    {
      "epoch": 0.3104021118266846,
      "grad_norm": 2.453125,
      "learning_rate": 1.4073936466654284e-05,
      "loss": 0.5709,
      "step": 13640
    },
    {
      "epoch": 0.3106296793573493,
      "grad_norm": 10.625,
      "learning_rate": 1.4069292216236301e-05,
      "loss": 0.8745,
      "step": 13650
    },
    {
      "epoch": 0.310857246888014,
      "grad_norm": 0.041748046875,
      "learning_rate": 1.406464796581832e-05,
      "loss": 0.7077,
      "step": 13660
    },
    {
      "epoch": 0.31108481441867875,
      "grad_norm": 0.0005950927734375,
      "learning_rate": 1.4060003715400337e-05,
      "loss": 1.3283,
      "step": 13670
    },
    {
      "epoch": 0.31131238194934346,
      "grad_norm": 204.0,
      "learning_rate": 1.4055359464982352e-05,
      "loss": 1.1514,
      "step": 13680
    },
    {
      "epoch": 0.3115399494800082,
      "grad_norm": 0.2451171875,
      "learning_rate": 1.405071521456437e-05,
      "loss": 0.8503,
      "step": 13690
    },
    {
      "epoch": 0.3117675170106729,
      "grad_norm": 0.0028533935546875,
      "learning_rate": 1.4046070964146388e-05,
      "loss": 0.5315,
      "step": 13700
    },
    {
      "epoch": 0.3119950845413376,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.4041426713728405e-05,
      "loss": 0.6903,
      "step": 13710
    },
    {
      "epoch": 0.3122226520720024,
      "grad_norm": 182.0,
      "learning_rate": 1.4036782463310424e-05,
      "loss": 1.4152,
      "step": 13720
    },
    {
      "epoch": 0.3124502196026671,
      "grad_norm": 28.25,
      "learning_rate": 1.403213821289244e-05,
      "loss": 0.7724,
      "step": 13730
    },
    {
      "epoch": 0.31267778713333183,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.402749396247446e-05,
      "loss": 0.4895,
      "step": 13740
    },
    {
      "epoch": 0.31290535466399655,
      "grad_norm": 1.578125,
      "learning_rate": 1.4022849712056475e-05,
      "loss": 0.5659,
      "step": 13750
    },
    {
      "epoch": 0.31313292219466127,
      "grad_norm": 374.0,
      "learning_rate": 1.4018205461638492e-05,
      "loss": 1.6393,
      "step": 13760
    },
    {
      "epoch": 0.313360489725326,
      "grad_norm": 46.0,
      "learning_rate": 1.401356121122051e-05,
      "loss": 1.1746,
      "step": 13770
    },
    {
      "epoch": 0.3135880572559907,
      "grad_norm": 75.0,
      "learning_rate": 1.4008916960802528e-05,
      "loss": 0.4692,
      "step": 13780
    },
    {
      "epoch": 0.3138156247866554,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.4004272710384545e-05,
      "loss": 1.1364,
      "step": 13790
    },
    {
      "epoch": 0.31404319231732014,
      "grad_norm": 28.125,
      "learning_rate": 1.3999628459966563e-05,
      "loss": 0.4576,
      "step": 13800
    },
    {
      "epoch": 0.3142707598479849,
      "grad_norm": 0.185546875,
      "learning_rate": 1.399498420954858e-05,
      "loss": 1.1306,
      "step": 13810
    },
    {
      "epoch": 0.31449832737864963,
      "grad_norm": 157.0,
      "learning_rate": 1.3990339959130596e-05,
      "loss": 0.6756,
      "step": 13820
    },
    {
      "epoch": 0.31472589490931435,
      "grad_norm": 0.064453125,
      "learning_rate": 1.3985695708712615e-05,
      "loss": 0.7268,
      "step": 13830
    },
    {
      "epoch": 0.31495346243997907,
      "grad_norm": 212.0,
      "learning_rate": 1.3981051458294632e-05,
      "loss": 0.4636,
      "step": 13840
    },
    {
      "epoch": 0.3151810299706438,
      "grad_norm": 442.0,
      "learning_rate": 1.397640720787665e-05,
      "loss": 1.8877,
      "step": 13850
    },
    {
      "epoch": 0.3154085975013085,
      "grad_norm": 193.0,
      "learning_rate": 1.3971762957458667e-05,
      "loss": 1.1208,
      "step": 13860
    },
    {
      "epoch": 0.3156361650319732,
      "grad_norm": 0.11474609375,
      "learning_rate": 1.3967118707040686e-05,
      "loss": 1.0451,
      "step": 13870
    },
    {
      "epoch": 0.31586373256263794,
      "grad_norm": 73.0,
      "learning_rate": 1.3962474456622703e-05,
      "loss": 1.3015,
      "step": 13880
    },
    {
      "epoch": 0.31609130009330266,
      "grad_norm": 98.0,
      "learning_rate": 1.3957830206204719e-05,
      "loss": 1.0351,
      "step": 13890
    },
    {
      "epoch": 0.31631886762396744,
      "grad_norm": 0.035400390625,
      "learning_rate": 1.3953185955786737e-05,
      "loss": 0.6703,
      "step": 13900
    },
    {
      "epoch": 0.31654643515463216,
      "grad_norm": 668.0,
      "learning_rate": 1.3948541705368754e-05,
      "loss": 0.6189,
      "step": 13910
    },
    {
      "epoch": 0.3167740026852969,
      "grad_norm": 8.1875,
      "learning_rate": 1.3943897454950772e-05,
      "loss": 0.8977,
      "step": 13920
    },
    {
      "epoch": 0.3170015702159616,
      "grad_norm": 10.6875,
      "learning_rate": 1.393925320453279e-05,
      "loss": 1.0117,
      "step": 13930
    },
    {
      "epoch": 0.3172291377466263,
      "grad_norm": 0.00933837890625,
      "learning_rate": 1.3934608954114807e-05,
      "loss": 0.579,
      "step": 13940
    },
    {
      "epoch": 0.31745670527729103,
      "grad_norm": 0.55859375,
      "learning_rate": 1.3929964703696826e-05,
      "loss": 0.4619,
      "step": 13950
    },
    {
      "epoch": 0.31768427280795575,
      "grad_norm": 157.0,
      "learning_rate": 1.3925320453278841e-05,
      "loss": 1.2644,
      "step": 13960
    },
    {
      "epoch": 0.31791184033862047,
      "grad_norm": 0.00106048583984375,
      "learning_rate": 1.3920676202860858e-05,
      "loss": 1.1834,
      "step": 13970
    },
    {
      "epoch": 0.3181394078692852,
      "grad_norm": 302.0,
      "learning_rate": 1.3916031952442877e-05,
      "loss": 1.6274,
      "step": 13980
    },
    {
      "epoch": 0.31836697539994996,
      "grad_norm": 16.125,
      "learning_rate": 1.3911387702024894e-05,
      "loss": 0.2547,
      "step": 13990
    },
    {
      "epoch": 0.3185945429306147,
      "grad_norm": 374.0,
      "learning_rate": 1.3906743451606913e-05,
      "loss": 1.3145,
      "step": 14000
    },
    {
      "epoch": 0.3188221104612794,
      "grad_norm": 0.004974365234375,
      "learning_rate": 1.390209920118893e-05,
      "loss": 0.8133,
      "step": 14010
    },
    {
      "epoch": 0.3190496779919441,
      "grad_norm": 0.09814453125,
      "learning_rate": 1.3897454950770947e-05,
      "loss": 0.5336,
      "step": 14020
    },
    {
      "epoch": 0.31927724552260883,
      "grad_norm": 159.0,
      "learning_rate": 1.3892810700352962e-05,
      "loss": 0.6873,
      "step": 14030
    },
    {
      "epoch": 0.31950481305327355,
      "grad_norm": 0.0294189453125,
      "learning_rate": 1.3888166449934981e-05,
      "loss": 1.072,
      "step": 14040
    },
    {
      "epoch": 0.31973238058393827,
      "grad_norm": 56.5,
      "learning_rate": 1.3883522199516998e-05,
      "loss": 0.7856,
      "step": 14050
    },
    {
      "epoch": 0.319959948114603,
      "grad_norm": 376.0,
      "learning_rate": 1.3878877949099017e-05,
      "loss": 0.5478,
      "step": 14060
    },
    {
      "epoch": 0.3201875156452677,
      "grad_norm": 51.25,
      "learning_rate": 1.3874233698681034e-05,
      "loss": 0.5286,
      "step": 14070
    },
    {
      "epoch": 0.3204150831759325,
      "grad_norm": 128.0,
      "learning_rate": 1.3869589448263053e-05,
      "loss": 1.1629,
      "step": 14080
    },
    {
      "epoch": 0.3206426507065972,
      "grad_norm": 280.0,
      "learning_rate": 1.386494519784507e-05,
      "loss": 1.0957,
      "step": 14090
    },
    {
      "epoch": 0.3208702182372619,
      "grad_norm": 0.022216796875,
      "learning_rate": 1.3860300947427085e-05,
      "loss": 0.6363,
      "step": 14100
    },
    {
      "epoch": 0.32109778576792664,
      "grad_norm": 103.0,
      "learning_rate": 1.3855656697009104e-05,
      "loss": 0.6626,
      "step": 14110
    },
    {
      "epoch": 0.32132535329859135,
      "grad_norm": 438.0,
      "learning_rate": 1.3851012446591121e-05,
      "loss": 0.7413,
      "step": 14120
    },
    {
      "epoch": 0.3215529208292561,
      "grad_norm": 264.0,
      "learning_rate": 1.3846368196173138e-05,
      "loss": 0.3535,
      "step": 14130
    },
    {
      "epoch": 0.3217804883599208,
      "grad_norm": 0.000667572021484375,
      "learning_rate": 1.3841723945755157e-05,
      "loss": 1.7759,
      "step": 14140
    },
    {
      "epoch": 0.3220080558905855,
      "grad_norm": 0.51171875,
      "learning_rate": 1.3837079695337174e-05,
      "loss": 1.8171,
      "step": 14150
    },
    {
      "epoch": 0.3222356234212503,
      "grad_norm": 60.75,
      "learning_rate": 1.3832435444919193e-05,
      "loss": 0.8028,
      "step": 14160
    },
    {
      "epoch": 0.322463190951915,
      "grad_norm": 0.054931640625,
      "learning_rate": 1.3827791194501208e-05,
      "loss": 0.7042,
      "step": 14170
    },
    {
      "epoch": 0.3226907584825797,
      "grad_norm": 340.0,
      "learning_rate": 1.3823146944083225e-05,
      "loss": 1.7306,
      "step": 14180
    },
    {
      "epoch": 0.32291832601324444,
      "grad_norm": 0.005157470703125,
      "learning_rate": 1.3818502693665244e-05,
      "loss": 1.1742,
      "step": 14190
    },
    {
      "epoch": 0.32314589354390916,
      "grad_norm": 386.0,
      "learning_rate": 1.3813858443247261e-05,
      "loss": 0.5203,
      "step": 14200
    },
    {
      "epoch": 0.3233734610745739,
      "grad_norm": 115.0,
      "learning_rate": 1.380921419282928e-05,
      "loss": 0.6467,
      "step": 14210
    },
    {
      "epoch": 0.3236010286052386,
      "grad_norm": 0.003692626953125,
      "learning_rate": 1.3804569942411297e-05,
      "loss": 0.9118,
      "step": 14220
    },
    {
      "epoch": 0.3238285961359033,
      "grad_norm": 235.0,
      "learning_rate": 1.3799925691993314e-05,
      "loss": 1.3569,
      "step": 14230
    },
    {
      "epoch": 0.32405616366656803,
      "grad_norm": 0.0028839111328125,
      "learning_rate": 1.379528144157533e-05,
      "loss": 1.1865,
      "step": 14240
    },
    {
      "epoch": 0.3242837311972328,
      "grad_norm": 32.75,
      "learning_rate": 1.3790637191157348e-05,
      "loss": 1.4558,
      "step": 14250
    },
    {
      "epoch": 0.3245112987278975,
      "grad_norm": 0.3984375,
      "learning_rate": 1.3785992940739365e-05,
      "loss": 0.7439,
      "step": 14260
    },
    {
      "epoch": 0.32473886625856224,
      "grad_norm": 77.0,
      "learning_rate": 1.3781348690321384e-05,
      "loss": 0.6287,
      "step": 14270
    },
    {
      "epoch": 0.32496643378922696,
      "grad_norm": 0.0074462890625,
      "learning_rate": 1.37767044399034e-05,
      "loss": 1.001,
      "step": 14280
    },
    {
      "epoch": 0.3251940013198917,
      "grad_norm": 0.002288818359375,
      "learning_rate": 1.377206018948542e-05,
      "loss": 0.6973,
      "step": 14290
    },
    {
      "epoch": 0.3254215688505564,
      "grad_norm": 284.0,
      "learning_rate": 1.3767415939067437e-05,
      "loss": 0.8186,
      "step": 14300
    },
    {
      "epoch": 0.3256491363812211,
      "grad_norm": 0.482421875,
      "learning_rate": 1.3762771688649452e-05,
      "loss": 0.4451,
      "step": 14310
    },
    {
      "epoch": 0.32587670391188583,
      "grad_norm": 87.5,
      "learning_rate": 1.375812743823147e-05,
      "loss": 1.2707,
      "step": 14320
    },
    {
      "epoch": 0.32610427144255055,
      "grad_norm": 318.0,
      "learning_rate": 1.3753483187813488e-05,
      "loss": 1.6368,
      "step": 14330
    },
    {
      "epoch": 0.3263318389732153,
      "grad_norm": 49.75,
      "learning_rate": 1.3748838937395505e-05,
      "loss": 0.5435,
      "step": 14340
    },
    {
      "epoch": 0.32655940650388005,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.3744194686977524e-05,
      "loss": 0.7773,
      "step": 14350
    },
    {
      "epoch": 0.32678697403454476,
      "grad_norm": 0.039306640625,
      "learning_rate": 1.373955043655954e-05,
      "loss": 0.9554,
      "step": 14360
    },
    {
      "epoch": 0.3270145415652095,
      "grad_norm": 71.0,
      "learning_rate": 1.373490618614156e-05,
      "loss": 0.8848,
      "step": 14370
    },
    {
      "epoch": 0.3272421090958742,
      "grad_norm": 1.296875,
      "learning_rate": 1.3730261935723575e-05,
      "loss": 0.3795,
      "step": 14380
    },
    {
      "epoch": 0.3274696766265389,
      "grad_norm": 3.34375,
      "learning_rate": 1.3725617685305592e-05,
      "loss": 1.2179,
      "step": 14390
    },
    {
      "epoch": 0.32769724415720364,
      "grad_norm": 243.0,
      "learning_rate": 1.372097343488761e-05,
      "loss": 1.0491,
      "step": 14400
    },
    {
      "epoch": 0.32792481168786836,
      "grad_norm": 229.0,
      "learning_rate": 1.3716329184469628e-05,
      "loss": 1.0979,
      "step": 14410
    },
    {
      "epoch": 0.3281523792185331,
      "grad_norm": 0.11669921875,
      "learning_rate": 1.3711684934051646e-05,
      "loss": 0.8361,
      "step": 14420
    },
    {
      "epoch": 0.32837994674919785,
      "grad_norm": 56.0,
      "learning_rate": 1.3707040683633663e-05,
      "loss": 0.9015,
      "step": 14430
    },
    {
      "epoch": 0.32860751427986257,
      "grad_norm": 0.00384521484375,
      "learning_rate": 1.370239643321568e-05,
      "loss": 1.1401,
      "step": 14440
    },
    {
      "epoch": 0.3288350818105273,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.3697752182797697e-05,
      "loss": 0.7148,
      "step": 14450
    },
    {
      "epoch": 0.329062649341192,
      "grad_norm": 0.0054931640625,
      "learning_rate": 1.3693107932379714e-05,
      "loss": 0.8983,
      "step": 14460
    },
    {
      "epoch": 0.3292902168718567,
      "grad_norm": 139.0,
      "learning_rate": 1.3688463681961732e-05,
      "loss": 2.2297,
      "step": 14470
    },
    {
      "epoch": 0.32951778440252144,
      "grad_norm": 294.0,
      "learning_rate": 1.368381943154375e-05,
      "loss": 0.6347,
      "step": 14480
    },
    {
      "epoch": 0.32974535193318616,
      "grad_norm": 67.0,
      "learning_rate": 1.3679175181125767e-05,
      "loss": 0.333,
      "step": 14490
    },
    {
      "epoch": 0.3299729194638509,
      "grad_norm": 72.0,
      "learning_rate": 1.3674530930707786e-05,
      "loss": 0.8417,
      "step": 14500
    },
    {
      "epoch": 0.3302004869945156,
      "grad_norm": 51.25,
      "learning_rate": 1.3669886680289803e-05,
      "loss": 1.0233,
      "step": 14510
    },
    {
      "epoch": 0.33042805452518037,
      "grad_norm": 0.0247802734375,
      "learning_rate": 1.3665242429871819e-05,
      "loss": 0.5366,
      "step": 14520
    },
    {
      "epoch": 0.3306556220558451,
      "grad_norm": 284.0,
      "learning_rate": 1.3660598179453837e-05,
      "loss": 1.2361,
      "step": 14530
    },
    {
      "epoch": 0.3308831895865098,
      "grad_norm": 135.0,
      "learning_rate": 1.3655953929035854e-05,
      "loss": 0.8806,
      "step": 14540
    },
    {
      "epoch": 0.3311107571171745,
      "grad_norm": 0.01171875,
      "learning_rate": 1.3651309678617873e-05,
      "loss": 0.6159,
      "step": 14550
    },
    {
      "epoch": 0.33133832464783924,
      "grad_norm": 472.0,
      "learning_rate": 1.364666542819989e-05,
      "loss": 0.3933,
      "step": 14560
    },
    {
      "epoch": 0.33156589217850396,
      "grad_norm": 3.8125,
      "learning_rate": 1.3642021177781907e-05,
      "loss": 0.8006,
      "step": 14570
    },
    {
      "epoch": 0.3317934597091687,
      "grad_norm": 278.0,
      "learning_rate": 1.3637376927363926e-05,
      "loss": 0.5984,
      "step": 14580
    },
    {
      "epoch": 0.3320210272398334,
      "grad_norm": 251.0,
      "learning_rate": 1.3632732676945941e-05,
      "loss": 1.2876,
      "step": 14590
    },
    {
      "epoch": 0.3322485947704981,
      "grad_norm": 0.11083984375,
      "learning_rate": 1.3628088426527958e-05,
      "loss": 0.9876,
      "step": 14600
    },
    {
      "epoch": 0.3324761623011629,
      "grad_norm": 174.0,
      "learning_rate": 1.3623444176109977e-05,
      "loss": 0.8395,
      "step": 14610
    },
    {
      "epoch": 0.3327037298318276,
      "grad_norm": 91.5,
      "learning_rate": 1.3618799925691994e-05,
      "loss": 0.427,
      "step": 14620
    },
    {
      "epoch": 0.33293129736249233,
      "grad_norm": 6.375,
      "learning_rate": 1.3614155675274013e-05,
      "loss": 1.1241,
      "step": 14630
    },
    {
      "epoch": 0.33315886489315705,
      "grad_norm": 217.0,
      "learning_rate": 1.360951142485603e-05,
      "loss": 1.4988,
      "step": 14640
    },
    {
      "epoch": 0.33338643242382177,
      "grad_norm": 264.0,
      "learning_rate": 1.3604867174438047e-05,
      "loss": 1.0041,
      "step": 14650
    },
    {
      "epoch": 0.3336139999544865,
      "grad_norm": 66.5,
      "learning_rate": 1.3600222924020064e-05,
      "loss": 0.5737,
      "step": 14660
    },
    {
      "epoch": 0.3338415674851512,
      "grad_norm": 296.0,
      "learning_rate": 1.3595578673602081e-05,
      "loss": 0.6479,
      "step": 14670
    },
    {
      "epoch": 0.3340691350158159,
      "grad_norm": 272.0,
      "learning_rate": 1.3590934423184098e-05,
      "loss": 1.3263,
      "step": 14680
    },
    {
      "epoch": 0.3342967025464807,
      "grad_norm": 55.75,
      "learning_rate": 1.3586290172766117e-05,
      "loss": 1.2984,
      "step": 14690
    },
    {
      "epoch": 0.3345242700771454,
      "grad_norm": 212.0,
      "learning_rate": 1.3581645922348134e-05,
      "loss": 1.2183,
      "step": 14700
    },
    {
      "epoch": 0.33475183760781013,
      "grad_norm": 0.0005645751953125,
      "learning_rate": 1.3577001671930153e-05,
      "loss": 1.0082,
      "step": 14710
    },
    {
      "epoch": 0.33497940513847485,
      "grad_norm": 136.0,
      "learning_rate": 1.357235742151217e-05,
      "loss": 0.6401,
      "step": 14720
    },
    {
      "epoch": 0.33520697266913957,
      "grad_norm": 199.0,
      "learning_rate": 1.3567713171094185e-05,
      "loss": 1.0214,
      "step": 14730
    },
    {
      "epoch": 0.3354345401998043,
      "grad_norm": 11.25,
      "learning_rate": 1.3563068920676204e-05,
      "loss": 0.798,
      "step": 14740
    },
    {
      "epoch": 0.335662107730469,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.3558424670258221e-05,
      "loss": 0.0564,
      "step": 14750
    },
    {
      "epoch": 0.3358896752611337,
      "grad_norm": 398.0,
      "learning_rate": 1.355378041984024e-05,
      "loss": 1.2777,
      "step": 14760
    },
    {
      "epoch": 0.33611724279179844,
      "grad_norm": 0.6171875,
      "learning_rate": 1.3549136169422257e-05,
      "loss": 0.2351,
      "step": 14770
    },
    {
      "epoch": 0.3363448103224632,
      "grad_norm": 336.0,
      "learning_rate": 1.3544491919004274e-05,
      "loss": 2.5492,
      "step": 14780
    },
    {
      "epoch": 0.33657237785312794,
      "grad_norm": 0.004791259765625,
      "learning_rate": 1.3539847668586293e-05,
      "loss": 0.8467,
      "step": 14790
    },
    {
      "epoch": 0.33679994538379265,
      "grad_norm": 0.01129150390625,
      "learning_rate": 1.3535203418168308e-05,
      "loss": 0.4928,
      "step": 14800
    },
    {
      "epoch": 0.3370275129144574,
      "grad_norm": 102.0,
      "learning_rate": 1.3530559167750325e-05,
      "loss": 0.9173,
      "step": 14810
    },
    {
      "epoch": 0.3372550804451221,
      "grad_norm": 0.07275390625,
      "learning_rate": 1.3525914917332344e-05,
      "loss": 0.2405,
      "step": 14820
    },
    {
      "epoch": 0.3374826479757868,
      "grad_norm": 0.0027313232421875,
      "learning_rate": 1.352127066691436e-05,
      "loss": 0.6303,
      "step": 14830
    },
    {
      "epoch": 0.3377102155064515,
      "grad_norm": 143.0,
      "learning_rate": 1.351662641649638e-05,
      "loss": 0.7115,
      "step": 14840
    },
    {
      "epoch": 0.33793778303711625,
      "grad_norm": 165.0,
      "learning_rate": 1.3511982166078397e-05,
      "loss": 1.4319,
      "step": 14850
    },
    {
      "epoch": 0.33816535056778096,
      "grad_norm": 7.84375,
      "learning_rate": 1.3507337915660415e-05,
      "loss": 0.7734,
      "step": 14860
    },
    {
      "epoch": 0.33839291809844574,
      "grad_norm": 2.03125,
      "learning_rate": 1.350269366524243e-05,
      "loss": 0.3553,
      "step": 14870
    },
    {
      "epoch": 0.33862048562911046,
      "grad_norm": 1.109375,
      "learning_rate": 1.3498049414824448e-05,
      "loss": 1.5449,
      "step": 14880
    },
    {
      "epoch": 0.3388480531597752,
      "grad_norm": 11.0,
      "learning_rate": 1.3493405164406465e-05,
      "loss": 0.1157,
      "step": 14890
    },
    {
      "epoch": 0.3390756206904399,
      "grad_norm": 42.5,
      "learning_rate": 1.3488760913988484e-05,
      "loss": 0.2758,
      "step": 14900
    },
    {
      "epoch": 0.3393031882211046,
      "grad_norm": 438.0,
      "learning_rate": 1.34841166635705e-05,
      "loss": 0.7988,
      "step": 14910
    },
    {
      "epoch": 0.33953075575176933,
      "grad_norm": 0.018310546875,
      "learning_rate": 1.347947241315252e-05,
      "loss": 1.3315,
      "step": 14920
    },
    {
      "epoch": 0.33975832328243405,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.3474828162734536e-05,
      "loss": 1.3078,
      "step": 14930
    },
    {
      "epoch": 0.33998589081309877,
      "grad_norm": 142.0,
      "learning_rate": 1.3470183912316552e-05,
      "loss": 1.1359,
      "step": 14940
    },
    {
      "epoch": 0.3402134583437635,
      "grad_norm": 132.0,
      "learning_rate": 1.346553966189857e-05,
      "loss": 1.3453,
      "step": 14950
    },
    {
      "epoch": 0.34044102587442826,
      "grad_norm": 150.0,
      "learning_rate": 1.3460895411480588e-05,
      "loss": 1.2023,
      "step": 14960
    },
    {
      "epoch": 0.340668593405093,
      "grad_norm": 17.5,
      "learning_rate": 1.3456251161062606e-05,
      "loss": 0.8902,
      "step": 14970
    },
    {
      "epoch": 0.3408961609357577,
      "grad_norm": 278.0,
      "learning_rate": 1.3451606910644623e-05,
      "loss": 0.9084,
      "step": 14980
    },
    {
      "epoch": 0.3411237284664224,
      "grad_norm": 0.875,
      "learning_rate": 1.344696266022664e-05,
      "loss": 1.4867,
      "step": 14990
    },
    {
      "epoch": 0.34135129599708713,
      "grad_norm": 24.375,
      "learning_rate": 1.344231840980866e-05,
      "loss": 1.5629,
      "step": 15000
    },
    {
      "epoch": 0.34157886352775185,
      "grad_norm": 0.007598876953125,
      "learning_rate": 1.3437674159390675e-05,
      "loss": 0.5011,
      "step": 15010
    },
    {
      "epoch": 0.34180643105841657,
      "grad_norm": 165.0,
      "learning_rate": 1.3433029908972692e-05,
      "loss": 0.4516,
      "step": 15020
    },
    {
      "epoch": 0.3420339985890813,
      "grad_norm": 71.0,
      "learning_rate": 1.342838565855471e-05,
      "loss": 0.3223,
      "step": 15030
    },
    {
      "epoch": 0.342261566119746,
      "grad_norm": 1.8515625,
      "learning_rate": 1.3423741408136727e-05,
      "loss": 0.7779,
      "step": 15040
    },
    {
      "epoch": 0.3424891336504108,
      "grad_norm": 111.0,
      "learning_rate": 1.3419097157718746e-05,
      "loss": 1.4144,
      "step": 15050
    },
    {
      "epoch": 0.3427167011810755,
      "grad_norm": 128.0,
      "learning_rate": 1.3414452907300763e-05,
      "loss": 1.0807,
      "step": 15060
    },
    {
      "epoch": 0.3429442687117402,
      "grad_norm": 48.0,
      "learning_rate": 1.3409808656882782e-05,
      "loss": 0.9897,
      "step": 15070
    },
    {
      "epoch": 0.34317183624240494,
      "grad_norm": 39.75,
      "learning_rate": 1.3405164406464797e-05,
      "loss": 0.6233,
      "step": 15080
    },
    {
      "epoch": 0.34339940377306966,
      "grad_norm": 4.75,
      "learning_rate": 1.3400520156046814e-05,
      "loss": 0.7961,
      "step": 15090
    },
    {
      "epoch": 0.3436269713037344,
      "grad_norm": 27.625,
      "learning_rate": 1.3395875905628833e-05,
      "loss": 0.258,
      "step": 15100
    },
    {
      "epoch": 0.3438545388343991,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.339123165521085e-05,
      "loss": 0.437,
      "step": 15110
    },
    {
      "epoch": 0.3440821063650638,
      "grad_norm": 0.00445556640625,
      "learning_rate": 1.3386587404792867e-05,
      "loss": 0.9132,
      "step": 15120
    },
    {
      "epoch": 0.3443096738957286,
      "grad_norm": 50.0,
      "learning_rate": 1.3381943154374886e-05,
      "loss": 0.4434,
      "step": 15130
    },
    {
      "epoch": 0.3445372414263933,
      "grad_norm": 123.5,
      "learning_rate": 1.3377298903956903e-05,
      "loss": 1.5043,
      "step": 15140
    },
    {
      "epoch": 0.344764808957058,
      "grad_norm": 193.0,
      "learning_rate": 1.3372654653538918e-05,
      "loss": 0.9092,
      "step": 15150
    },
    {
      "epoch": 0.34499237648772274,
      "grad_norm": 0.11572265625,
      "learning_rate": 1.3368010403120937e-05,
      "loss": 0.5477,
      "step": 15160
    },
    {
      "epoch": 0.34521994401838746,
      "grad_norm": 22.0,
      "learning_rate": 1.3363366152702954e-05,
      "loss": 0.9493,
      "step": 15170
    },
    {
      "epoch": 0.3454475115490522,
      "grad_norm": 0.01043701171875,
      "learning_rate": 1.3358721902284973e-05,
      "loss": 0.5923,
      "step": 15180
    },
    {
      "epoch": 0.3456750790797169,
      "grad_norm": 233.0,
      "learning_rate": 1.335407765186699e-05,
      "loss": 0.4275,
      "step": 15190
    },
    {
      "epoch": 0.3459026466103816,
      "grad_norm": 0.00121307373046875,
      "learning_rate": 1.3349433401449009e-05,
      "loss": 0.8316,
      "step": 15200
    },
    {
      "epoch": 0.34613021414104633,
      "grad_norm": 0.0054931640625,
      "learning_rate": 1.3344789151031026e-05,
      "loss": 0.6201,
      "step": 15210
    },
    {
      "epoch": 0.3463577816717111,
      "grad_norm": 388.0,
      "learning_rate": 1.3340144900613041e-05,
      "loss": 1.2672,
      "step": 15220
    },
    {
      "epoch": 0.3465853492023758,
      "grad_norm": 140.0,
      "learning_rate": 1.3335500650195058e-05,
      "loss": 1.0552,
      "step": 15230
    },
    {
      "epoch": 0.34681291673304054,
      "grad_norm": 0.09619140625,
      "learning_rate": 1.3330856399777077e-05,
      "loss": 0.6529,
      "step": 15240
    },
    {
      "epoch": 0.34704048426370526,
      "grad_norm": 73.0,
      "learning_rate": 1.3326212149359094e-05,
      "loss": 0.3012,
      "step": 15250
    },
    {
      "epoch": 0.34726805179437,
      "grad_norm": 173.0,
      "learning_rate": 1.3321567898941113e-05,
      "loss": 1.1495,
      "step": 15260
    },
    {
      "epoch": 0.3474956193250347,
      "grad_norm": 386.0,
      "learning_rate": 1.331692364852313e-05,
      "loss": 1.9857,
      "step": 15270
    },
    {
      "epoch": 0.3477231868556994,
      "grad_norm": 135.0,
      "learning_rate": 1.3312279398105149e-05,
      "loss": 1.5766,
      "step": 15280
    },
    {
      "epoch": 0.34795075438636414,
      "grad_norm": 207.0,
      "learning_rate": 1.3307635147687164e-05,
      "loss": 0.8038,
      "step": 15290
    },
    {
      "epoch": 0.34817832191702885,
      "grad_norm": 548.0,
      "learning_rate": 1.3302990897269181e-05,
      "loss": 0.6681,
      "step": 15300
    },
    {
      "epoch": 0.34840588944769363,
      "grad_norm": 402.0,
      "learning_rate": 1.32983466468512e-05,
      "loss": 1.1897,
      "step": 15310
    },
    {
      "epoch": 0.34863345697835835,
      "grad_norm": 0.0184326171875,
      "learning_rate": 1.3293702396433217e-05,
      "loss": 0.909,
      "step": 15320
    },
    {
      "epoch": 0.34886102450902307,
      "grad_norm": 0.03759765625,
      "learning_rate": 1.3289058146015234e-05,
      "loss": 0.4867,
      "step": 15330
    },
    {
      "epoch": 0.3490885920396878,
      "grad_norm": 0.0179443359375,
      "learning_rate": 1.3284413895597253e-05,
      "loss": 0.1549,
      "step": 15340
    },
    {
      "epoch": 0.3493161595703525,
      "grad_norm": 67.5,
      "learning_rate": 1.327976964517927e-05,
      "loss": 2.1047,
      "step": 15350
    },
    {
      "epoch": 0.3495437271010172,
      "grad_norm": 0.003692626953125,
      "learning_rate": 1.3275125394761285e-05,
      "loss": 0.5222,
      "step": 15360
    },
    {
      "epoch": 0.34977129463168194,
      "grad_norm": 179.0,
      "learning_rate": 1.3270481144343304e-05,
      "loss": 0.6977,
      "step": 15370
    },
    {
      "epoch": 0.34999886216234666,
      "grad_norm": 46.25,
      "learning_rate": 1.326583689392532e-05,
      "loss": 0.7688,
      "step": 15380
    },
    {
      "epoch": 0.3502264296930114,
      "grad_norm": 0.00335693359375,
      "learning_rate": 1.326119264350734e-05,
      "loss": 0.0765,
      "step": 15390
    },
    {
      "epoch": 0.35045399722367615,
      "grad_norm": 3.25,
      "learning_rate": 1.3256548393089357e-05,
      "loss": 0.9989,
      "step": 15400
    },
    {
      "epoch": 0.35068156475434087,
      "grad_norm": 245.0,
      "learning_rate": 1.3251904142671375e-05,
      "loss": 1.2773,
      "step": 15410
    },
    {
      "epoch": 0.3509091322850056,
      "grad_norm": 384.0,
      "learning_rate": 1.3247259892253392e-05,
      "loss": 0.7924,
      "step": 15420
    },
    {
      "epoch": 0.3511366998156703,
      "grad_norm": 101.5,
      "learning_rate": 1.3242615641835408e-05,
      "loss": 0.5162,
      "step": 15430
    },
    {
      "epoch": 0.351364267346335,
      "grad_norm": 20.75,
      "learning_rate": 1.3237971391417427e-05,
      "loss": 1.9466,
      "step": 15440
    },
    {
      "epoch": 0.35159183487699974,
      "grad_norm": 246.0,
      "learning_rate": 1.3233327140999444e-05,
      "loss": 1.6041,
      "step": 15450
    },
    {
      "epoch": 0.35181940240766446,
      "grad_norm": 168.0,
      "learning_rate": 1.322868289058146e-05,
      "loss": 0.7777,
      "step": 15460
    },
    {
      "epoch": 0.3520469699383292,
      "grad_norm": 298.0,
      "learning_rate": 1.322403864016348e-05,
      "loss": 0.6025,
      "step": 15470
    },
    {
      "epoch": 0.3522745374689939,
      "grad_norm": 0.0272216796875,
      "learning_rate": 1.3219394389745496e-05,
      "loss": 0.895,
      "step": 15480
    },
    {
      "epoch": 0.35250210499965867,
      "grad_norm": 4.09375,
      "learning_rate": 1.3214750139327515e-05,
      "loss": 0.8177,
      "step": 15490
    },
    {
      "epoch": 0.3527296725303234,
      "grad_norm": 225.0,
      "learning_rate": 1.321010588890953e-05,
      "loss": 1.0028,
      "step": 15500
    },
    {
      "epoch": 0.3529572400609881,
      "grad_norm": 71.5,
      "learning_rate": 1.3205461638491548e-05,
      "loss": 0.7376,
      "step": 15510
    },
    {
      "epoch": 0.3531848075916528,
      "grad_norm": 0.2333984375,
      "learning_rate": 1.3200817388073566e-05,
      "loss": 1.1548,
      "step": 15520
    },
    {
      "epoch": 0.35341237512231755,
      "grad_norm": 1.3984375,
      "learning_rate": 1.3196173137655583e-05,
      "loss": 1.4933,
      "step": 15530
    },
    {
      "epoch": 0.35363994265298226,
      "grad_norm": 8.25,
      "learning_rate": 1.31915288872376e-05,
      "loss": 0.7798,
      "step": 15540
    },
    {
      "epoch": 0.353867510183647,
      "grad_norm": 0.0189208984375,
      "learning_rate": 1.318688463681962e-05,
      "loss": 0.4567,
      "step": 15550
    },
    {
      "epoch": 0.3540950777143117,
      "grad_norm": 0.1357421875,
      "learning_rate": 1.3182240386401636e-05,
      "loss": 0.6848,
      "step": 15560
    },
    {
      "epoch": 0.3543226452449764,
      "grad_norm": 0.01025390625,
      "learning_rate": 1.3177596135983652e-05,
      "loss": 1.4235,
      "step": 15570
    },
    {
      "epoch": 0.3545502127756412,
      "grad_norm": 194.0,
      "learning_rate": 1.317295188556567e-05,
      "loss": 0.5879,
      "step": 15580
    },
    {
      "epoch": 0.3547777803063059,
      "grad_norm": 0.004913330078125,
      "learning_rate": 1.3168307635147687e-05,
      "loss": 0.3398,
      "step": 15590
    },
    {
      "epoch": 0.35500534783697063,
      "grad_norm": 1.3828125,
      "learning_rate": 1.3163663384729706e-05,
      "loss": 0.5632,
      "step": 15600
    },
    {
      "epoch": 0.35523291536763535,
      "grad_norm": 0.019287109375,
      "learning_rate": 1.3159019134311723e-05,
      "loss": 0.6453,
      "step": 15610
    },
    {
      "epoch": 0.35546048289830007,
      "grad_norm": 0.015625,
      "learning_rate": 1.3154374883893742e-05,
      "loss": 0.8242,
      "step": 15620
    },
    {
      "epoch": 0.3556880504289648,
      "grad_norm": 185.0,
      "learning_rate": 1.3149730633475759e-05,
      "loss": 1.1618,
      "step": 15630
    },
    {
      "epoch": 0.3559156179596295,
      "grad_norm": 211.0,
      "learning_rate": 1.3145086383057774e-05,
      "loss": 0.3499,
      "step": 15640
    },
    {
      "epoch": 0.3561431854902942,
      "grad_norm": 210.0,
      "learning_rate": 1.3140442132639793e-05,
      "loss": 0.7016,
      "step": 15650
    },
    {
      "epoch": 0.356370753020959,
      "grad_norm": 0.0025787353515625,
      "learning_rate": 1.313579788222181e-05,
      "loss": 0.6714,
      "step": 15660
    },
    {
      "epoch": 0.3565983205516237,
      "grad_norm": 252.0,
      "learning_rate": 1.3131153631803827e-05,
      "loss": 1.6999,
      "step": 15670
    },
    {
      "epoch": 0.35682588808228843,
      "grad_norm": 182.0,
      "learning_rate": 1.3126509381385846e-05,
      "loss": 0.4642,
      "step": 15680
    },
    {
      "epoch": 0.35705345561295315,
      "grad_norm": 130.0,
      "learning_rate": 1.3121865130967863e-05,
      "loss": 1.4242,
      "step": 15690
    },
    {
      "epoch": 0.35728102314361787,
      "grad_norm": 109.5,
      "learning_rate": 1.3117220880549882e-05,
      "loss": 0.0546,
      "step": 15700
    },
    {
      "epoch": 0.3575085906742826,
      "grad_norm": 2.671875,
      "learning_rate": 1.3112576630131897e-05,
      "loss": 0.758,
      "step": 15710
    },
    {
      "epoch": 0.3577361582049473,
      "grad_norm": 0.4609375,
      "learning_rate": 1.3107932379713914e-05,
      "loss": 1.1305,
      "step": 15720
    },
    {
      "epoch": 0.357963725735612,
      "grad_norm": 286.0,
      "learning_rate": 1.3103288129295933e-05,
      "loss": 0.503,
      "step": 15730
    },
    {
      "epoch": 0.35819129326627674,
      "grad_norm": 0.00213623046875,
      "learning_rate": 1.309864387887795e-05,
      "loss": 1.7633,
      "step": 15740
    },
    {
      "epoch": 0.3584188607969415,
      "grad_norm": 15.5,
      "learning_rate": 1.3093999628459969e-05,
      "loss": 0.8071,
      "step": 15750
    },
    {
      "epoch": 0.35864642832760624,
      "grad_norm": 0.546875,
      "learning_rate": 1.3089355378041986e-05,
      "loss": 0.4278,
      "step": 15760
    },
    {
      "epoch": 0.35887399585827096,
      "grad_norm": 0.0238037109375,
      "learning_rate": 1.3084711127624003e-05,
      "loss": 0.9018,
      "step": 15770
    },
    {
      "epoch": 0.3591015633889357,
      "grad_norm": 98.5,
      "learning_rate": 1.3080066877206018e-05,
      "loss": 0.7321,
      "step": 15780
    },
    {
      "epoch": 0.3593291309196004,
      "grad_norm": 237.0,
      "learning_rate": 1.3075422626788037e-05,
      "loss": 0.5638,
      "step": 15790
    },
    {
      "epoch": 0.3595566984502651,
      "grad_norm": 340.0,
      "learning_rate": 1.3070778376370054e-05,
      "loss": 1.0392,
      "step": 15800
    },
    {
      "epoch": 0.35978426598092983,
      "grad_norm": 163.0,
      "learning_rate": 1.3066134125952073e-05,
      "loss": 0.9563,
      "step": 15810
    },
    {
      "epoch": 0.36001183351159455,
      "grad_norm": 8.440017700195312e-05,
      "learning_rate": 1.306148987553409e-05,
      "loss": 1.5171,
      "step": 15820
    },
    {
      "epoch": 0.36023940104225927,
      "grad_norm": 157.0,
      "learning_rate": 1.3056845625116109e-05,
      "loss": 0.4192,
      "step": 15830
    },
    {
      "epoch": 0.36046696857292404,
      "grad_norm": 148.0,
      "learning_rate": 1.3052201374698126e-05,
      "loss": 1.3199,
      "step": 15840
    },
    {
      "epoch": 0.36069453610358876,
      "grad_norm": 38.0,
      "learning_rate": 1.3047557124280141e-05,
      "loss": 1.0828,
      "step": 15850
    },
    {
      "epoch": 0.3609221036342535,
      "grad_norm": 0.71875,
      "learning_rate": 1.304291287386216e-05,
      "loss": 0.3923,
      "step": 15860
    },
    {
      "epoch": 0.3611496711649182,
      "grad_norm": 174.0,
      "learning_rate": 1.3038268623444177e-05,
      "loss": 0.5717,
      "step": 15870
    },
    {
      "epoch": 0.3613772386955829,
      "grad_norm": 191.0,
      "learning_rate": 1.3033624373026194e-05,
      "loss": 0.9921,
      "step": 15880
    },
    {
      "epoch": 0.36160480622624763,
      "grad_norm": 142.0,
      "learning_rate": 1.3028980122608213e-05,
      "loss": 1.1172,
      "step": 15890
    },
    {
      "epoch": 0.36183237375691235,
      "grad_norm": 0.0006103515625,
      "learning_rate": 1.302433587219023e-05,
      "loss": 0.3797,
      "step": 15900
    },
    {
      "epoch": 0.36205994128757707,
      "grad_norm": 0.0034637451171875,
      "learning_rate": 1.3019691621772248e-05,
      "loss": 0.8292,
      "step": 15910
    },
    {
      "epoch": 0.3622875088182418,
      "grad_norm": 236.0,
      "learning_rate": 1.3015047371354264e-05,
      "loss": 1.1466,
      "step": 15920
    },
    {
      "epoch": 0.36251507634890656,
      "grad_norm": 0.00099945068359375,
      "learning_rate": 1.3010403120936281e-05,
      "loss": 0.2944,
      "step": 15930
    },
    {
      "epoch": 0.3627426438795713,
      "grad_norm": 392.0,
      "learning_rate": 1.30057588705183e-05,
      "loss": 0.5735,
      "step": 15940
    },
    {
      "epoch": 0.362970211410236,
      "grad_norm": 87.0,
      "learning_rate": 1.3001114620100317e-05,
      "loss": 0.9794,
      "step": 15950
    },
    {
      "epoch": 0.3631977789409007,
      "grad_norm": 0.0036468505859375,
      "learning_rate": 1.2996470369682335e-05,
      "loss": 1.6257,
      "step": 15960
    },
    {
      "epoch": 0.36342534647156544,
      "grad_norm": 161.0,
      "learning_rate": 1.2991826119264352e-05,
      "loss": 1.2111,
      "step": 15970
    },
    {
      "epoch": 0.36365291400223015,
      "grad_norm": 33.0,
      "learning_rate": 1.298718186884637e-05,
      "loss": 0.8252,
      "step": 15980
    },
    {
      "epoch": 0.3638804815328949,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.2982537618428387e-05,
      "loss": 0.5141,
      "step": 15990
    },
    {
      "epoch": 0.3641080490635596,
      "grad_norm": 170.0,
      "learning_rate": 1.2977893368010404e-05,
      "loss": 0.4366,
      "step": 16000
    },
    {
      "epoch": 0.3643356165942243,
      "grad_norm": 76.0,
      "learning_rate": 1.297324911759242e-05,
      "loss": 0.2637,
      "step": 16010
    },
    {
      "epoch": 0.3645631841248891,
      "grad_norm": 53.5,
      "learning_rate": 1.296860486717444e-05,
      "loss": 1.5652,
      "step": 16020
    },
    {
      "epoch": 0.3647907516555538,
      "grad_norm": 0.0008392333984375,
      "learning_rate": 1.2963960616756456e-05,
      "loss": 0.4035,
      "step": 16030
    },
    {
      "epoch": 0.3650183191862185,
      "grad_norm": 107.0,
      "learning_rate": 1.2959316366338475e-05,
      "loss": 0.9229,
      "step": 16040
    },
    {
      "epoch": 0.36524588671688324,
      "grad_norm": 376.0,
      "learning_rate": 1.2954672115920492e-05,
      "loss": 0.3609,
      "step": 16050
    },
    {
      "epoch": 0.36547345424754796,
      "grad_norm": 160.0,
      "learning_rate": 1.2950027865502508e-05,
      "loss": 0.8992,
      "step": 16060
    },
    {
      "epoch": 0.3657010217782127,
      "grad_norm": 200.0,
      "learning_rate": 1.2945383615084526e-05,
      "loss": 0.5738,
      "step": 16070
    },
    {
      "epoch": 0.3659285893088774,
      "grad_norm": 0.01483154296875,
      "learning_rate": 1.2940739364666543e-05,
      "loss": 0.7051,
      "step": 16080
    },
    {
      "epoch": 0.3661561568395421,
      "grad_norm": 177.0,
      "learning_rate": 1.293609511424856e-05,
      "loss": 1.3206,
      "step": 16090
    },
    {
      "epoch": 0.36638372437020683,
      "grad_norm": 47.0,
      "learning_rate": 1.293145086383058e-05,
      "loss": 0.4525,
      "step": 16100
    },
    {
      "epoch": 0.3666112919008716,
      "grad_norm": 24.25,
      "learning_rate": 1.2926806613412596e-05,
      "loss": 0.5652,
      "step": 16110
    },
    {
      "epoch": 0.3668388594315363,
      "grad_norm": 0.1396484375,
      "learning_rate": 1.2922162362994615e-05,
      "loss": 0.312,
      "step": 16120
    },
    {
      "epoch": 0.36706642696220104,
      "grad_norm": 422.0,
      "learning_rate": 1.291751811257663e-05,
      "loss": 0.9173,
      "step": 16130
    },
    {
      "epoch": 0.36729399449286576,
      "grad_norm": 208.0,
      "learning_rate": 1.2912873862158647e-05,
      "loss": 1.8416,
      "step": 16140
    },
    {
      "epoch": 0.3675215620235305,
      "grad_norm": 0.0245361328125,
      "learning_rate": 1.2908229611740666e-05,
      "loss": 0.6649,
      "step": 16150
    },
    {
      "epoch": 0.3677491295541952,
      "grad_norm": 280.0,
      "learning_rate": 1.2903585361322683e-05,
      "loss": 0.4746,
      "step": 16160
    },
    {
      "epoch": 0.3679766970848599,
      "grad_norm": 82.0,
      "learning_rate": 1.2898941110904702e-05,
      "loss": 1.9868,
      "step": 16170
    },
    {
      "epoch": 0.36820426461552463,
      "grad_norm": 0.000858306884765625,
      "learning_rate": 1.2894296860486719e-05,
      "loss": 0.3607,
      "step": 16180
    },
    {
      "epoch": 0.3684318321461894,
      "grad_norm": 17.5,
      "learning_rate": 1.2889652610068736e-05,
      "loss": 0.0988,
      "step": 16190
    },
    {
      "epoch": 0.3686593996768541,
      "grad_norm": 231.0,
      "learning_rate": 1.2885008359650753e-05,
      "loss": 0.5711,
      "step": 16200
    },
    {
      "epoch": 0.36888696720751885,
      "grad_norm": 0.001190185546875,
      "learning_rate": 1.288036410923277e-05,
      "loss": 0.4288,
      "step": 16210
    },
    {
      "epoch": 0.36911453473818356,
      "grad_norm": 75.5,
      "learning_rate": 1.2875719858814787e-05,
      "loss": 0.7394,
      "step": 16220
    },
    {
      "epoch": 0.3693421022688483,
      "grad_norm": 0.11962890625,
      "learning_rate": 1.2871075608396806e-05,
      "loss": 1.2715,
      "step": 16230
    },
    {
      "epoch": 0.369569669799513,
      "grad_norm": 40.75,
      "learning_rate": 1.2866431357978823e-05,
      "loss": 0.9984,
      "step": 16240
    },
    {
      "epoch": 0.3697972373301777,
      "grad_norm": 256.0,
      "learning_rate": 1.2861787107560842e-05,
      "loss": 1.1072,
      "step": 16250
    },
    {
      "epoch": 0.37002480486084244,
      "grad_norm": 105.0,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 0.6409,
      "step": 16260
    },
    {
      "epoch": 0.37025237239150716,
      "grad_norm": 1664.0,
      "learning_rate": 1.2852498606724874e-05,
      "loss": 1.2922,
      "step": 16270
    },
    {
      "epoch": 0.37047993992217193,
      "grad_norm": 0.458984375,
      "learning_rate": 1.2847854356306893e-05,
      "loss": 1.2397,
      "step": 16280
    },
    {
      "epoch": 0.37070750745283665,
      "grad_norm": 0.002105712890625,
      "learning_rate": 1.284321010588891e-05,
      "loss": 0.5196,
      "step": 16290
    },
    {
      "epoch": 0.37093507498350137,
      "grad_norm": 11.4375,
      "learning_rate": 1.2838565855470929e-05,
      "loss": 1.0441,
      "step": 16300
    },
    {
      "epoch": 0.3711626425141661,
      "grad_norm": 61.75,
      "learning_rate": 1.2833921605052946e-05,
      "loss": 0.3559,
      "step": 16310
    },
    {
      "epoch": 0.3713902100448308,
      "grad_norm": 0.0732421875,
      "learning_rate": 1.2829277354634963e-05,
      "loss": 1.1553,
      "step": 16320
    },
    {
      "epoch": 0.3716177775754955,
      "grad_norm": 159.0,
      "learning_rate": 1.2824633104216982e-05,
      "loss": 1.2631,
      "step": 16330
    },
    {
      "epoch": 0.37184534510616024,
      "grad_norm": 111.5,
      "learning_rate": 1.2819988853798997e-05,
      "loss": 0.7138,
      "step": 16340
    },
    {
      "epoch": 0.37207291263682496,
      "grad_norm": 1.3125,
      "learning_rate": 1.2815344603381014e-05,
      "loss": 0.3249,
      "step": 16350
    },
    {
      "epoch": 0.3723004801674897,
      "grad_norm": 37.75,
      "learning_rate": 1.2810700352963033e-05,
      "loss": 0.6052,
      "step": 16360
    },
    {
      "epoch": 0.37252804769815445,
      "grad_norm": 28.625,
      "learning_rate": 1.280605610254505e-05,
      "loss": 0.8426,
      "step": 16370
    },
    {
      "epoch": 0.37275561522881917,
      "grad_norm": 79.0,
      "learning_rate": 1.2801411852127069e-05,
      "loss": 0.9621,
      "step": 16380
    },
    {
      "epoch": 0.3729831827594839,
      "grad_norm": 144.0,
      "learning_rate": 1.2796767601709086e-05,
      "loss": 1.7628,
      "step": 16390
    },
    {
      "epoch": 0.3732107502901486,
      "grad_norm": 87.5,
      "learning_rate": 1.2792123351291103e-05,
      "loss": 0.7972,
      "step": 16400
    },
    {
      "epoch": 0.3734383178208133,
      "grad_norm": 190.0,
      "learning_rate": 1.278747910087312e-05,
      "loss": 0.6429,
      "step": 16410
    },
    {
      "epoch": 0.37366588535147804,
      "grad_norm": 87.5,
      "learning_rate": 1.2782834850455137e-05,
      "loss": 0.9118,
      "step": 16420
    },
    {
      "epoch": 0.37389345288214276,
      "grad_norm": 54.5,
      "learning_rate": 1.2778190600037154e-05,
      "loss": 0.3812,
      "step": 16430
    },
    {
      "epoch": 0.3741210204128075,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.2773546349619173e-05,
      "loss": 0.7153,
      "step": 16440
    },
    {
      "epoch": 0.3743485879434722,
      "grad_norm": 0.0289306640625,
      "learning_rate": 1.276890209920119e-05,
      "loss": 0.8649,
      "step": 16450
    },
    {
      "epoch": 0.374576155474137,
      "grad_norm": 168.0,
      "learning_rate": 1.2764257848783208e-05,
      "loss": 0.9267,
      "step": 16460
    },
    {
      "epoch": 0.3748037230048017,
      "grad_norm": 322.0,
      "learning_rate": 1.2759613598365226e-05,
      "loss": 0.9965,
      "step": 16470
    },
    {
      "epoch": 0.3750312905354664,
      "grad_norm": 215.0,
      "learning_rate": 1.2754969347947241e-05,
      "loss": 1.285,
      "step": 16480
    },
    {
      "epoch": 0.37525885806613113,
      "grad_norm": 16.75,
      "learning_rate": 1.275032509752926e-05,
      "loss": 0.913,
      "step": 16490
    },
    {
      "epoch": 0.37548642559679585,
      "grad_norm": 211.0,
      "learning_rate": 1.2745680847111277e-05,
      "loss": 0.7308,
      "step": 16500
    },
    {
      "epoch": 0.37571399312746057,
      "grad_norm": 34.25,
      "learning_rate": 1.2741036596693295e-05,
      "loss": 0.7919,
      "step": 16510
    },
    {
      "epoch": 0.3759415606581253,
      "grad_norm": 73.0,
      "learning_rate": 1.2736392346275313e-05,
      "loss": 0.397,
      "step": 16520
    },
    {
      "epoch": 0.37616912818879,
      "grad_norm": 61.25,
      "learning_rate": 1.273174809585733e-05,
      "loss": 0.6596,
      "step": 16530
    },
    {
      "epoch": 0.3763966957194547,
      "grad_norm": 144.0,
      "learning_rate": 1.2727103845439348e-05,
      "loss": 1.3412,
      "step": 16540
    },
    {
      "epoch": 0.3766242632501195,
      "grad_norm": 0.055908203125,
      "learning_rate": 1.2722459595021364e-05,
      "loss": 0.1395,
      "step": 16550
    },
    {
      "epoch": 0.3768518307807842,
      "grad_norm": 0.068359375,
      "learning_rate": 1.271781534460338e-05,
      "loss": 0.3578,
      "step": 16560
    },
    {
      "epoch": 0.37707939831144893,
      "grad_norm": 123.0,
      "learning_rate": 1.27131710941854e-05,
      "loss": 0.4674,
      "step": 16570
    },
    {
      "epoch": 0.37730696584211365,
      "grad_norm": 96.0,
      "learning_rate": 1.2708526843767417e-05,
      "loss": 0.7044,
      "step": 16580
    },
    {
      "epoch": 0.37753453337277837,
      "grad_norm": 4.96875,
      "learning_rate": 1.2703882593349435e-05,
      "loss": 0.2028,
      "step": 16590
    },
    {
      "epoch": 0.3777621009034431,
      "grad_norm": 91.5,
      "learning_rate": 1.2699238342931452e-05,
      "loss": 1.0424,
      "step": 16600
    },
    {
      "epoch": 0.3779896684341078,
      "grad_norm": 38.0,
      "learning_rate": 1.2694594092513471e-05,
      "loss": 0.4607,
      "step": 16610
    },
    {
      "epoch": 0.3782172359647725,
      "grad_norm": 0.00043487548828125,
      "learning_rate": 1.2689949842095486e-05,
      "loss": 1.3067,
      "step": 16620
    },
    {
      "epoch": 0.3784448034954373,
      "grad_norm": 39.75,
      "learning_rate": 1.2685305591677504e-05,
      "loss": 0.8811,
      "step": 16630
    },
    {
      "epoch": 0.378672371026102,
      "grad_norm": 0.0169677734375,
      "learning_rate": 1.2680661341259522e-05,
      "loss": 0.3443,
      "step": 16640
    },
    {
      "epoch": 0.37889993855676674,
      "grad_norm": 93.0,
      "learning_rate": 1.267601709084154e-05,
      "loss": 0.8494,
      "step": 16650
    },
    {
      "epoch": 0.37912750608743145,
      "grad_norm": 91.0,
      "learning_rate": 1.2671372840423556e-05,
      "loss": 1.0365,
      "step": 16660
    },
    {
      "epoch": 0.3793550736180962,
      "grad_norm": 2.546875,
      "learning_rate": 1.2666728590005575e-05,
      "loss": 0.6799,
      "step": 16670
    },
    {
      "epoch": 0.3795826411487609,
      "grad_norm": 0.00103759765625,
      "learning_rate": 1.2662084339587592e-05,
      "loss": 0.259,
      "step": 16680
    },
    {
      "epoch": 0.3798102086794256,
      "grad_norm": 249.0,
      "learning_rate": 1.2657440089169608e-05,
      "loss": 0.7308,
      "step": 16690
    },
    {
      "epoch": 0.38003777621009033,
      "grad_norm": 268.0,
      "learning_rate": 1.2652795838751626e-05,
      "loss": 0.9347,
      "step": 16700
    },
    {
      "epoch": 0.38026534374075505,
      "grad_norm": 151.0,
      "learning_rate": 1.2648151588333643e-05,
      "loss": 0.8446,
      "step": 16710
    },
    {
      "epoch": 0.3804929112714198,
      "grad_norm": 0.00032806396484375,
      "learning_rate": 1.2643507337915662e-05,
      "loss": 0.1795,
      "step": 16720
    },
    {
      "epoch": 0.38072047880208454,
      "grad_norm": 5.9375,
      "learning_rate": 1.2638863087497679e-05,
      "loss": 0.8374,
      "step": 16730
    },
    {
      "epoch": 0.38094804633274926,
      "grad_norm": 21.5,
      "learning_rate": 1.2634218837079696e-05,
      "loss": 1.006,
      "step": 16740
    },
    {
      "epoch": 0.381175613863414,
      "grad_norm": 0.85546875,
      "learning_rate": 1.2629574586661715e-05,
      "loss": 1.1695,
      "step": 16750
    },
    {
      "epoch": 0.3814031813940787,
      "grad_norm": 0.0019989013671875,
      "learning_rate": 1.262493033624373e-05,
      "loss": 0.7222,
      "step": 16760
    },
    {
      "epoch": 0.3816307489247434,
      "grad_norm": 127.5,
      "learning_rate": 1.2620286085825747e-05,
      "loss": 0.71,
      "step": 16770
    },
    {
      "epoch": 0.38185831645540813,
      "grad_norm": 57.0,
      "learning_rate": 1.2615641835407766e-05,
      "loss": 1.4763,
      "step": 16780
    },
    {
      "epoch": 0.38208588398607285,
      "grad_norm": 61.5,
      "learning_rate": 1.2610997584989783e-05,
      "loss": 0.6749,
      "step": 16790
    },
    {
      "epoch": 0.38231345151673757,
      "grad_norm": 6.03125,
      "learning_rate": 1.2606353334571802e-05,
      "loss": 0.6963,
      "step": 16800
    },
    {
      "epoch": 0.38254101904740234,
      "grad_norm": 0.080078125,
      "learning_rate": 1.2601709084153819e-05,
      "loss": 0.7662,
      "step": 16810
    },
    {
      "epoch": 0.38276858657806706,
      "grad_norm": 84.5,
      "learning_rate": 1.2597064833735838e-05,
      "loss": 0.5923,
      "step": 16820
    },
    {
      "epoch": 0.3829961541087318,
      "grad_norm": 104.0,
      "learning_rate": 1.2592420583317853e-05,
      "loss": 1.259,
      "step": 16830
    },
    {
      "epoch": 0.3832237216393965,
      "grad_norm": 91.5,
      "learning_rate": 1.258777633289987e-05,
      "loss": 1.4961,
      "step": 16840
    },
    {
      "epoch": 0.3834512891700612,
      "grad_norm": 151.0,
      "learning_rate": 1.2583132082481889e-05,
      "loss": 0.8115,
      "step": 16850
    },
    {
      "epoch": 0.38367885670072593,
      "grad_norm": 27.0,
      "learning_rate": 1.2578487832063906e-05,
      "loss": 0.8399,
      "step": 16860
    },
    {
      "epoch": 0.38390642423139065,
      "grad_norm": 0.052978515625,
      "learning_rate": 1.2573843581645923e-05,
      "loss": 0.6785,
      "step": 16870
    },
    {
      "epoch": 0.38413399176205537,
      "grad_norm": 128.0,
      "learning_rate": 1.2569199331227942e-05,
      "loss": 1.0389,
      "step": 16880
    },
    {
      "epoch": 0.3843615592927201,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.2564555080809959e-05,
      "loss": 1.2284,
      "step": 16890
    },
    {
      "epoch": 0.38458912682338486,
      "grad_norm": 3.015625,
      "learning_rate": 1.2559910830391974e-05,
      "loss": 0.882,
      "step": 16900
    },
    {
      "epoch": 0.3848166943540496,
      "grad_norm": 95.0,
      "learning_rate": 1.2555266579973993e-05,
      "loss": 0.9268,
      "step": 16910
    },
    {
      "epoch": 0.3850442618847143,
      "grad_norm": 4.71875,
      "learning_rate": 1.255062232955601e-05,
      "loss": 0.3932,
      "step": 16920
    },
    {
      "epoch": 0.385271829415379,
      "grad_norm": 0.01953125,
      "learning_rate": 1.2545978079138029e-05,
      "loss": 0.3272,
      "step": 16930
    },
    {
      "epoch": 0.38549939694604374,
      "grad_norm": 0.578125,
      "learning_rate": 1.2541333828720046e-05,
      "loss": 0.9968,
      "step": 16940
    },
    {
      "epoch": 0.38572696447670846,
      "grad_norm": 0.3984375,
      "learning_rate": 1.2536689578302065e-05,
      "loss": 0.4099,
      "step": 16950
    },
    {
      "epoch": 0.3859545320073732,
      "grad_norm": 354.0,
      "learning_rate": 1.2532045327884082e-05,
      "loss": 1.3693,
      "step": 16960
    },
    {
      "epoch": 0.3861820995380379,
      "grad_norm": 486.0,
      "learning_rate": 1.2527401077466097e-05,
      "loss": 1.1435,
      "step": 16970
    },
    {
      "epoch": 0.3864096670687026,
      "grad_norm": 55.5,
      "learning_rate": 1.2522756827048114e-05,
      "loss": 0.6448,
      "step": 16980
    },
    {
      "epoch": 0.3866372345993674,
      "grad_norm": 135.0,
      "learning_rate": 1.2518112576630133e-05,
      "loss": 0.5282,
      "step": 16990
    },
    {
      "epoch": 0.3868648021300321,
      "grad_norm": 171.0,
      "learning_rate": 1.251346832621215e-05,
      "loss": 1.2159,
      "step": 17000
    },
    {
      "epoch": 0.3870923696606968,
      "grad_norm": 350.0,
      "learning_rate": 1.2508824075794169e-05,
      "loss": 0.8245,
      "step": 17010
    },
    {
      "epoch": 0.38731993719136154,
      "grad_norm": 0.00015926361083984375,
      "learning_rate": 1.2504179825376186e-05,
      "loss": 0.5061,
      "step": 17020
    },
    {
      "epoch": 0.38754750472202626,
      "grad_norm": 141.0,
      "learning_rate": 1.2499535574958204e-05,
      "loss": 0.6716,
      "step": 17030
    },
    {
      "epoch": 0.387775072252691,
      "grad_norm": 94.5,
      "learning_rate": 1.249489132454022e-05,
      "loss": 0.7097,
      "step": 17040
    },
    {
      "epoch": 0.3880026397833557,
      "grad_norm": 0.00060272216796875,
      "learning_rate": 1.2490247074122237e-05,
      "loss": 0.7681,
      "step": 17050
    },
    {
      "epoch": 0.3882302073140204,
      "grad_norm": 253.0,
      "learning_rate": 1.2485602823704256e-05,
      "loss": 0.8042,
      "step": 17060
    },
    {
      "epoch": 0.38845777484468513,
      "grad_norm": 54.25,
      "learning_rate": 1.2480958573286273e-05,
      "loss": 0.8486,
      "step": 17070
    },
    {
      "epoch": 0.3886853423753499,
      "grad_norm": 0.001007080078125,
      "learning_rate": 1.247631432286829e-05,
      "loss": 0.7343,
      "step": 17080
    },
    {
      "epoch": 0.3889129099060146,
      "grad_norm": 133.0,
      "learning_rate": 1.2471670072450308e-05,
      "loss": 0.6121,
      "step": 17090
    },
    {
      "epoch": 0.38914047743667934,
      "grad_norm": 99.5,
      "learning_rate": 1.2467025822032325e-05,
      "loss": 0.3583,
      "step": 17100
    },
    {
      "epoch": 0.38936804496734406,
      "grad_norm": 0.0751953125,
      "learning_rate": 1.246238157161434e-05,
      "loss": 1.037,
      "step": 17110
    },
    {
      "epoch": 0.3895956124980088,
      "grad_norm": 3.640625,
      "learning_rate": 1.245773732119636e-05,
      "loss": 0.7493,
      "step": 17120
    },
    {
      "epoch": 0.3898231800286735,
      "grad_norm": 0.08544921875,
      "learning_rate": 1.2453093070778377e-05,
      "loss": 0.138,
      "step": 17130
    },
    {
      "epoch": 0.3900507475593382,
      "grad_norm": 97.0,
      "learning_rate": 1.2448448820360395e-05,
      "loss": 1.4574,
      "step": 17140
    },
    {
      "epoch": 0.39027831509000294,
      "grad_norm": 81.0,
      "learning_rate": 1.2443804569942412e-05,
      "loss": 0.9045,
      "step": 17150
    },
    {
      "epoch": 0.3905058826206677,
      "grad_norm": 0.013671875,
      "learning_rate": 1.2439160319524431e-05,
      "loss": 0.5006,
      "step": 17160
    },
    {
      "epoch": 0.39073345015133243,
      "grad_norm": 0.0040283203125,
      "learning_rate": 1.2434516069106448e-05,
      "loss": 0.8424,
      "step": 17170
    },
    {
      "epoch": 0.39096101768199715,
      "grad_norm": 1.0625,
      "learning_rate": 1.2429871818688464e-05,
      "loss": 1.3774,
      "step": 17180
    },
    {
      "epoch": 0.39118858521266187,
      "grad_norm": 25.875,
      "learning_rate": 1.2425227568270482e-05,
      "loss": 0.7609,
      "step": 17190
    },
    {
      "epoch": 0.3914161527433266,
      "grad_norm": 89.0,
      "learning_rate": 1.24205833178525e-05,
      "loss": 0.6416,
      "step": 17200
    },
    {
      "epoch": 0.3916437202739913,
      "grad_norm": 22.875,
      "learning_rate": 1.2415939067434516e-05,
      "loss": 0.2466,
      "step": 17210
    },
    {
      "epoch": 0.391871287804656,
      "grad_norm": 0.0001354217529296875,
      "learning_rate": 1.2411294817016535e-05,
      "loss": 0.8901,
      "step": 17220
    },
    {
      "epoch": 0.39209885533532074,
      "grad_norm": 354.0,
      "learning_rate": 1.2406650566598552e-05,
      "loss": 1.217,
      "step": 17230
    },
    {
      "epoch": 0.39232642286598546,
      "grad_norm": 0.11865234375,
      "learning_rate": 1.2402006316180571e-05,
      "loss": 0.9975,
      "step": 17240
    },
    {
      "epoch": 0.39255399039665023,
      "grad_norm": 170.0,
      "learning_rate": 1.2397362065762586e-05,
      "loss": 0.2857,
      "step": 17250
    },
    {
      "epoch": 0.39278155792731495,
      "grad_norm": 0.002777099609375,
      "learning_rate": 1.2392717815344603e-05,
      "loss": 0.2327,
      "step": 17260
    },
    {
      "epoch": 0.39300912545797967,
      "grad_norm": 272.0,
      "learning_rate": 1.2388073564926622e-05,
      "loss": 1.4568,
      "step": 17270
    },
    {
      "epoch": 0.3932366929886444,
      "grad_norm": 51.75,
      "learning_rate": 1.238342931450864e-05,
      "loss": 1.0954,
      "step": 17280
    },
    {
      "epoch": 0.3934642605193091,
      "grad_norm": 120.0,
      "learning_rate": 1.2378785064090656e-05,
      "loss": 1.5833,
      "step": 17290
    },
    {
      "epoch": 0.3936918280499738,
      "grad_norm": 466.0,
      "learning_rate": 1.2374140813672675e-05,
      "loss": 1.3734,
      "step": 17300
    },
    {
      "epoch": 0.39391939558063854,
      "grad_norm": 0.005218505859375,
      "learning_rate": 1.2369496563254692e-05,
      "loss": 1.0348,
      "step": 17310
    },
    {
      "epoch": 0.39414696311130326,
      "grad_norm": 146.0,
      "learning_rate": 1.2364852312836707e-05,
      "loss": 1.0445,
      "step": 17320
    },
    {
      "epoch": 0.394374530641968,
      "grad_norm": 0.004730224609375,
      "learning_rate": 1.2360208062418726e-05,
      "loss": 0.541,
      "step": 17330
    },
    {
      "epoch": 0.39460209817263275,
      "grad_norm": 124.0,
      "learning_rate": 1.2355563812000743e-05,
      "loss": 0.8074,
      "step": 17340
    },
    {
      "epoch": 0.39482966570329747,
      "grad_norm": 7.75,
      "learning_rate": 1.2350919561582762e-05,
      "loss": 0.9686,
      "step": 17350
    },
    {
      "epoch": 0.3950572332339622,
      "grad_norm": 83.0,
      "learning_rate": 1.2346275311164779e-05,
      "loss": 1.1944,
      "step": 17360
    },
    {
      "epoch": 0.3952848007646269,
      "grad_norm": 90.0,
      "learning_rate": 1.2341631060746798e-05,
      "loss": 0.4076,
      "step": 17370
    },
    {
      "epoch": 0.3955123682952916,
      "grad_norm": 205.0,
      "learning_rate": 1.2336986810328815e-05,
      "loss": 1.4992,
      "step": 17380
    },
    {
      "epoch": 0.39573993582595635,
      "grad_norm": 169.0,
      "learning_rate": 1.233234255991083e-05,
      "loss": 0.8125,
      "step": 17390
    },
    {
      "epoch": 0.39596750335662106,
      "grad_norm": 0.00469970703125,
      "learning_rate": 1.2327698309492849e-05,
      "loss": 1.4201,
      "step": 17400
    },
    {
      "epoch": 0.3961950708872858,
      "grad_norm": 0.000423431396484375,
      "learning_rate": 1.2323054059074866e-05,
      "loss": 0.3692,
      "step": 17410
    },
    {
      "epoch": 0.3964226384179505,
      "grad_norm": 81.5,
      "learning_rate": 1.2318409808656883e-05,
      "loss": 0.7759,
      "step": 17420
    },
    {
      "epoch": 0.3966502059486153,
      "grad_norm": 32.25,
      "learning_rate": 1.2313765558238902e-05,
      "loss": 0.708,
      "step": 17430
    },
    {
      "epoch": 0.39687777347928,
      "grad_norm": 0.0223388671875,
      "learning_rate": 1.2309121307820919e-05,
      "loss": 1.0007,
      "step": 17440
    },
    {
      "epoch": 0.3971053410099447,
      "grad_norm": 68.5,
      "learning_rate": 1.2304477057402938e-05,
      "loss": 0.1409,
      "step": 17450
    },
    {
      "epoch": 0.39733290854060943,
      "grad_norm": 134.0,
      "learning_rate": 1.2299832806984953e-05,
      "loss": 0.705,
      "step": 17460
    },
    {
      "epoch": 0.39756047607127415,
      "grad_norm": 0.001495361328125,
      "learning_rate": 1.229518855656697e-05,
      "loss": 0.7679,
      "step": 17470
    },
    {
      "epoch": 0.39778804360193887,
      "grad_norm": 544.0,
      "learning_rate": 1.2290544306148989e-05,
      "loss": 0.9176,
      "step": 17480
    },
    {
      "epoch": 0.3980156111326036,
      "grad_norm": 156.0,
      "learning_rate": 1.2285900055731006e-05,
      "loss": 1.0391,
      "step": 17490
    },
    {
      "epoch": 0.3982431786632683,
      "grad_norm": 0.0133056640625,
      "learning_rate": 1.2281255805313025e-05,
      "loss": 1.1167,
      "step": 17500
    },
    {
      "epoch": 0.398470746193933,
      "grad_norm": 83.0,
      "learning_rate": 1.2276611554895042e-05,
      "loss": 0.3517,
      "step": 17510
    },
    {
      "epoch": 0.3986983137245978,
      "grad_norm": 0.0018157958984375,
      "learning_rate": 1.2271967304477059e-05,
      "loss": 0.3915,
      "step": 17520
    },
    {
      "epoch": 0.3989258812552625,
      "grad_norm": 118.5,
      "learning_rate": 1.2267323054059074e-05,
      "loss": 0.8845,
      "step": 17530
    },
    {
      "epoch": 0.39915344878592723,
      "grad_norm": 159.0,
      "learning_rate": 1.2262678803641093e-05,
      "loss": 0.8344,
      "step": 17540
    },
    {
      "epoch": 0.39938101631659195,
      "grad_norm": 262.0,
      "learning_rate": 1.225803455322311e-05,
      "loss": 0.8815,
      "step": 17550
    },
    {
      "epoch": 0.39960858384725667,
      "grad_norm": 348.0,
      "learning_rate": 1.2253390302805129e-05,
      "loss": 0.8194,
      "step": 17560
    },
    {
      "epoch": 0.3998361513779214,
      "grad_norm": 38.5,
      "learning_rate": 1.2248746052387146e-05,
      "loss": 0.6572,
      "step": 17570
    },
    {
      "epoch": 0.4000637189085861,
      "grad_norm": 0.0010528564453125,
      "learning_rate": 1.2244101801969164e-05,
      "loss": 0.1868,
      "step": 17580
    },
    {
      "epoch": 0.4002912864392508,
      "grad_norm": 0.0091552734375,
      "learning_rate": 1.2239457551551181e-05,
      "loss": 0.4269,
      "step": 17590
    },
    {
      "epoch": 0.40051885396991554,
      "grad_norm": 2.0625,
      "learning_rate": 1.2234813301133197e-05,
      "loss": 0.9737,
      "step": 17600
    },
    {
      "epoch": 0.4007464215005803,
      "grad_norm": 0.5390625,
      "learning_rate": 1.2230169050715216e-05,
      "loss": 0.7785,
      "step": 17610
    },
    {
      "epoch": 0.40097398903124504,
      "grad_norm": 4.59375,
      "learning_rate": 1.2225524800297233e-05,
      "loss": 0.7506,
      "step": 17620
    },
    {
      "epoch": 0.40120155656190976,
      "grad_norm": 294.0,
      "learning_rate": 1.222088054987925e-05,
      "loss": 1.9791,
      "step": 17630
    },
    {
      "epoch": 0.4014291240925745,
      "grad_norm": 1296.0,
      "learning_rate": 1.2216236299461268e-05,
      "loss": 1.2024,
      "step": 17640
    },
    {
      "epoch": 0.4016566916232392,
      "grad_norm": 1.2265625,
      "learning_rate": 1.2211592049043285e-05,
      "loss": 0.1902,
      "step": 17650
    },
    {
      "epoch": 0.4018842591539039,
      "grad_norm": 14.0625,
      "learning_rate": 1.2206947798625304e-05,
      "loss": 0.7949,
      "step": 17660
    },
    {
      "epoch": 0.40211182668456863,
      "grad_norm": 0.00244140625,
      "learning_rate": 1.220230354820732e-05,
      "loss": 0.4611,
      "step": 17670
    },
    {
      "epoch": 0.40233939421523335,
      "grad_norm": 0.00592041015625,
      "learning_rate": 1.2197659297789337e-05,
      "loss": 1.031,
      "step": 17680
    },
    {
      "epoch": 0.4025669617458981,
      "grad_norm": 1.3984375,
      "learning_rate": 1.2193015047371355e-05,
      "loss": 1.5722,
      "step": 17690
    },
    {
      "epoch": 0.40279452927656284,
      "grad_norm": 0.0006561279296875,
      "learning_rate": 1.2188370796953372e-05,
      "loss": 0.6163,
      "step": 17700
    },
    {
      "epoch": 0.40302209680722756,
      "grad_norm": 127.0,
      "learning_rate": 1.2183726546535391e-05,
      "loss": 0.3253,
      "step": 17710
    },
    {
      "epoch": 0.4032496643378923,
      "grad_norm": 76.5,
      "learning_rate": 1.2179082296117408e-05,
      "loss": 0.4038,
      "step": 17720
    },
    {
      "epoch": 0.403477231868557,
      "grad_norm": 53.0,
      "learning_rate": 1.2174438045699425e-05,
      "loss": 0.9599,
      "step": 17730
    },
    {
      "epoch": 0.4037047993992217,
      "grad_norm": 139.0,
      "learning_rate": 1.2169793795281442e-05,
      "loss": 0.7909,
      "step": 17740
    },
    {
      "epoch": 0.40393236692988643,
      "grad_norm": 0.197265625,
      "learning_rate": 1.216514954486346e-05,
      "loss": 0.7193,
      "step": 17750
    },
    {
      "epoch": 0.40415993446055115,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.2160505294445476e-05,
      "loss": 0.7582,
      "step": 17760
    },
    {
      "epoch": 0.40438750199121587,
      "grad_norm": 65.0,
      "learning_rate": 1.2155861044027495e-05,
      "loss": 0.6586,
      "step": 17770
    },
    {
      "epoch": 0.40461506952188064,
      "grad_norm": 0.00225830078125,
      "learning_rate": 1.2151216793609512e-05,
      "loss": 0.9581,
      "step": 17780
    },
    {
      "epoch": 0.40484263705254536,
      "grad_norm": 0.006317138671875,
      "learning_rate": 1.2146572543191531e-05,
      "loss": 0.2644,
      "step": 17790
    },
    {
      "epoch": 0.4050702045832101,
      "grad_norm": 1.46875,
      "learning_rate": 1.2141928292773548e-05,
      "loss": 0.8513,
      "step": 17800
    },
    {
      "epoch": 0.4052977721138748,
      "grad_norm": 232.0,
      "learning_rate": 1.2137284042355563e-05,
      "loss": 1.0833,
      "step": 17810
    },
    {
      "epoch": 0.4055253396445395,
      "grad_norm": 37.75,
      "learning_rate": 1.2132639791937582e-05,
      "loss": 0.9023,
      "step": 17820
    },
    {
      "epoch": 0.40575290717520424,
      "grad_norm": 0.001556396484375,
      "learning_rate": 1.21279955415196e-05,
      "loss": 0.6931,
      "step": 17830
    },
    {
      "epoch": 0.40598047470586895,
      "grad_norm": 161.0,
      "learning_rate": 1.2123351291101616e-05,
      "loss": 0.3631,
      "step": 17840
    },
    {
      "epoch": 0.4062080422365337,
      "grad_norm": 91.5,
      "learning_rate": 1.2118707040683635e-05,
      "loss": 1.409,
      "step": 17850
    },
    {
      "epoch": 0.4064356097671984,
      "grad_norm": 51.5,
      "learning_rate": 1.2114062790265652e-05,
      "loss": 1.5302,
      "step": 17860
    },
    {
      "epoch": 0.40666317729786317,
      "grad_norm": 226.0,
      "learning_rate": 1.210941853984767e-05,
      "loss": 0.906,
      "step": 17870
    },
    {
      "epoch": 0.4068907448285279,
      "grad_norm": 0.1484375,
      "learning_rate": 1.2104774289429686e-05,
      "loss": 0.4803,
      "step": 17880
    },
    {
      "epoch": 0.4071183123591926,
      "grad_norm": 0.3671875,
      "learning_rate": 1.2100130039011703e-05,
      "loss": 0.3903,
      "step": 17890
    },
    {
      "epoch": 0.4073458798898573,
      "grad_norm": 202.0,
      "learning_rate": 1.2095485788593722e-05,
      "loss": 0.6071,
      "step": 17900
    },
    {
      "epoch": 0.40757344742052204,
      "grad_norm": 486.0,
      "learning_rate": 1.2090841538175739e-05,
      "loss": 1.4786,
      "step": 17910
    },
    {
      "epoch": 0.40780101495118676,
      "grad_norm": 17.375,
      "learning_rate": 1.2086197287757758e-05,
      "loss": 0.3555,
      "step": 17920
    },
    {
      "epoch": 0.4080285824818515,
      "grad_norm": 198.0,
      "learning_rate": 1.2081553037339775e-05,
      "loss": 0.5354,
      "step": 17930
    },
    {
      "epoch": 0.4082561500125162,
      "grad_norm": 39.5,
      "learning_rate": 1.2076908786921792e-05,
      "loss": 0.9613,
      "step": 17940
    },
    {
      "epoch": 0.4084837175431809,
      "grad_norm": 0.04052734375,
      "learning_rate": 1.2072264536503809e-05,
      "loss": 0.9074,
      "step": 17950
    },
    {
      "epoch": 0.4087112850738457,
      "grad_norm": 169.0,
      "learning_rate": 1.2067620286085826e-05,
      "loss": 0.5313,
      "step": 17960
    },
    {
      "epoch": 0.4089388526045104,
      "grad_norm": 178.0,
      "learning_rate": 1.2062976035667843e-05,
      "loss": 1.4092,
      "step": 17970
    },
    {
      "epoch": 0.4091664201351751,
      "grad_norm": 93.0,
      "learning_rate": 1.2058331785249862e-05,
      "loss": 1.6036,
      "step": 17980
    },
    {
      "epoch": 0.40939398766583984,
      "grad_norm": 12.875,
      "learning_rate": 1.2053687534831879e-05,
      "loss": 1.1937,
      "step": 17990
    },
    {
      "epoch": 0.40962155519650456,
      "grad_norm": 0.0031585693359375,
      "learning_rate": 1.2049043284413898e-05,
      "loss": 0.5844,
      "step": 18000
    },
    {
      "epoch": 0.4098491227271693,
      "grad_norm": 139.0,
      "learning_rate": 1.2044399033995915e-05,
      "loss": 0.9902,
      "step": 18010
    },
    {
      "epoch": 0.410076690257834,
      "grad_norm": 125.0,
      "learning_rate": 1.203975478357793e-05,
      "loss": 0.5762,
      "step": 18020
    },
    {
      "epoch": 0.4103042577884987,
      "grad_norm": 96.5,
      "learning_rate": 1.2035110533159949e-05,
      "loss": 0.7045,
      "step": 18030
    },
    {
      "epoch": 0.41053182531916343,
      "grad_norm": 38.25,
      "learning_rate": 1.2030466282741966e-05,
      "loss": 0.6972,
      "step": 18040
    },
    {
      "epoch": 0.4107593928498282,
      "grad_norm": 0.00531005859375,
      "learning_rate": 1.2025822032323985e-05,
      "loss": 0.7883,
      "step": 18050
    },
    {
      "epoch": 0.4109869603804929,
      "grad_norm": 0.013671875,
      "learning_rate": 1.2021177781906002e-05,
      "loss": 1.0242,
      "step": 18060
    },
    {
      "epoch": 0.41121452791115765,
      "grad_norm": 32.75,
      "learning_rate": 1.2016533531488019e-05,
      "loss": 0.7443,
      "step": 18070
    },
    {
      "epoch": 0.41144209544182236,
      "grad_norm": 2.640625,
      "learning_rate": 1.2011889281070037e-05,
      "loss": 0.5149,
      "step": 18080
    },
    {
      "epoch": 0.4116696629724871,
      "grad_norm": 306.0,
      "learning_rate": 1.2007245030652053e-05,
      "loss": 0.6479,
      "step": 18090
    },
    {
      "epoch": 0.4118972305031518,
      "grad_norm": 3.234375,
      "learning_rate": 1.200260078023407e-05,
      "loss": 0.5257,
      "step": 18100
    },
    {
      "epoch": 0.4121247980338165,
      "grad_norm": 0.00095367431640625,
      "learning_rate": 1.1997956529816089e-05,
      "loss": 0.5818,
      "step": 18110
    },
    {
      "epoch": 0.41235236556448124,
      "grad_norm": 121.5,
      "learning_rate": 1.1993312279398106e-05,
      "loss": 0.5247,
      "step": 18120
    },
    {
      "epoch": 0.412579933095146,
      "grad_norm": 187.0,
      "learning_rate": 1.1988668028980124e-05,
      "loss": 1.983,
      "step": 18130
    },
    {
      "epoch": 0.41280750062581073,
      "grad_norm": 1.03125,
      "learning_rate": 1.1984023778562141e-05,
      "loss": 1.2094,
      "step": 18140
    },
    {
      "epoch": 0.41303506815647545,
      "grad_norm": 2.453125,
      "learning_rate": 1.197937952814416e-05,
      "loss": 0.6829,
      "step": 18150
    },
    {
      "epoch": 0.41326263568714017,
      "grad_norm": 54.25,
      "learning_rate": 1.1974735277726176e-05,
      "loss": 1.4937,
      "step": 18160
    },
    {
      "epoch": 0.4134902032178049,
      "grad_norm": 0.00299072265625,
      "learning_rate": 1.1970091027308193e-05,
      "loss": 0.6557,
      "step": 18170
    },
    {
      "epoch": 0.4137177707484696,
      "grad_norm": 192.0,
      "learning_rate": 1.196544677689021e-05,
      "loss": 0.9217,
      "step": 18180
    },
    {
      "epoch": 0.4139453382791343,
      "grad_norm": 0.0634765625,
      "learning_rate": 1.1960802526472228e-05,
      "loss": 1.0493,
      "step": 18190
    },
    {
      "epoch": 0.41417290580979904,
      "grad_norm": 105.0,
      "learning_rate": 1.1956158276054246e-05,
      "loss": 0.4544,
      "step": 18200
    },
    {
      "epoch": 0.41440047334046376,
      "grad_norm": 0.0299072265625,
      "learning_rate": 1.1951514025636264e-05,
      "loss": 0.8943,
      "step": 18210
    },
    {
      "epoch": 0.41462804087112853,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.1946869775218281e-05,
      "loss": 0.5173,
      "step": 18220
    },
    {
      "epoch": 0.41485560840179325,
      "grad_norm": 52.5,
      "learning_rate": 1.1942225524800297e-05,
      "loss": 0.9757,
      "step": 18230
    },
    {
      "epoch": 0.41508317593245797,
      "grad_norm": 56.75,
      "learning_rate": 1.1937581274382315e-05,
      "loss": 1.082,
      "step": 18240
    },
    {
      "epoch": 0.4153107434631227,
      "grad_norm": 390.0,
      "learning_rate": 1.1932937023964332e-05,
      "loss": 1.3435,
      "step": 18250
    },
    {
      "epoch": 0.4155383109937874,
      "grad_norm": 118.0,
      "learning_rate": 1.1928292773546351e-05,
      "loss": 0.4047,
      "step": 18260
    },
    {
      "epoch": 0.4157658785244521,
      "grad_norm": 0.00640869140625,
      "learning_rate": 1.1923648523128368e-05,
      "loss": 0.8279,
      "step": 18270
    },
    {
      "epoch": 0.41599344605511684,
      "grad_norm": 0.0067138671875,
      "learning_rate": 1.1919004272710385e-05,
      "loss": 0.513,
      "step": 18280
    },
    {
      "epoch": 0.41622101358578156,
      "grad_norm": 90.0,
      "learning_rate": 1.1914360022292404e-05,
      "loss": 0.9149,
      "step": 18290
    },
    {
      "epoch": 0.4164485811164463,
      "grad_norm": 67.0,
      "learning_rate": 1.190971577187442e-05,
      "loss": 0.5235,
      "step": 18300
    },
    {
      "epoch": 0.41667614864711106,
      "grad_norm": 197.0,
      "learning_rate": 1.1905071521456436e-05,
      "loss": 0.8308,
      "step": 18310
    },
    {
      "epoch": 0.4169037161777758,
      "grad_norm": 0.02490234375,
      "learning_rate": 1.1900427271038455e-05,
      "loss": 0.6498,
      "step": 18320
    },
    {
      "epoch": 0.4171312837084405,
      "grad_norm": 0.2216796875,
      "learning_rate": 1.1895783020620472e-05,
      "loss": 0.4057,
      "step": 18330
    },
    {
      "epoch": 0.4173588512391052,
      "grad_norm": 13.6875,
      "learning_rate": 1.1891138770202491e-05,
      "loss": 0.4702,
      "step": 18340
    },
    {
      "epoch": 0.41758641876976993,
      "grad_norm": 103.5,
      "learning_rate": 1.1886494519784508e-05,
      "loss": 0.5368,
      "step": 18350
    },
    {
      "epoch": 0.41781398630043465,
      "grad_norm": 137.0,
      "learning_rate": 1.1881850269366527e-05,
      "loss": 0.5947,
      "step": 18360
    },
    {
      "epoch": 0.41804155383109937,
      "grad_norm": 0.060546875,
      "learning_rate": 1.1877206018948542e-05,
      "loss": 0.5945,
      "step": 18370
    },
    {
      "epoch": 0.4182691213617641,
      "grad_norm": 39.5,
      "learning_rate": 1.187256176853056e-05,
      "loss": 1.4374,
      "step": 18380
    },
    {
      "epoch": 0.4184966888924288,
      "grad_norm": 168.0,
      "learning_rate": 1.1867917518112578e-05,
      "loss": 0.523,
      "step": 18390
    },
    {
      "epoch": 0.4187242564230936,
      "grad_norm": 0.0111083984375,
      "learning_rate": 1.1863273267694595e-05,
      "loss": 0.4927,
      "step": 18400
    },
    {
      "epoch": 0.4189518239537583,
      "grad_norm": 11.75,
      "learning_rate": 1.1858629017276612e-05,
      "loss": 0.5641,
      "step": 18410
    },
    {
      "epoch": 0.419179391484423,
      "grad_norm": 71.0,
      "learning_rate": 1.1853984766858631e-05,
      "loss": 0.9872,
      "step": 18420
    },
    {
      "epoch": 0.41940695901508773,
      "grad_norm": 4.375,
      "learning_rate": 1.1849340516440648e-05,
      "loss": 0.6054,
      "step": 18430
    },
    {
      "epoch": 0.41963452654575245,
      "grad_norm": 167.0,
      "learning_rate": 1.1844696266022663e-05,
      "loss": 0.4379,
      "step": 18440
    },
    {
      "epoch": 0.41986209407641717,
      "grad_norm": 0.00119781494140625,
      "learning_rate": 1.1840052015604682e-05,
      "loss": 0.1155,
      "step": 18450
    },
    {
      "epoch": 0.4200896616070819,
      "grad_norm": 0.400390625,
      "learning_rate": 1.1835407765186699e-05,
      "loss": 0.9431,
      "step": 18460
    },
    {
      "epoch": 0.4203172291377466,
      "grad_norm": 0.01470947265625,
      "learning_rate": 1.1830763514768718e-05,
      "loss": 0.6564,
      "step": 18470
    },
    {
      "epoch": 0.4205447966684113,
      "grad_norm": 40.25,
      "learning_rate": 1.1826119264350735e-05,
      "loss": 0.7337,
      "step": 18480
    },
    {
      "epoch": 0.4207723641990761,
      "grad_norm": 318.0,
      "learning_rate": 1.1821475013932752e-05,
      "loss": 0.5047,
      "step": 18490
    },
    {
      "epoch": 0.4209999317297408,
      "grad_norm": 82.0,
      "learning_rate": 1.181683076351477e-05,
      "loss": 0.7341,
      "step": 18500
    },
    {
      "epoch": 0.42122749926040554,
      "grad_norm": 1.2421875,
      "learning_rate": 1.1812186513096786e-05,
      "loss": 0.7416,
      "step": 18510
    },
    {
      "epoch": 0.42145506679107025,
      "grad_norm": 292.0,
      "learning_rate": 1.1807542262678803e-05,
      "loss": 0.8252,
      "step": 18520
    },
    {
      "epoch": 0.421682634321735,
      "grad_norm": 0.07958984375,
      "learning_rate": 1.1802898012260822e-05,
      "loss": 0.3252,
      "step": 18530
    },
    {
      "epoch": 0.4219102018523997,
      "grad_norm": 282.0,
      "learning_rate": 1.1798253761842839e-05,
      "loss": 1.8573,
      "step": 18540
    },
    {
      "epoch": 0.4221377693830644,
      "grad_norm": 0.0101318359375,
      "learning_rate": 1.1793609511424858e-05,
      "loss": 0.4494,
      "step": 18550
    },
    {
      "epoch": 0.42236533691372913,
      "grad_norm": 139.0,
      "learning_rate": 1.1788965261006875e-05,
      "loss": 0.8999,
      "step": 18560
    },
    {
      "epoch": 0.42259290444439385,
      "grad_norm": 177.0,
      "learning_rate": 1.1784321010588893e-05,
      "loss": 1.5469,
      "step": 18570
    },
    {
      "epoch": 0.4228204719750586,
      "grad_norm": 37.0,
      "learning_rate": 1.1779676760170909e-05,
      "loss": 0.716,
      "step": 18580
    },
    {
      "epoch": 0.42304803950572334,
      "grad_norm": 54.25,
      "learning_rate": 1.1775032509752926e-05,
      "loss": 1.7722,
      "step": 18590
    },
    {
      "epoch": 0.42327560703638806,
      "grad_norm": 0.001922607421875,
      "learning_rate": 1.1770388259334945e-05,
      "loss": 1.4776,
      "step": 18600
    },
    {
      "epoch": 0.4235031745670528,
      "grad_norm": 141.0,
      "learning_rate": 1.1765744008916962e-05,
      "loss": 1.3288,
      "step": 18610
    },
    {
      "epoch": 0.4237307420977175,
      "grad_norm": 165.0,
      "learning_rate": 1.1761099758498979e-05,
      "loss": 1.0274,
      "step": 18620
    },
    {
      "epoch": 0.4239583096283822,
      "grad_norm": 0.1591796875,
      "learning_rate": 1.1756455508080998e-05,
      "loss": 0.8924,
      "step": 18630
    },
    {
      "epoch": 0.42418587715904693,
      "grad_norm": 1.7421875,
      "learning_rate": 1.1751811257663015e-05,
      "loss": 0.6444,
      "step": 18640
    },
    {
      "epoch": 0.42441344468971165,
      "grad_norm": 276.0,
      "learning_rate": 1.174716700724503e-05,
      "loss": 1.1335,
      "step": 18650
    },
    {
      "epoch": 0.4246410122203764,
      "grad_norm": 0.00148773193359375,
      "learning_rate": 1.1742522756827049e-05,
      "loss": 1.1421,
      "step": 18660
    },
    {
      "epoch": 0.42486857975104114,
      "grad_norm": 89.5,
      "learning_rate": 1.1737878506409066e-05,
      "loss": 0.5837,
      "step": 18670
    },
    {
      "epoch": 0.42509614728170586,
      "grad_norm": 185.0,
      "learning_rate": 1.1733234255991084e-05,
      "loss": 0.5884,
      "step": 18680
    },
    {
      "epoch": 0.4253237148123706,
      "grad_norm": 276.0,
      "learning_rate": 1.1728590005573102e-05,
      "loss": 0.8734,
      "step": 18690
    },
    {
      "epoch": 0.4255512823430353,
      "grad_norm": 17.375,
      "learning_rate": 1.172394575515512e-05,
      "loss": 0.3176,
      "step": 18700
    },
    {
      "epoch": 0.4257788498737,
      "grad_norm": 159.0,
      "learning_rate": 1.1719301504737137e-05,
      "loss": 1.1005,
      "step": 18710
    },
    {
      "epoch": 0.42600641740436473,
      "grad_norm": 352.0,
      "learning_rate": 1.1714657254319153e-05,
      "loss": 0.9345,
      "step": 18720
    },
    {
      "epoch": 0.42623398493502945,
      "grad_norm": 172.0,
      "learning_rate": 1.171001300390117e-05,
      "loss": 0.925,
      "step": 18730
    },
    {
      "epoch": 0.42646155246569417,
      "grad_norm": 113.5,
      "learning_rate": 1.1705368753483188e-05,
      "loss": 1.256,
      "step": 18740
    },
    {
      "epoch": 0.42668911999635895,
      "grad_norm": 111.5,
      "learning_rate": 1.1700724503065206e-05,
      "loss": 1.2663,
      "step": 18750
    },
    {
      "epoch": 0.42691668752702366,
      "grad_norm": 0.0732421875,
      "learning_rate": 1.1696080252647224e-05,
      "loss": 0.3866,
      "step": 18760
    },
    {
      "epoch": 0.4271442550576884,
      "grad_norm": 133.0,
      "learning_rate": 1.1691436002229241e-05,
      "loss": 1.3242,
      "step": 18770
    },
    {
      "epoch": 0.4273718225883531,
      "grad_norm": 0.0072021484375,
      "learning_rate": 1.168679175181126e-05,
      "loss": 1.0018,
      "step": 18780
    },
    {
      "epoch": 0.4275993901190178,
      "grad_norm": 130.0,
      "learning_rate": 1.1682147501393275e-05,
      "loss": 1.1158,
      "step": 18790
    },
    {
      "epoch": 0.42782695764968254,
      "grad_norm": 63.25,
      "learning_rate": 1.1677503250975293e-05,
      "loss": 0.4795,
      "step": 18800
    },
    {
      "epoch": 0.42805452518034726,
      "grad_norm": 266.0,
      "learning_rate": 1.1672859000557311e-05,
      "loss": 0.6352,
      "step": 18810
    },
    {
      "epoch": 0.428282092711012,
      "grad_norm": 136.0,
      "learning_rate": 1.1668214750139328e-05,
      "loss": 0.6358,
      "step": 18820
    },
    {
      "epoch": 0.4285096602416767,
      "grad_norm": 61.25,
      "learning_rate": 1.1663570499721345e-05,
      "loss": 0.9743,
      "step": 18830
    },
    {
      "epoch": 0.42873722777234147,
      "grad_norm": 0.51953125,
      "learning_rate": 1.1658926249303364e-05,
      "loss": 0.4739,
      "step": 18840
    },
    {
      "epoch": 0.4289647953030062,
      "grad_norm": 78.0,
      "learning_rate": 1.1654281998885381e-05,
      "loss": 0.664,
      "step": 18850
    },
    {
      "epoch": 0.4291923628336709,
      "grad_norm": 40.5,
      "learning_rate": 1.1649637748467397e-05,
      "loss": 0.5024,
      "step": 18860
    },
    {
      "epoch": 0.4294199303643356,
      "grad_norm": 1.25,
      "learning_rate": 1.1644993498049415e-05,
      "loss": 1.2693,
      "step": 18870
    },
    {
      "epoch": 0.42964749789500034,
      "grad_norm": 57.0,
      "learning_rate": 1.1640349247631432e-05,
      "loss": 0.455,
      "step": 18880
    },
    {
      "epoch": 0.42987506542566506,
      "grad_norm": 0.00061798095703125,
      "learning_rate": 1.1635704997213451e-05,
      "loss": 1.2344,
      "step": 18890
    },
    {
      "epoch": 0.4301026329563298,
      "grad_norm": 2.625,
      "learning_rate": 1.1631060746795468e-05,
      "loss": 0.3564,
      "step": 18900
    },
    {
      "epoch": 0.4303302004869945,
      "grad_norm": 75.5,
      "learning_rate": 1.1626416496377487e-05,
      "loss": 1.2859,
      "step": 18910
    },
    {
      "epoch": 0.4305577680176592,
      "grad_norm": 0.0140380859375,
      "learning_rate": 1.1621772245959504e-05,
      "loss": 1.2539,
      "step": 18920
    },
    {
      "epoch": 0.430785335548324,
      "grad_norm": 184.0,
      "learning_rate": 1.161712799554152e-05,
      "loss": 0.4846,
      "step": 18930
    },
    {
      "epoch": 0.4310129030789887,
      "grad_norm": 15.5625,
      "learning_rate": 1.1612483745123538e-05,
      "loss": 0.4321,
      "step": 18940
    },
    {
      "epoch": 0.4312404706096534,
      "grad_norm": 304.0,
      "learning_rate": 1.1607839494705555e-05,
      "loss": 1.1572,
      "step": 18950
    },
    {
      "epoch": 0.43146803814031814,
      "grad_norm": 169.0,
      "learning_rate": 1.1603195244287572e-05,
      "loss": 0.7285,
      "step": 18960
    },
    {
      "epoch": 0.43169560567098286,
      "grad_norm": 163.0,
      "learning_rate": 1.1598550993869591e-05,
      "loss": 0.9689,
      "step": 18970
    },
    {
      "epoch": 0.4319231732016476,
      "grad_norm": 0.06982421875,
      "learning_rate": 1.1593906743451608e-05,
      "loss": 0.8096,
      "step": 18980
    },
    {
      "epoch": 0.4321507407323123,
      "grad_norm": 300.0,
      "learning_rate": 1.1589262493033627e-05,
      "loss": 0.9258,
      "step": 18990
    },
    {
      "epoch": 0.432378308262977,
      "grad_norm": 17.375,
      "learning_rate": 1.1584618242615642e-05,
      "loss": 0.4611,
      "step": 19000
    },
    {
      "epoch": 0.43260587579364174,
      "grad_norm": 828.0,
      "learning_rate": 1.1579973992197659e-05,
      "loss": 0.8808,
      "step": 19010
    },
    {
      "epoch": 0.4328334433243065,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.1575329741779678e-05,
      "loss": 0.5733,
      "step": 19020
    },
    {
      "epoch": 0.43306101085497123,
      "grad_norm": 127.5,
      "learning_rate": 1.1570685491361695e-05,
      "loss": 0.8962,
      "step": 19030
    },
    {
      "epoch": 0.43328857838563595,
      "grad_norm": 0.0296630859375,
      "learning_rate": 1.1566041240943712e-05,
      "loss": 0.7101,
      "step": 19040
    },
    {
      "epoch": 0.43351614591630067,
      "grad_norm": 28.125,
      "learning_rate": 1.156139699052573e-05,
      "loss": 0.9459,
      "step": 19050
    },
    {
      "epoch": 0.4337437134469654,
      "grad_norm": 0.0185546875,
      "learning_rate": 1.1556752740107748e-05,
      "loss": 1.2666,
      "step": 19060
    },
    {
      "epoch": 0.4339712809776301,
      "grad_norm": 0.06396484375,
      "learning_rate": 1.1552108489689763e-05,
      "loss": 1.1724,
      "step": 19070
    },
    {
      "epoch": 0.4341988485082948,
      "grad_norm": 73.0,
      "learning_rate": 1.1547464239271782e-05,
      "loss": 0.938,
      "step": 19080
    },
    {
      "epoch": 0.43442641603895954,
      "grad_norm": 75.0,
      "learning_rate": 1.1542819988853799e-05,
      "loss": 0.6739,
      "step": 19090
    },
    {
      "epoch": 0.4346539835696243,
      "grad_norm": 101.0,
      "learning_rate": 1.1538175738435818e-05,
      "loss": 0.4455,
      "step": 19100
    },
    {
      "epoch": 0.43488155110028903,
      "grad_norm": 79.5,
      "learning_rate": 1.1533531488017835e-05,
      "loss": 0.7655,
      "step": 19110
    },
    {
      "epoch": 0.43510911863095375,
      "grad_norm": 294.0,
      "learning_rate": 1.1528887237599854e-05,
      "loss": 1.2181,
      "step": 19120
    },
    {
      "epoch": 0.43533668616161847,
      "grad_norm": 0.0010986328125,
      "learning_rate": 1.152424298718187e-05,
      "loss": 0.8817,
      "step": 19130
    },
    {
      "epoch": 0.4355642536922832,
      "grad_norm": 52.25,
      "learning_rate": 1.1519598736763886e-05,
      "loss": 0.5193,
      "step": 19140
    },
    {
      "epoch": 0.4357918212229479,
      "grad_norm": 55.0,
      "learning_rate": 1.1514954486345905e-05,
      "loss": 1.073,
      "step": 19150
    },
    {
      "epoch": 0.4360193887536126,
      "grad_norm": 980.0,
      "learning_rate": 1.1510310235927922e-05,
      "loss": 1.5026,
      "step": 19160
    },
    {
      "epoch": 0.43624695628427734,
      "grad_norm": 34.5,
      "learning_rate": 1.1505665985509939e-05,
      "loss": 0.2761,
      "step": 19170
    },
    {
      "epoch": 0.43647452381494206,
      "grad_norm": 1.8359375,
      "learning_rate": 1.1501021735091958e-05,
      "loss": 0.431,
      "step": 19180
    },
    {
      "epoch": 0.43670209134560684,
      "grad_norm": 0.58203125,
      "learning_rate": 1.1496377484673975e-05,
      "loss": 0.4126,
      "step": 19190
    },
    {
      "epoch": 0.43692965887627155,
      "grad_norm": 127.0,
      "learning_rate": 1.1491733234255993e-05,
      "loss": 0.2884,
      "step": 19200
    },
    {
      "epoch": 0.4371572264069363,
      "grad_norm": 150.0,
      "learning_rate": 1.1487088983838009e-05,
      "loss": 1.3961,
      "step": 19210
    },
    {
      "epoch": 0.437384793937601,
      "grad_norm": 246.0,
      "learning_rate": 1.1482444733420026e-05,
      "loss": 1.74,
      "step": 19220
    },
    {
      "epoch": 0.4376123614682657,
      "grad_norm": 0.00147247314453125,
      "learning_rate": 1.1477800483002045e-05,
      "loss": 0.3923,
      "step": 19230
    },
    {
      "epoch": 0.4378399289989304,
      "grad_norm": 64.5,
      "learning_rate": 1.1473156232584062e-05,
      "loss": 0.6952,
      "step": 19240
    },
    {
      "epoch": 0.43806749652959515,
      "grad_norm": 127.0,
      "learning_rate": 1.146851198216608e-05,
      "loss": 1.0268,
      "step": 19250
    },
    {
      "epoch": 0.43829506406025986,
      "grad_norm": 104.0,
      "learning_rate": 1.1463867731748097e-05,
      "loss": 0.3342,
      "step": 19260
    },
    {
      "epoch": 0.4385226315909246,
      "grad_norm": 149.0,
      "learning_rate": 1.1459223481330114e-05,
      "loss": 1.2058,
      "step": 19270
    },
    {
      "epoch": 0.43875019912158936,
      "grad_norm": 0.11962890625,
      "learning_rate": 1.1454579230912131e-05,
      "loss": 0.5729,
      "step": 19280
    },
    {
      "epoch": 0.4389777666522541,
      "grad_norm": 184.0,
      "learning_rate": 1.1449934980494149e-05,
      "loss": 0.6421,
      "step": 19290
    },
    {
      "epoch": 0.4392053341829188,
      "grad_norm": 358.0,
      "learning_rate": 1.1445290730076166e-05,
      "loss": 0.8291,
      "step": 19300
    },
    {
      "epoch": 0.4394329017135835,
      "grad_norm": 96.5,
      "learning_rate": 1.1440646479658184e-05,
      "loss": 0.6095,
      "step": 19310
    },
    {
      "epoch": 0.43966046924424823,
      "grad_norm": 89.0,
      "learning_rate": 1.1436002229240201e-05,
      "loss": 0.8361,
      "step": 19320
    },
    {
      "epoch": 0.43988803677491295,
      "grad_norm": 386.0,
      "learning_rate": 1.143135797882222e-05,
      "loss": 0.7607,
      "step": 19330
    },
    {
      "epoch": 0.44011560430557767,
      "grad_norm": 119.0,
      "learning_rate": 1.1426713728404237e-05,
      "loss": 0.7973,
      "step": 19340
    },
    {
      "epoch": 0.4403431718362424,
      "grad_norm": 239.0,
      "learning_rate": 1.1422069477986253e-05,
      "loss": 1.4038,
      "step": 19350
    },
    {
      "epoch": 0.4405707393669071,
      "grad_norm": 146.0,
      "learning_rate": 1.1417425227568271e-05,
      "loss": 0.5638,
      "step": 19360
    },
    {
      "epoch": 0.4407983068975719,
      "grad_norm": 0.00201416015625,
      "learning_rate": 1.1412780977150288e-05,
      "loss": 0.829,
      "step": 19370
    },
    {
      "epoch": 0.4410258744282366,
      "grad_norm": 42.25,
      "learning_rate": 1.1408136726732305e-05,
      "loss": 0.5206,
      "step": 19380
    },
    {
      "epoch": 0.4412534419589013,
      "grad_norm": 166.0,
      "learning_rate": 1.1403492476314324e-05,
      "loss": 1.0747,
      "step": 19390
    },
    {
      "epoch": 0.44148100948956603,
      "grad_norm": 0.21875,
      "learning_rate": 1.1398848225896341e-05,
      "loss": 0.6069,
      "step": 19400
    },
    {
      "epoch": 0.44170857702023075,
      "grad_norm": 94.5,
      "learning_rate": 1.139420397547836e-05,
      "loss": 0.3724,
      "step": 19410
    },
    {
      "epoch": 0.44193614455089547,
      "grad_norm": 85.0,
      "learning_rate": 1.1389559725060375e-05,
      "loss": 0.8044,
      "step": 19420
    },
    {
      "epoch": 0.4421637120815602,
      "grad_norm": 428.0,
      "learning_rate": 1.1384915474642392e-05,
      "loss": 0.7198,
      "step": 19430
    },
    {
      "epoch": 0.4423912796122249,
      "grad_norm": 0.031494140625,
      "learning_rate": 1.1380271224224411e-05,
      "loss": 1.1538,
      "step": 19440
    },
    {
      "epoch": 0.4426188471428896,
      "grad_norm": 84.5,
      "learning_rate": 1.1375626973806428e-05,
      "loss": 0.3187,
      "step": 19450
    },
    {
      "epoch": 0.4428464146735544,
      "grad_norm": 0.0115966796875,
      "learning_rate": 1.1370982723388447e-05,
      "loss": 1.1246,
      "step": 19460
    },
    {
      "epoch": 0.4430739822042191,
      "grad_norm": 109.5,
      "learning_rate": 1.1366338472970464e-05,
      "loss": 0.7567,
      "step": 19470
    },
    {
      "epoch": 0.44330154973488384,
      "grad_norm": 31.875,
      "learning_rate": 1.1361694222552481e-05,
      "loss": 0.4627,
      "step": 19480
    },
    {
      "epoch": 0.44352911726554856,
      "grad_norm": 2.578125,
      "learning_rate": 1.1357049972134498e-05,
      "loss": 0.6854,
      "step": 19490
    },
    {
      "epoch": 0.4437566847962133,
      "grad_norm": 73.5,
      "learning_rate": 1.1352405721716515e-05,
      "loss": 0.5787,
      "step": 19500
    },
    {
      "epoch": 0.443984252326878,
      "grad_norm": 107.5,
      "learning_rate": 1.1347761471298532e-05,
      "loss": 0.7934,
      "step": 19510
    },
    {
      "epoch": 0.4442118198575427,
      "grad_norm": 272.0,
      "learning_rate": 1.1343117220880551e-05,
      "loss": 0.7152,
      "step": 19520
    },
    {
      "epoch": 0.44443938738820743,
      "grad_norm": 145.0,
      "learning_rate": 1.1338472970462568e-05,
      "loss": 0.8625,
      "step": 19530
    },
    {
      "epoch": 0.44466695491887215,
      "grad_norm": 97.0,
      "learning_rate": 1.1333828720044587e-05,
      "loss": 0.4079,
      "step": 19540
    },
    {
      "epoch": 0.4448945224495369,
      "grad_norm": 1048.0,
      "learning_rate": 1.1329184469626604e-05,
      "loss": 1.1945,
      "step": 19550
    },
    {
      "epoch": 0.44512208998020164,
      "grad_norm": 146.0,
      "learning_rate": 1.132454021920862e-05,
      "loss": 0.9906,
      "step": 19560
    },
    {
      "epoch": 0.44534965751086636,
      "grad_norm": 76.5,
      "learning_rate": 1.1319895968790638e-05,
      "loss": 0.7638,
      "step": 19570
    },
    {
      "epoch": 0.4455772250415311,
      "grad_norm": 76.5,
      "learning_rate": 1.1315251718372655e-05,
      "loss": 1.132,
      "step": 19580
    },
    {
      "epoch": 0.4458047925721958,
      "grad_norm": 131.0,
      "learning_rate": 1.1310607467954674e-05,
      "loss": 0.5304,
      "step": 19590
    },
    {
      "epoch": 0.4460323601028605,
      "grad_norm": 99.5,
      "learning_rate": 1.130596321753669e-05,
      "loss": 1.0953,
      "step": 19600
    },
    {
      "epoch": 0.44625992763352523,
      "grad_norm": 184.0,
      "learning_rate": 1.1301318967118708e-05,
      "loss": 0.5375,
      "step": 19610
    },
    {
      "epoch": 0.44648749516418995,
      "grad_norm": 203.0,
      "learning_rate": 1.1296674716700727e-05,
      "loss": 0.4984,
      "step": 19620
    },
    {
      "epoch": 0.4467150626948547,
      "grad_norm": 0.00066375732421875,
      "learning_rate": 1.1292030466282742e-05,
      "loss": 1.0009,
      "step": 19630
    },
    {
      "epoch": 0.44694263022551944,
      "grad_norm": 139.0,
      "learning_rate": 1.1287386215864759e-05,
      "loss": 1.4666,
      "step": 19640
    },
    {
      "epoch": 0.44717019775618416,
      "grad_norm": 34.0,
      "learning_rate": 1.1282741965446778e-05,
      "loss": 0.1788,
      "step": 19650
    },
    {
      "epoch": 0.4473977652868489,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.1278097715028795e-05,
      "loss": 0.1462,
      "step": 19660
    },
    {
      "epoch": 0.4476253328175136,
      "grad_norm": 0.169921875,
      "learning_rate": 1.1273453464610814e-05,
      "loss": 0.8998,
      "step": 19670
    },
    {
      "epoch": 0.4478529003481783,
      "grad_norm": 232.0,
      "learning_rate": 1.126880921419283e-05,
      "loss": 0.9663,
      "step": 19680
    },
    {
      "epoch": 0.44808046787884304,
      "grad_norm": 948.0,
      "learning_rate": 1.1264164963774848e-05,
      "loss": 0.8919,
      "step": 19690
    },
    {
      "epoch": 0.44830803540950775,
      "grad_norm": 0.08642578125,
      "learning_rate": 1.1259520713356865e-05,
      "loss": 0.7653,
      "step": 19700
    },
    {
      "epoch": 0.4485356029401725,
      "grad_norm": 1.46875,
      "learning_rate": 1.1254876462938882e-05,
      "loss": 0.6203,
      "step": 19710
    },
    {
      "epoch": 0.44876317047083725,
      "grad_norm": 103.5,
      "learning_rate": 1.1250232212520899e-05,
      "loss": 0.8217,
      "step": 19720
    },
    {
      "epoch": 0.44899073800150197,
      "grad_norm": 0.00927734375,
      "learning_rate": 1.1245587962102918e-05,
      "loss": 0.8687,
      "step": 19730
    },
    {
      "epoch": 0.4492183055321667,
      "grad_norm": 213.0,
      "learning_rate": 1.1240943711684935e-05,
      "loss": 0.5222,
      "step": 19740
    },
    {
      "epoch": 0.4494458730628314,
      "grad_norm": 75.0,
      "learning_rate": 1.1236299461266953e-05,
      "loss": 0.7772,
      "step": 19750
    },
    {
      "epoch": 0.4496734405934961,
      "grad_norm": 0.0032958984375,
      "learning_rate": 1.123165521084897e-05,
      "loss": 1.6588,
      "step": 19760
    },
    {
      "epoch": 0.44990100812416084,
      "grad_norm": 31.75,
      "learning_rate": 1.1227010960430986e-05,
      "loss": 0.8505,
      "step": 19770
    },
    {
      "epoch": 0.45012857565482556,
      "grad_norm": 548.0,
      "learning_rate": 1.1222366710013005e-05,
      "loss": 1.2073,
      "step": 19780
    },
    {
      "epoch": 0.4503561431854903,
      "grad_norm": 180.0,
      "learning_rate": 1.1217722459595022e-05,
      "loss": 1.5482,
      "step": 19790
    },
    {
      "epoch": 0.450583710716155,
      "grad_norm": 0.00592041015625,
      "learning_rate": 1.121307820917704e-05,
      "loss": 1.486,
      "step": 19800
    },
    {
      "epoch": 0.45081127824681977,
      "grad_norm": 217.0,
      "learning_rate": 1.1208433958759057e-05,
      "loss": 0.7636,
      "step": 19810
    },
    {
      "epoch": 0.4510388457774845,
      "grad_norm": 26.75,
      "learning_rate": 1.1203789708341074e-05,
      "loss": 0.3005,
      "step": 19820
    },
    {
      "epoch": 0.4512664133081492,
      "grad_norm": 172.0,
      "learning_rate": 1.1199145457923093e-05,
      "loss": 1.0971,
      "step": 19830
    },
    {
      "epoch": 0.4514939808388139,
      "grad_norm": 143.0,
      "learning_rate": 1.1194501207505109e-05,
      "loss": 1.3792,
      "step": 19840
    },
    {
      "epoch": 0.45172154836947864,
      "grad_norm": 23.0,
      "learning_rate": 1.1189856957087126e-05,
      "loss": 0.8751,
      "step": 19850
    },
    {
      "epoch": 0.45194911590014336,
      "grad_norm": 240.0,
      "learning_rate": 1.1185212706669144e-05,
      "loss": 0.6203,
      "step": 19860
    },
    {
      "epoch": 0.4521766834308081,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.1180568456251161e-05,
      "loss": 0.9038,
      "step": 19870
    },
    {
      "epoch": 0.4524042509614728,
      "grad_norm": 18.5,
      "learning_rate": 1.117592420583318e-05,
      "loss": 0.559,
      "step": 19880
    },
    {
      "epoch": 0.4526318184921375,
      "grad_norm": 340.0,
      "learning_rate": 1.1171279955415197e-05,
      "loss": 1.44,
      "step": 19890
    },
    {
      "epoch": 0.4528593860228023,
      "grad_norm": 0.0025177001953125,
      "learning_rate": 1.1166635704997216e-05,
      "loss": 0.3392,
      "step": 19900
    },
    {
      "epoch": 0.453086953553467,
      "grad_norm": 144.0,
      "learning_rate": 1.1161991454579231e-05,
      "loss": 0.5448,
      "step": 19910
    },
    {
      "epoch": 0.4533145210841317,
      "grad_norm": 174.0,
      "learning_rate": 1.1157347204161248e-05,
      "loss": 0.3945,
      "step": 19920
    },
    {
      "epoch": 0.45354208861479645,
      "grad_norm": 103.5,
      "learning_rate": 1.1152702953743265e-05,
      "loss": 0.8188,
      "step": 19930
    },
    {
      "epoch": 0.45376965614546116,
      "grad_norm": 161.0,
      "learning_rate": 1.1148058703325284e-05,
      "loss": 0.8045,
      "step": 19940
    },
    {
      "epoch": 0.4539972236761259,
      "grad_norm": 117.0,
      "learning_rate": 1.1143414452907301e-05,
      "loss": 0.4256,
      "step": 19950
    },
    {
      "epoch": 0.4542247912067906,
      "grad_norm": 114.5,
      "learning_rate": 1.113877020248932e-05,
      "loss": 0.3405,
      "step": 19960
    },
    {
      "epoch": 0.4544523587374553,
      "grad_norm": 180.0,
      "learning_rate": 1.1134125952071337e-05,
      "loss": 0.8737,
      "step": 19970
    },
    {
      "epoch": 0.45467992626812004,
      "grad_norm": 1.0859375,
      "learning_rate": 1.1129481701653352e-05,
      "loss": 0.7511,
      "step": 19980
    },
    {
      "epoch": 0.4549074937987848,
      "grad_norm": 0.0074462890625,
      "learning_rate": 1.1124837451235371e-05,
      "loss": 0.4989,
      "step": 19990
    },
    {
      "epoch": 0.45513506132944953,
      "grad_norm": 159.0,
      "learning_rate": 1.1120193200817388e-05,
      "loss": 0.7811,
      "step": 20000
    },
    {
      "epoch": 0.45536262886011425,
      "grad_norm": 136.0,
      "learning_rate": 1.1115548950399407e-05,
      "loss": 0.9338,
      "step": 20010
    },
    {
      "epoch": 0.45559019639077897,
      "grad_norm": 0.039794921875,
      "learning_rate": 1.1110904699981424e-05,
      "loss": 0.6635,
      "step": 20020
    },
    {
      "epoch": 0.4558177639214437,
      "grad_norm": 100.0,
      "learning_rate": 1.1106260449563441e-05,
      "loss": 1.1098,
      "step": 20030
    },
    {
      "epoch": 0.4560453314521084,
      "grad_norm": 276.0,
      "learning_rate": 1.110161619914546e-05,
      "loss": 1.3526,
      "step": 20040
    },
    {
      "epoch": 0.4562728989827731,
      "grad_norm": 59.25,
      "learning_rate": 1.1096971948727475e-05,
      "loss": 0.9512,
      "step": 20050
    },
    {
      "epoch": 0.45650046651343784,
      "grad_norm": 10.375,
      "learning_rate": 1.1092327698309492e-05,
      "loss": 0.1804,
      "step": 20060
    },
    {
      "epoch": 0.45672803404410256,
      "grad_norm": 182.0,
      "learning_rate": 1.1087683447891511e-05,
      "loss": 0.8179,
      "step": 20070
    },
    {
      "epoch": 0.45695560157476733,
      "grad_norm": 158.0,
      "learning_rate": 1.1083039197473528e-05,
      "loss": 0.9248,
      "step": 20080
    },
    {
      "epoch": 0.45718316910543205,
      "grad_norm": 39.5,
      "learning_rate": 1.1078394947055547e-05,
      "loss": 0.8427,
      "step": 20090
    },
    {
      "epoch": 0.45741073663609677,
      "grad_norm": 0.1875,
      "learning_rate": 1.1073750696637564e-05,
      "loss": 0.7494,
      "step": 20100
    },
    {
      "epoch": 0.4576383041667615,
      "grad_norm": 0.041015625,
      "learning_rate": 1.1069106446219583e-05,
      "loss": 0.3357,
      "step": 20110
    },
    {
      "epoch": 0.4578658716974262,
      "grad_norm": 85.0,
      "learning_rate": 1.10644621958016e-05,
      "loss": 0.7039,
      "step": 20120
    },
    {
      "epoch": 0.4580934392280909,
      "grad_norm": 102.5,
      "learning_rate": 1.1059817945383615e-05,
      "loss": 0.3867,
      "step": 20130
    },
    {
      "epoch": 0.45832100675875564,
      "grad_norm": 130.0,
      "learning_rate": 1.1055173694965634e-05,
      "loss": 0.3962,
      "step": 20140
    },
    {
      "epoch": 0.45854857428942036,
      "grad_norm": 340.0,
      "learning_rate": 1.105052944454765e-05,
      "loss": 0.782,
      "step": 20150
    },
    {
      "epoch": 0.45877614182008514,
      "grad_norm": 11.9375,
      "learning_rate": 1.1045885194129668e-05,
      "loss": 0.9465,
      "step": 20160
    },
    {
      "epoch": 0.45900370935074986,
      "grad_norm": 0.095703125,
      "learning_rate": 1.1041240943711687e-05,
      "loss": 0.9731,
      "step": 20170
    },
    {
      "epoch": 0.4592312768814146,
      "grad_norm": 136.0,
      "learning_rate": 1.1036596693293704e-05,
      "loss": 0.7633,
      "step": 20180
    },
    {
      "epoch": 0.4594588444120793,
      "grad_norm": 268.0,
      "learning_rate": 1.1031952442875722e-05,
      "loss": 1.1034,
      "step": 20190
    },
    {
      "epoch": 0.459686411942744,
      "grad_norm": 8.75,
      "learning_rate": 1.1027308192457738e-05,
      "loss": 0.3671,
      "step": 20200
    },
    {
      "epoch": 0.45991397947340873,
      "grad_norm": 142.0,
      "learning_rate": 1.1022663942039755e-05,
      "loss": 1.1564,
      "step": 20210
    },
    {
      "epoch": 0.46014154700407345,
      "grad_norm": 43.25,
      "learning_rate": 1.1018019691621774e-05,
      "loss": 1.7516,
      "step": 20220
    },
    {
      "epoch": 0.46036911453473817,
      "grad_norm": 0.06103515625,
      "learning_rate": 1.101337544120379e-05,
      "loss": 1.276,
      "step": 20230
    },
    {
      "epoch": 0.4605966820654029,
      "grad_norm": 0.345703125,
      "learning_rate": 1.1008731190785808e-05,
      "loss": 0.8124,
      "step": 20240
    },
    {
      "epoch": 0.46082424959606766,
      "grad_norm": 164.0,
      "learning_rate": 1.1004086940367826e-05,
      "loss": 0.6578,
      "step": 20250
    },
    {
      "epoch": 0.4610518171267324,
      "grad_norm": 57.0,
      "learning_rate": 1.0999442689949844e-05,
      "loss": 1.6422,
      "step": 20260
    },
    {
      "epoch": 0.4612793846573971,
      "grad_norm": 0.002044677734375,
      "learning_rate": 1.0994798439531859e-05,
      "loss": 0.2822,
      "step": 20270
    },
    {
      "epoch": 0.4615069521880618,
      "grad_norm": 0.0034027099609375,
      "learning_rate": 1.0990154189113878e-05,
      "loss": 0.1614,
      "step": 20280
    },
    {
      "epoch": 0.46173451971872653,
      "grad_norm": 21.625,
      "learning_rate": 1.0985509938695895e-05,
      "loss": 0.6907,
      "step": 20290
    },
    {
      "epoch": 0.46196208724939125,
      "grad_norm": 199.0,
      "learning_rate": 1.0980865688277913e-05,
      "loss": 0.4411,
      "step": 20300
    },
    {
      "epoch": 0.46218965478005597,
      "grad_norm": 146.0,
      "learning_rate": 1.097622143785993e-05,
      "loss": 0.9095,
      "step": 20310
    },
    {
      "epoch": 0.4624172223107207,
      "grad_norm": 760.0,
      "learning_rate": 1.097157718744195e-05,
      "loss": 0.4325,
      "step": 20320
    },
    {
      "epoch": 0.4626447898413854,
      "grad_norm": 31.375,
      "learning_rate": 1.0966932937023966e-05,
      "loss": 0.4029,
      "step": 20330
    },
    {
      "epoch": 0.4628723573720502,
      "grad_norm": 9.0,
      "learning_rate": 1.0962288686605982e-05,
      "loss": 0.9176,
      "step": 20340
    },
    {
      "epoch": 0.4630999249027149,
      "grad_norm": 121.0,
      "learning_rate": 1.0957644436188e-05,
      "loss": 1.0816,
      "step": 20350
    },
    {
      "epoch": 0.4633274924333796,
      "grad_norm": 136.0,
      "learning_rate": 1.0953000185770017e-05,
      "loss": 0.8189,
      "step": 20360
    },
    {
      "epoch": 0.46355505996404434,
      "grad_norm": 0.0023651123046875,
      "learning_rate": 1.0948355935352035e-05,
      "loss": 0.6467,
      "step": 20370
    },
    {
      "epoch": 0.46378262749470905,
      "grad_norm": 224.0,
      "learning_rate": 1.0943711684934053e-05,
      "loss": 1.314,
      "step": 20380
    },
    {
      "epoch": 0.4640101950253738,
      "grad_norm": 17.125,
      "learning_rate": 1.093906743451607e-05,
      "loss": 0.368,
      "step": 20390
    },
    {
      "epoch": 0.4642377625560385,
      "grad_norm": 0.0255126953125,
      "learning_rate": 1.0934423184098089e-05,
      "loss": 0.7684,
      "step": 20400
    },
    {
      "epoch": 0.4644653300867032,
      "grad_norm": 0.000606536865234375,
      "learning_rate": 1.0929778933680104e-05,
      "loss": 0.7655,
      "step": 20410
    },
    {
      "epoch": 0.46469289761736793,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.0925134683262121e-05,
      "loss": 0.5796,
      "step": 20420
    },
    {
      "epoch": 0.4649204651480327,
      "grad_norm": 93.5,
      "learning_rate": 1.092049043284414e-05,
      "loss": 1.0314,
      "step": 20430
    },
    {
      "epoch": 0.4651480326786974,
      "grad_norm": 67.0,
      "learning_rate": 1.0915846182426157e-05,
      "loss": 0.9501,
      "step": 20440
    },
    {
      "epoch": 0.46537560020936214,
      "grad_norm": 65.5,
      "learning_rate": 1.0911201932008176e-05,
      "loss": 0.9243,
      "step": 20450
    },
    {
      "epoch": 0.46560316774002686,
      "grad_norm": 189.0,
      "learning_rate": 1.0906557681590193e-05,
      "loss": 0.9611,
      "step": 20460
    },
    {
      "epoch": 0.4658307352706916,
      "grad_norm": 8.1875,
      "learning_rate": 1.090191343117221e-05,
      "loss": 0.9024,
      "step": 20470
    },
    {
      "epoch": 0.4660583028013563,
      "grad_norm": 0.0003566741943359375,
      "learning_rate": 1.0897269180754226e-05,
      "loss": 0.132,
      "step": 20480
    },
    {
      "epoch": 0.466285870332021,
      "grad_norm": 0.36328125,
      "learning_rate": 1.0892624930336244e-05,
      "loss": 0.8165,
      "step": 20490
    },
    {
      "epoch": 0.46651343786268573,
      "grad_norm": 0.002716064453125,
      "learning_rate": 1.0887980679918261e-05,
      "loss": 0.9115,
      "step": 20500
    },
    {
      "epoch": 0.46674100539335045,
      "grad_norm": 264.0,
      "learning_rate": 1.088333642950028e-05,
      "loss": 1.2762,
      "step": 20510
    },
    {
      "epoch": 0.4669685729240152,
      "grad_norm": 448.0,
      "learning_rate": 1.0878692179082297e-05,
      "loss": 0.8869,
      "step": 20520
    },
    {
      "epoch": 0.46719614045467994,
      "grad_norm": 91.0,
      "learning_rate": 1.0874047928664316e-05,
      "loss": 0.3867,
      "step": 20530
    },
    {
      "epoch": 0.46742370798534466,
      "grad_norm": 11.75,
      "learning_rate": 1.0869403678246333e-05,
      "loss": 0.4353,
      "step": 20540
    },
    {
      "epoch": 0.4676512755160094,
      "grad_norm": 0.0022430419921875,
      "learning_rate": 1.0864759427828348e-05,
      "loss": 0.5758,
      "step": 20550
    },
    {
      "epoch": 0.4678788430466741,
      "grad_norm": 147.0,
      "learning_rate": 1.0860115177410367e-05,
      "loss": 1.1936,
      "step": 20560
    },
    {
      "epoch": 0.4681064105773388,
      "grad_norm": 0.00116729736328125,
      "learning_rate": 1.0855470926992384e-05,
      "loss": 1.103,
      "step": 20570
    },
    {
      "epoch": 0.46833397810800353,
      "grad_norm": 98.0,
      "learning_rate": 1.0850826676574401e-05,
      "loss": 0.6932,
      "step": 20580
    },
    {
      "epoch": 0.46856154563866825,
      "grad_norm": 0.004180908203125,
      "learning_rate": 1.084618242615642e-05,
      "loss": 1.0805,
      "step": 20590
    },
    {
      "epoch": 0.468789113169333,
      "grad_norm": 8.0,
      "learning_rate": 1.0841538175738437e-05,
      "loss": 0.0198,
      "step": 20600
    },
    {
      "epoch": 0.46901668069999775,
      "grad_norm": 0.2734375,
      "learning_rate": 1.0836893925320456e-05,
      "loss": 0.8611,
      "step": 20610
    },
    {
      "epoch": 0.46924424823066246,
      "grad_norm": 0.05517578125,
      "learning_rate": 1.0832249674902471e-05,
      "loss": 0.9615,
      "step": 20620
    },
    {
      "epoch": 0.4694718157613272,
      "grad_norm": 0.9375,
      "learning_rate": 1.0827605424484488e-05,
      "loss": 0.6407,
      "step": 20630
    },
    {
      "epoch": 0.4696993832919919,
      "grad_norm": 13.125,
      "learning_rate": 1.0822961174066507e-05,
      "loss": 0.7758,
      "step": 20640
    },
    {
      "epoch": 0.4699269508226566,
      "grad_norm": 75.5,
      "learning_rate": 1.0818316923648524e-05,
      "loss": 1.1411,
      "step": 20650
    },
    {
      "epoch": 0.47015451835332134,
      "grad_norm": 0.140625,
      "learning_rate": 1.0813672673230543e-05,
      "loss": 0.6177,
      "step": 20660
    },
    {
      "epoch": 0.47038208588398606,
      "grad_norm": 59.25,
      "learning_rate": 1.080902842281256e-05,
      "loss": 0.4813,
      "step": 20670
    },
    {
      "epoch": 0.4706096534146508,
      "grad_norm": 233.0,
      "learning_rate": 1.0804384172394577e-05,
      "loss": 0.7685,
      "step": 20680
    },
    {
      "epoch": 0.47083722094531555,
      "grad_norm": 80.5,
      "learning_rate": 1.0799739921976594e-05,
      "loss": 2.3902,
      "step": 20690
    },
    {
      "epoch": 0.47106478847598027,
      "grad_norm": 195.0,
      "learning_rate": 1.0795095671558611e-05,
      "loss": 2.2526,
      "step": 20700
    },
    {
      "epoch": 0.471292356006645,
      "grad_norm": 103.5,
      "learning_rate": 1.0790451421140628e-05,
      "loss": 0.8075,
      "step": 20710
    },
    {
      "epoch": 0.4715199235373097,
      "grad_norm": 159.0,
      "learning_rate": 1.0785807170722647e-05,
      "loss": 0.2975,
      "step": 20720
    },
    {
      "epoch": 0.4717474910679744,
      "grad_norm": 0.0191650390625,
      "learning_rate": 1.0781162920304664e-05,
      "loss": 0.3904,
      "step": 20730
    },
    {
      "epoch": 0.47197505859863914,
      "grad_norm": 94.0,
      "learning_rate": 1.0776518669886682e-05,
      "loss": 0.9472,
      "step": 20740
    },
    {
      "epoch": 0.47220262612930386,
      "grad_norm": 147.0,
      "learning_rate": 1.07718744194687e-05,
      "loss": 1.5331,
      "step": 20750
    },
    {
      "epoch": 0.4724301936599686,
      "grad_norm": 23.0,
      "learning_rate": 1.0767230169050715e-05,
      "loss": 0.7488,
      "step": 20760
    },
    {
      "epoch": 0.4726577611906333,
      "grad_norm": 56.0,
      "learning_rate": 1.0762585918632734e-05,
      "loss": 0.7023,
      "step": 20770
    },
    {
      "epoch": 0.47288532872129807,
      "grad_norm": 428.0,
      "learning_rate": 1.075794166821475e-05,
      "loss": 1.1723,
      "step": 20780
    },
    {
      "epoch": 0.4731128962519628,
      "grad_norm": 5.75,
      "learning_rate": 1.075329741779677e-05,
      "loss": 1.0845,
      "step": 20790
    },
    {
      "epoch": 0.4733404637826275,
      "grad_norm": 380.0,
      "learning_rate": 1.0748653167378787e-05,
      "loss": 0.8337,
      "step": 20800
    },
    {
      "epoch": 0.4735680313132922,
      "grad_norm": 136.0,
      "learning_rate": 1.0744008916960804e-05,
      "loss": 1.8259,
      "step": 20810
    },
    {
      "epoch": 0.47379559884395694,
      "grad_norm": 43.0,
      "learning_rate": 1.0739364666542822e-05,
      "loss": 0.8661,
      "step": 20820
    },
    {
      "epoch": 0.47402316637462166,
      "grad_norm": 138.0,
      "learning_rate": 1.0734720416124838e-05,
      "loss": 0.3491,
      "step": 20830
    },
    {
      "epoch": 0.4742507339052864,
      "grad_norm": 93.5,
      "learning_rate": 1.0730076165706855e-05,
      "loss": 0.1836,
      "step": 20840
    },
    {
      "epoch": 0.4744783014359511,
      "grad_norm": 0.2119140625,
      "learning_rate": 1.0725431915288873e-05,
      "loss": 0.5908,
      "step": 20850
    },
    {
      "epoch": 0.4747058689666158,
      "grad_norm": 54.0,
      "learning_rate": 1.072078766487089e-05,
      "loss": 0.2546,
      "step": 20860
    },
    {
      "epoch": 0.4749334364972806,
      "grad_norm": 0.003814697265625,
      "learning_rate": 1.071614341445291e-05,
      "loss": 0.2712,
      "step": 20870
    },
    {
      "epoch": 0.4751610040279453,
      "grad_norm": 0.5,
      "learning_rate": 1.0711499164034926e-05,
      "loss": 0.6389,
      "step": 20880
    },
    {
      "epoch": 0.47538857155861003,
      "grad_norm": 0.259765625,
      "learning_rate": 1.0706854913616943e-05,
      "loss": 0.2218,
      "step": 20890
    },
    {
      "epoch": 0.47561613908927475,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.070221066319896e-05,
      "loss": 1.0831,
      "step": 20900
    },
    {
      "epoch": 0.47584370661993947,
      "grad_norm": 104.0,
      "learning_rate": 1.0697566412780978e-05,
      "loss": 0.3012,
      "step": 20910
    },
    {
      "epoch": 0.4760712741506042,
      "grad_norm": 0.0703125,
      "learning_rate": 1.0692922162362995e-05,
      "loss": 0.4566,
      "step": 20920
    },
    {
      "epoch": 0.4762988416812689,
      "grad_norm": 452.0,
      "learning_rate": 1.0688277911945013e-05,
      "loss": 0.811,
      "step": 20930
    },
    {
      "epoch": 0.4765264092119336,
      "grad_norm": 206.0,
      "learning_rate": 1.068363366152703e-05,
      "loss": 0.4731,
      "step": 20940
    },
    {
      "epoch": 0.47675397674259834,
      "grad_norm": 418.0,
      "learning_rate": 1.0678989411109049e-05,
      "loss": 1.0727,
      "step": 20950
    },
    {
      "epoch": 0.4769815442732631,
      "grad_norm": 0.1103515625,
      "learning_rate": 1.0674345160691066e-05,
      "loss": 0.9641,
      "step": 20960
    },
    {
      "epoch": 0.47720911180392783,
      "grad_norm": 0.03564453125,
      "learning_rate": 1.0669700910273082e-05,
      "loss": 0.7316,
      "step": 20970
    },
    {
      "epoch": 0.47743667933459255,
      "grad_norm": 106.5,
      "learning_rate": 1.06650566598551e-05,
      "loss": 0.7342,
      "step": 20980
    },
    {
      "epoch": 0.47766424686525727,
      "grad_norm": 55.25,
      "learning_rate": 1.0660412409437117e-05,
      "loss": 0.4166,
      "step": 20990
    },
    {
      "epoch": 0.477891814395922,
      "grad_norm": 5.4375,
      "learning_rate": 1.0655768159019136e-05,
      "loss": 0.6077,
      "step": 21000
    },
    {
      "epoch": 0.4781193819265867,
      "grad_norm": 62.5,
      "learning_rate": 1.0651123908601153e-05,
      "loss": 0.9724,
      "step": 21010
    },
    {
      "epoch": 0.4783469494572514,
      "grad_norm": 9.5625,
      "learning_rate": 1.064647965818317e-05,
      "loss": 0.6352,
      "step": 21020
    },
    {
      "epoch": 0.47857451698791614,
      "grad_norm": 32.25,
      "learning_rate": 1.0641835407765189e-05,
      "loss": 0.9509,
      "step": 21030
    },
    {
      "epoch": 0.47880208451858086,
      "grad_norm": 123.0,
      "learning_rate": 1.0637191157347204e-05,
      "loss": 0.8612,
      "step": 21040
    },
    {
      "epoch": 0.47902965204924564,
      "grad_norm": 0.004669189453125,
      "learning_rate": 1.0632546906929221e-05,
      "loss": 1.062,
      "step": 21050
    },
    {
      "epoch": 0.47925721957991035,
      "grad_norm": 0.205078125,
      "learning_rate": 1.062790265651124e-05,
      "loss": 0.4264,
      "step": 21060
    },
    {
      "epoch": 0.4794847871105751,
      "grad_norm": 124.0,
      "learning_rate": 1.0623258406093257e-05,
      "loss": 1.4206,
      "step": 21070
    },
    {
      "epoch": 0.4797123546412398,
      "grad_norm": 131.0,
      "learning_rate": 1.0618614155675276e-05,
      "loss": 1.0404,
      "step": 21080
    },
    {
      "epoch": 0.4799399221719045,
      "grad_norm": 0.89453125,
      "learning_rate": 1.0613969905257293e-05,
      "loss": 0.5529,
      "step": 21090
    },
    {
      "epoch": 0.48016748970256923,
      "grad_norm": 62.25,
      "learning_rate": 1.0609325654839312e-05,
      "loss": 0.6335,
      "step": 21100
    },
    {
      "epoch": 0.48039505723323395,
      "grad_norm": 226.0,
      "learning_rate": 1.0604681404421327e-05,
      "loss": 1.5953,
      "step": 21110
    },
    {
      "epoch": 0.48062262476389866,
      "grad_norm": 84.0,
      "learning_rate": 1.0600037154003344e-05,
      "loss": 0.4039,
      "step": 21120
    },
    {
      "epoch": 0.48085019229456344,
      "grad_norm": 0.01226806640625,
      "learning_rate": 1.0595392903585361e-05,
      "loss": 0.494,
      "step": 21130
    },
    {
      "epoch": 0.48107775982522816,
      "grad_norm": 4.28125,
      "learning_rate": 1.059074865316738e-05,
      "loss": 1.2988,
      "step": 21140
    },
    {
      "epoch": 0.4813053273558929,
      "grad_norm": 105.5,
      "learning_rate": 1.0586104402749397e-05,
      "loss": 0.6353,
      "step": 21150
    },
    {
      "epoch": 0.4815328948865576,
      "grad_norm": 193.0,
      "learning_rate": 1.0581460152331416e-05,
      "loss": 0.756,
      "step": 21160
    },
    {
      "epoch": 0.4817604624172223,
      "grad_norm": 114.5,
      "learning_rate": 1.0576815901913433e-05,
      "loss": 0.8195,
      "step": 21170
    },
    {
      "epoch": 0.48198802994788703,
      "grad_norm": 0.0024261474609375,
      "learning_rate": 1.0572171651495448e-05,
      "loss": 0.6462,
      "step": 21180
    },
    {
      "epoch": 0.48221559747855175,
      "grad_norm": 1.5390625,
      "learning_rate": 1.0567527401077467e-05,
      "loss": 0.3883,
      "step": 21190
    },
    {
      "epoch": 0.48244316500921647,
      "grad_norm": 239.0,
      "learning_rate": 1.0562883150659484e-05,
      "loss": 0.9778,
      "step": 21200
    },
    {
      "epoch": 0.4826707325398812,
      "grad_norm": 35.5,
      "learning_rate": 1.0558238900241503e-05,
      "loss": 1.1536,
      "step": 21210
    },
    {
      "epoch": 0.48289830007054596,
      "grad_norm": 1.0078125,
      "learning_rate": 1.055359464982352e-05,
      "loss": 0.6755,
      "step": 21220
    },
    {
      "epoch": 0.4831258676012107,
      "grad_norm": 109.0,
      "learning_rate": 1.0548950399405537e-05,
      "loss": 0.97,
      "step": 21230
    },
    {
      "epoch": 0.4833534351318754,
      "grad_norm": 0.546875,
      "learning_rate": 1.0544306148987556e-05,
      "loss": 0.0725,
      "step": 21240
    },
    {
      "epoch": 0.4835810026625401,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.0539661898569571e-05,
      "loss": 0.9112,
      "step": 21250
    },
    {
      "epoch": 0.48380857019320483,
      "grad_norm": 142.0,
      "learning_rate": 1.0535017648151588e-05,
      "loss": 0.4495,
      "step": 21260
    },
    {
      "epoch": 0.48403613772386955,
      "grad_norm": 98.5,
      "learning_rate": 1.0530373397733607e-05,
      "loss": 0.84,
      "step": 21270
    },
    {
      "epoch": 0.48426370525453427,
      "grad_norm": 66.5,
      "learning_rate": 1.0525729147315624e-05,
      "loss": 0.9419,
      "step": 21280
    },
    {
      "epoch": 0.484491272785199,
      "grad_norm": 620.0,
      "learning_rate": 1.0521084896897643e-05,
      "loss": 1.346,
      "step": 21290
    },
    {
      "epoch": 0.4847188403158637,
      "grad_norm": 12.25,
      "learning_rate": 1.051644064647966e-05,
      "loss": 0.5872,
      "step": 21300
    },
    {
      "epoch": 0.4849464078465285,
      "grad_norm": 243.0,
      "learning_rate": 1.0511796396061678e-05,
      "loss": 1.2269,
      "step": 21310
    },
    {
      "epoch": 0.4851739753771932,
      "grad_norm": 286.0,
      "learning_rate": 1.0507152145643694e-05,
      "loss": 1.1672,
      "step": 21320
    },
    {
      "epoch": 0.4854015429078579,
      "grad_norm": 408.0,
      "learning_rate": 1.050250789522571e-05,
      "loss": 0.8389,
      "step": 21330
    },
    {
      "epoch": 0.48562911043852264,
      "grad_norm": 392.0,
      "learning_rate": 1.049786364480773e-05,
      "loss": 0.988,
      "step": 21340
    },
    {
      "epoch": 0.48585667796918736,
      "grad_norm": 60.0,
      "learning_rate": 1.0493219394389747e-05,
      "loss": 1.0203,
      "step": 21350
    },
    {
      "epoch": 0.4860842454998521,
      "grad_norm": 0.02197265625,
      "learning_rate": 1.0488575143971764e-05,
      "loss": 0.4647,
      "step": 21360
    },
    {
      "epoch": 0.4863118130305168,
      "grad_norm": 11.0625,
      "learning_rate": 1.0483930893553782e-05,
      "loss": 0.3202,
      "step": 21370
    },
    {
      "epoch": 0.4865393805611815,
      "grad_norm": 199.0,
      "learning_rate": 1.04792866431358e-05,
      "loss": 1.0421,
      "step": 21380
    },
    {
      "epoch": 0.48676694809184623,
      "grad_norm": 115.5,
      "learning_rate": 1.0474642392717815e-05,
      "loss": 0.2598,
      "step": 21390
    },
    {
      "epoch": 0.486994515622511,
      "grad_norm": 0.23828125,
      "learning_rate": 1.0469998142299834e-05,
      "loss": 0.2841,
      "step": 21400
    },
    {
      "epoch": 0.4872220831531757,
      "grad_norm": 118.0,
      "learning_rate": 1.046535389188185e-05,
      "loss": 0.2243,
      "step": 21410
    },
    {
      "epoch": 0.48744965068384044,
      "grad_norm": 240.0,
      "learning_rate": 1.046070964146387e-05,
      "loss": 1.147,
      "step": 21420
    },
    {
      "epoch": 0.48767721821450516,
      "grad_norm": 88.0,
      "learning_rate": 1.0456065391045886e-05,
      "loss": 0.5047,
      "step": 21430
    },
    {
      "epoch": 0.4879047857451699,
      "grad_norm": 193.0,
      "learning_rate": 1.0451421140627903e-05,
      "loss": 1.4793,
      "step": 21440
    },
    {
      "epoch": 0.4881323532758346,
      "grad_norm": 0.236328125,
      "learning_rate": 1.0446776890209922e-05,
      "loss": 0.3805,
      "step": 21450
    },
    {
      "epoch": 0.4883599208064993,
      "grad_norm": 0.002716064453125,
      "learning_rate": 1.0442132639791938e-05,
      "loss": 0.4939,
      "step": 21460
    },
    {
      "epoch": 0.48858748833716403,
      "grad_norm": 0.02001953125,
      "learning_rate": 1.0437488389373955e-05,
      "loss": 0.6053,
      "step": 21470
    },
    {
      "epoch": 0.48881505586782875,
      "grad_norm": 0.84375,
      "learning_rate": 1.0432844138955973e-05,
      "loss": 0.8262,
      "step": 21480
    },
    {
      "epoch": 0.4890426233984935,
      "grad_norm": 91.0,
      "learning_rate": 1.042819988853799e-05,
      "loss": 0.8544,
      "step": 21490
    },
    {
      "epoch": 0.48927019092915824,
      "grad_norm": 132.0,
      "learning_rate": 1.042355563812001e-05,
      "loss": 0.6098,
      "step": 21500
    },
    {
      "epoch": 0.48949775845982296,
      "grad_norm": 308.0,
      "learning_rate": 1.0418911387702026e-05,
      "loss": 1.414,
      "step": 21510
    },
    {
      "epoch": 0.4897253259904877,
      "grad_norm": 216.0,
      "learning_rate": 1.0414267137284045e-05,
      "loss": 0.759,
      "step": 21520
    },
    {
      "epoch": 0.4899528935211524,
      "grad_norm": 370.0,
      "learning_rate": 1.040962288686606e-05,
      "loss": 0.8881,
      "step": 21530
    },
    {
      "epoch": 0.4901804610518171,
      "grad_norm": 187.0,
      "learning_rate": 1.0404978636448077e-05,
      "loss": 0.6893,
      "step": 21540
    },
    {
      "epoch": 0.49040802858248184,
      "grad_norm": 0.6875,
      "learning_rate": 1.0400334386030096e-05,
      "loss": 0.6515,
      "step": 21550
    },
    {
      "epoch": 0.49063559611314655,
      "grad_norm": 51.0,
      "learning_rate": 1.0395690135612113e-05,
      "loss": 1.3642,
      "step": 21560
    },
    {
      "epoch": 0.4908631636438113,
      "grad_norm": 0.0009002685546875,
      "learning_rate": 1.039104588519413e-05,
      "loss": 0.1521,
      "step": 21570
    },
    {
      "epoch": 0.49109073117447605,
      "grad_norm": 244.0,
      "learning_rate": 1.0386401634776149e-05,
      "loss": 1.5877,
      "step": 21580
    },
    {
      "epoch": 0.49131829870514077,
      "grad_norm": 184.0,
      "learning_rate": 1.0381757384358166e-05,
      "loss": 0.7904,
      "step": 21590
    },
    {
      "epoch": 0.4915458662358055,
      "grad_norm": 0.1923828125,
      "learning_rate": 1.0377113133940181e-05,
      "loss": 1.0193,
      "step": 21600
    },
    {
      "epoch": 0.4917734337664702,
      "grad_norm": 155.0,
      "learning_rate": 1.03724688835222e-05,
      "loss": 1.4354,
      "step": 21610
    },
    {
      "epoch": 0.4920010012971349,
      "grad_norm": 180.0,
      "learning_rate": 1.0367824633104217e-05,
      "loss": 1.8819,
      "step": 21620
    },
    {
      "epoch": 0.49222856882779964,
      "grad_norm": 47.5,
      "learning_rate": 1.0363180382686236e-05,
      "loss": 0.6474,
      "step": 21630
    },
    {
      "epoch": 0.49245613635846436,
      "grad_norm": 0.0380859375,
      "learning_rate": 1.0358536132268253e-05,
      "loss": 0.4997,
      "step": 21640
    },
    {
      "epoch": 0.4926837038891291,
      "grad_norm": 75.0,
      "learning_rate": 1.0353891881850272e-05,
      "loss": 0.9377,
      "step": 21650
    },
    {
      "epoch": 0.49291127141979385,
      "grad_norm": 286.0,
      "learning_rate": 1.0349247631432289e-05,
      "loss": 0.8414,
      "step": 21660
    },
    {
      "epoch": 0.49313883895045857,
      "grad_norm": 23.25,
      "learning_rate": 1.0344603381014304e-05,
      "loss": 0.5696,
      "step": 21670
    },
    {
      "epoch": 0.4933664064811233,
      "grad_norm": 0.29296875,
      "learning_rate": 1.0339959130596321e-05,
      "loss": 0.6264,
      "step": 21680
    },
    {
      "epoch": 0.493593974011788,
      "grad_norm": 0.01458740234375,
      "learning_rate": 1.033531488017834e-05,
      "loss": 1.57,
      "step": 21690
    },
    {
      "epoch": 0.4938215415424527,
      "grad_norm": 0.796875,
      "learning_rate": 1.0330670629760357e-05,
      "loss": 1.0515,
      "step": 21700
    },
    {
      "epoch": 0.49404910907311744,
      "grad_norm": 0.00439453125,
      "learning_rate": 1.0326026379342376e-05,
      "loss": 0.8273,
      "step": 21710
    },
    {
      "epoch": 0.49427667660378216,
      "grad_norm": 251.0,
      "learning_rate": 1.0321382128924393e-05,
      "loss": 1.6267,
      "step": 21720
    },
    {
      "epoch": 0.4945042441344469,
      "grad_norm": 6.03125,
      "learning_rate": 1.0316737878506412e-05,
      "loss": 0.6707,
      "step": 21730
    },
    {
      "epoch": 0.4947318116651116,
      "grad_norm": 10.1875,
      "learning_rate": 1.0312093628088427e-05,
      "loss": 0.8452,
      "step": 21740
    },
    {
      "epoch": 0.49495937919577637,
      "grad_norm": 228.0,
      "learning_rate": 1.0307449377670444e-05,
      "loss": 0.5863,
      "step": 21750
    },
    {
      "epoch": 0.4951869467264411,
      "grad_norm": 7.28125,
      "learning_rate": 1.0302805127252463e-05,
      "loss": 1.1342,
      "step": 21760
    },
    {
      "epoch": 0.4954145142571058,
      "grad_norm": 0.0014801025390625,
      "learning_rate": 1.029816087683448e-05,
      "loss": 0.5463,
      "step": 21770
    },
    {
      "epoch": 0.4956420817877705,
      "grad_norm": 0.5625,
      "learning_rate": 1.0293516626416497e-05,
      "loss": 1.4651,
      "step": 21780
    },
    {
      "epoch": 0.49586964931843525,
      "grad_norm": 17.875,
      "learning_rate": 1.0288872375998516e-05,
      "loss": 0.5195,
      "step": 21790
    },
    {
      "epoch": 0.49609721684909996,
      "grad_norm": 5.46875,
      "learning_rate": 1.0284228125580533e-05,
      "loss": 0.5849,
      "step": 21800
    },
    {
      "epoch": 0.4963247843797647,
      "grad_norm": 202.0,
      "learning_rate": 1.0279583875162548e-05,
      "loss": 0.4254,
      "step": 21810
    },
    {
      "epoch": 0.4965523519104294,
      "grad_norm": 0.000896453857421875,
      "learning_rate": 1.0274939624744567e-05,
      "loss": 1.0013,
      "step": 21820
    },
    {
      "epoch": 0.4967799194410941,
      "grad_norm": 74.5,
      "learning_rate": 1.0270295374326584e-05,
      "loss": 0.94,
      "step": 21830
    },
    {
      "epoch": 0.4970074869717589,
      "grad_norm": 0.00089263916015625,
      "learning_rate": 1.0265651123908603e-05,
      "loss": 0.7385,
      "step": 21840
    },
    {
      "epoch": 0.4972350545024236,
      "grad_norm": 100.0,
      "learning_rate": 1.026100687349062e-05,
      "loss": 0.3209,
      "step": 21850
    },
    {
      "epoch": 0.49746262203308833,
      "grad_norm": 51.25,
      "learning_rate": 1.0256362623072638e-05,
      "loss": 0.7604,
      "step": 21860
    },
    {
      "epoch": 0.49769018956375305,
      "grad_norm": 142.0,
      "learning_rate": 1.0251718372654655e-05,
      "loss": 0.6284,
      "step": 21870
    },
    {
      "epoch": 0.49791775709441777,
      "grad_norm": 195.0,
      "learning_rate": 1.024707412223667e-05,
      "loss": 0.6945,
      "step": 21880
    },
    {
      "epoch": 0.4981453246250825,
      "grad_norm": 172.0,
      "learning_rate": 1.024242987181869e-05,
      "loss": 1.3041,
      "step": 21890
    },
    {
      "epoch": 0.4983728921557472,
      "grad_norm": 132.0,
      "learning_rate": 1.0237785621400707e-05,
      "loss": 0.7655,
      "step": 21900
    },
    {
      "epoch": 0.4986004596864119,
      "grad_norm": 194.0,
      "learning_rate": 1.0233141370982724e-05,
      "loss": 0.5915,
      "step": 21910
    },
    {
      "epoch": 0.49882802721707664,
      "grad_norm": 80.0,
      "learning_rate": 1.0228497120564742e-05,
      "loss": 0.4883,
      "step": 21920
    },
    {
      "epoch": 0.4990555947477414,
      "grad_norm": 134.0,
      "learning_rate": 1.022385287014676e-05,
      "loss": 0.647,
      "step": 21930
    },
    {
      "epoch": 0.49928316227840613,
      "grad_norm": 194.0,
      "learning_rate": 1.0219208619728778e-05,
      "loss": 1.392,
      "step": 21940
    },
    {
      "epoch": 0.49951072980907085,
      "grad_norm": 72.5,
      "learning_rate": 1.0214564369310794e-05,
      "loss": 0.6214,
      "step": 21950
    },
    {
      "epoch": 0.49973829733973557,
      "grad_norm": 221.0,
      "learning_rate": 1.020992011889281e-05,
      "loss": 1.1992,
      "step": 21960
    },
    {
      "epoch": 0.4999658648704003,
      "grad_norm": 214.0,
      "learning_rate": 1.020527586847483e-05,
      "loss": 0.2643,
      "step": 21970
    },
    {
      "epoch": 0.5001934324010651,
      "grad_norm": 454.0,
      "learning_rate": 1.0200631618056846e-05,
      "loss": 0.5856,
      "step": 21980
    },
    {
      "epoch": 0.5004209999317297,
      "grad_norm": 0.84765625,
      "learning_rate": 1.0195987367638863e-05,
      "loss": 0.2545,
      "step": 21990
    },
    {
      "epoch": 0.5006485674623945,
      "grad_norm": 0.0257568359375,
      "learning_rate": 1.0191343117220882e-05,
      "loss": 0.5556,
      "step": 22000
    },
    {
      "epoch": 0.5008761349930592,
      "grad_norm": 0.0439453125,
      "learning_rate": 1.01866988668029e-05,
      "loss": 0.4605,
      "step": 22010
    },
    {
      "epoch": 0.5011037025237239,
      "grad_norm": 120.0,
      "learning_rate": 1.0182054616384915e-05,
      "loss": 1.0527,
      "step": 22020
    },
    {
      "epoch": 0.5013312700543886,
      "grad_norm": 32.5,
      "learning_rate": 1.0177410365966933e-05,
      "loss": 0.8035,
      "step": 22030
    },
    {
      "epoch": 0.5015588375850534,
      "grad_norm": 0.90234375,
      "learning_rate": 1.017276611554895e-05,
      "loss": 0.3515,
      "step": 22040
    },
    {
      "epoch": 0.501786405115718,
      "grad_norm": 164.0,
      "learning_rate": 1.016812186513097e-05,
      "loss": 0.7802,
      "step": 22050
    },
    {
      "epoch": 0.5020139726463828,
      "grad_norm": 684.0,
      "learning_rate": 1.0163477614712986e-05,
      "loss": 0.5205,
      "step": 22060
    },
    {
      "epoch": 0.5022415401770476,
      "grad_norm": 0.66015625,
      "learning_rate": 1.0158833364295005e-05,
      "loss": 0.99,
      "step": 22070
    },
    {
      "epoch": 0.5024691077077122,
      "grad_norm": 79.5,
      "learning_rate": 1.0154189113877022e-05,
      "loss": 1.0367,
      "step": 22080
    },
    {
      "epoch": 0.502696675238377,
      "grad_norm": 155.0,
      "learning_rate": 1.0149544863459037e-05,
      "loss": 0.6468,
      "step": 22090
    },
    {
      "epoch": 0.5029242427690417,
      "grad_norm": 51.25,
      "learning_rate": 1.0144900613041056e-05,
      "loss": 0.6571,
      "step": 22100
    },
    {
      "epoch": 0.5031518102997065,
      "grad_norm": 0.06201171875,
      "learning_rate": 1.0140256362623073e-05,
      "loss": 0.668,
      "step": 22110
    },
    {
      "epoch": 0.5033793778303711,
      "grad_norm": 155.0,
      "learning_rate": 1.013561211220509e-05,
      "loss": 1.2999,
      "step": 22120
    },
    {
      "epoch": 0.5036069453610359,
      "grad_norm": 151.0,
      "learning_rate": 1.0130967861787109e-05,
      "loss": 0.6087,
      "step": 22130
    },
    {
      "epoch": 0.5038345128917006,
      "grad_norm": 216.0,
      "learning_rate": 1.0126323611369126e-05,
      "loss": 0.5312,
      "step": 22140
    },
    {
      "epoch": 0.5040620804223653,
      "grad_norm": 752.0,
      "learning_rate": 1.0121679360951145e-05,
      "loss": 1.02,
      "step": 22150
    },
    {
      "epoch": 0.5042896479530301,
      "grad_norm": 0.00160980224609375,
      "learning_rate": 1.011703511053316e-05,
      "loss": 0.5102,
      "step": 22160
    },
    {
      "epoch": 0.5045172154836948,
      "grad_norm": 1.3515625,
      "learning_rate": 1.0112390860115177e-05,
      "loss": 1.1732,
      "step": 22170
    },
    {
      "epoch": 0.5047447830143595,
      "grad_norm": 0.07177734375,
      "learning_rate": 1.0107746609697196e-05,
      "loss": 0.329,
      "step": 22180
    },
    {
      "epoch": 0.5049723505450242,
      "grad_norm": 148.0,
      "learning_rate": 1.0103102359279213e-05,
      "loss": 0.4687,
      "step": 22190
    },
    {
      "epoch": 0.505199918075689,
      "grad_norm": 228.0,
      "learning_rate": 1.0098458108861232e-05,
      "loss": 0.5934,
      "step": 22200
    },
    {
      "epoch": 0.5054274856063536,
      "grad_norm": 168.0,
      "learning_rate": 1.0093813858443249e-05,
      "loss": 0.6435,
      "step": 22210
    },
    {
      "epoch": 0.5056550531370184,
      "grad_norm": 396.0,
      "learning_rate": 1.0089169608025266e-05,
      "loss": 0.8552,
      "step": 22220
    },
    {
      "epoch": 0.5058826206676831,
      "grad_norm": 0.058349609375,
      "learning_rate": 1.0084525357607283e-05,
      "loss": 0.6292,
      "step": 22230
    },
    {
      "epoch": 0.5061101881983479,
      "grad_norm": 50.5,
      "learning_rate": 1.00798811071893e-05,
      "loss": 0.5905,
      "step": 22240
    },
    {
      "epoch": 0.5063377557290126,
      "grad_norm": 6.65625,
      "learning_rate": 1.0075236856771317e-05,
      "loss": 0.5234,
      "step": 22250
    },
    {
      "epoch": 0.5065653232596773,
      "grad_norm": 1.2109375,
      "learning_rate": 1.0070592606353336e-05,
      "loss": 0.8258,
      "step": 22260
    },
    {
      "epoch": 0.5067928907903421,
      "grad_norm": 162.0,
      "learning_rate": 1.0065948355935353e-05,
      "loss": 0.2971,
      "step": 22270
    },
    {
      "epoch": 0.5070204583210067,
      "grad_norm": 143.0,
      "learning_rate": 1.0061304105517372e-05,
      "loss": 1.2603,
      "step": 22280
    },
    {
      "epoch": 0.5072480258516715,
      "grad_norm": 6.09375,
      "learning_rate": 1.0056659855099389e-05,
      "loss": 0.897,
      "step": 22290
    },
    {
      "epoch": 0.5074755933823362,
      "grad_norm": 212.0,
      "learning_rate": 1.0052015604681404e-05,
      "loss": 0.9263,
      "step": 22300
    },
    {
      "epoch": 0.5077031609130009,
      "grad_norm": 0.000972747802734375,
      "learning_rate": 1.0047371354263423e-05,
      "loss": 0.3934,
      "step": 22310
    },
    {
      "epoch": 0.5079307284436657,
      "grad_norm": 2.828125,
      "learning_rate": 1.004272710384544e-05,
      "loss": 0.2122,
      "step": 22320
    },
    {
      "epoch": 0.5081582959743304,
      "grad_norm": 0.0010986328125,
      "learning_rate": 1.0038082853427457e-05,
      "loss": 1.3057,
      "step": 22330
    },
    {
      "epoch": 0.5083858635049952,
      "grad_norm": 18.75,
      "learning_rate": 1.0033438603009476e-05,
      "loss": 1.0187,
      "step": 22340
    },
    {
      "epoch": 0.5086134310356598,
      "grad_norm": 213.0,
      "learning_rate": 1.0028794352591493e-05,
      "loss": 1.6953,
      "step": 22350
    },
    {
      "epoch": 0.5088409985663246,
      "grad_norm": 6.5625,
      "learning_rate": 1.0024150102173511e-05,
      "loss": 1.0856,
      "step": 22360
    },
    {
      "epoch": 0.5090685660969893,
      "grad_norm": 79.5,
      "learning_rate": 1.0019505851755527e-05,
      "loss": 1.5442,
      "step": 22370
    },
    {
      "epoch": 0.509296133627654,
      "grad_norm": 207.0,
      "learning_rate": 1.0014861601337544e-05,
      "loss": 0.6377,
      "step": 22380
    },
    {
      "epoch": 0.5095237011583187,
      "grad_norm": 0.000926971435546875,
      "learning_rate": 1.0010217350919563e-05,
      "loss": 0.8089,
      "step": 22390
    },
    {
      "epoch": 0.5097512686889835,
      "grad_norm": 0.0169677734375,
      "learning_rate": 1.000557310050158e-05,
      "loss": 1.1553,
      "step": 22400
    },
    {
      "epoch": 0.5099788362196482,
      "grad_norm": 164.0,
      "learning_rate": 1.0000928850083598e-05,
      "loss": 1.5977,
      "step": 22410
    },
    {
      "epoch": 0.5102064037503129,
      "grad_norm": 66.0,
      "learning_rate": 9.996284599665614e-06,
      "loss": 0.9482,
      "step": 22420
    },
    {
      "epoch": 0.5104339712809777,
      "grad_norm": 247.0,
      "learning_rate": 9.991640349247633e-06,
      "loss": 0.9757,
      "step": 22430
    },
    {
      "epoch": 0.5106615388116423,
      "grad_norm": 78.0,
      "learning_rate": 9.98699609882965e-06,
      "loss": 0.4563,
      "step": 22440
    },
    {
      "epoch": 0.5108891063423071,
      "grad_norm": 185.0,
      "learning_rate": 9.982351848411668e-06,
      "loss": 1.4184,
      "step": 22450
    },
    {
      "epoch": 0.5111166738729718,
      "grad_norm": 129.0,
      "learning_rate": 9.977707597993684e-06,
      "loss": 0.8275,
      "step": 22460
    },
    {
      "epoch": 0.5113442414036365,
      "grad_norm": 0.0625,
      "learning_rate": 9.973063347575702e-06,
      "loss": 0.592,
      "step": 22470
    },
    {
      "epoch": 0.5115718089343012,
      "grad_norm": 218.0,
      "learning_rate": 9.96841909715772e-06,
      "loss": 0.7545,
      "step": 22480
    },
    {
      "epoch": 0.511799376464966,
      "grad_norm": 92.5,
      "learning_rate": 9.963774846739737e-06,
      "loss": 0.3991,
      "step": 22490
    },
    {
      "epoch": 0.5120269439956308,
      "grad_norm": 81.5,
      "learning_rate": 9.959130596321754e-06,
      "loss": 0.9894,
      "step": 22500
    },
    {
      "epoch": 0.5122545115262954,
      "grad_norm": 394.0,
      "learning_rate": 9.954486345903772e-06,
      "loss": 1.798,
      "step": 22510
    },
    {
      "epoch": 0.5124820790569602,
      "grad_norm": 262.0,
      "learning_rate": 9.94984209548579e-06,
      "loss": 0.5383,
      "step": 22520
    },
    {
      "epoch": 0.5127096465876249,
      "grad_norm": 57.75,
      "learning_rate": 9.945197845067806e-06,
      "loss": 1.9652,
      "step": 22530
    },
    {
      "epoch": 0.5129372141182896,
      "grad_norm": 0.04541015625,
      "learning_rate": 9.940553594649825e-06,
      "loss": 0.8438,
      "step": 22540
    },
    {
      "epoch": 0.5131647816489543,
      "grad_norm": 14.4375,
      "learning_rate": 9.935909344231842e-06,
      "loss": 0.7467,
      "step": 22550
    },
    {
      "epoch": 0.5133923491796191,
      "grad_norm": 0.10791015625,
      "learning_rate": 9.93126509381386e-06,
      "loss": 0.8841,
      "step": 22560
    },
    {
      "epoch": 0.5136199167102837,
      "grad_norm": 0.392578125,
      "learning_rate": 9.926620843395876e-06,
      "loss": 0.8254,
      "step": 22570
    },
    {
      "epoch": 0.5138474842409485,
      "grad_norm": 72.0,
      "learning_rate": 9.921976592977895e-06,
      "loss": 0.3011,
      "step": 22580
    },
    {
      "epoch": 0.5140750517716133,
      "grad_norm": 20.0,
      "learning_rate": 9.917332342559912e-06,
      "loss": 0.7301,
      "step": 22590
    },
    {
      "epoch": 0.5143026193022779,
      "grad_norm": 71.0,
      "learning_rate": 9.91268809214193e-06,
      "loss": 0.5231,
      "step": 22600
    },
    {
      "epoch": 0.5145301868329427,
      "grad_norm": 183.0,
      "learning_rate": 9.908043841723946e-06,
      "loss": 0.7222,
      "step": 22610
    },
    {
      "epoch": 0.5147577543636074,
      "grad_norm": 190.0,
      "learning_rate": 9.903399591305965e-06,
      "loss": 0.7516,
      "step": 22620
    },
    {
      "epoch": 0.5149853218942722,
      "grad_norm": 6.84375,
      "learning_rate": 9.89875534088798e-06,
      "loss": 0.6681,
      "step": 22630
    },
    {
      "epoch": 0.5152128894249368,
      "grad_norm": 181.0,
      "learning_rate": 9.89411109047e-06,
      "loss": 0.718,
      "step": 22640
    },
    {
      "epoch": 0.5154404569556016,
      "grad_norm": 54.25,
      "learning_rate": 9.889466840052016e-06,
      "loss": 0.9276,
      "step": 22650
    },
    {
      "epoch": 0.5156680244862663,
      "grad_norm": 0.0091552734375,
      "learning_rate": 9.884822589634035e-06,
      "loss": 1.1055,
      "step": 22660
    },
    {
      "epoch": 0.515895592016931,
      "grad_norm": 147.0,
      "learning_rate": 9.88017833921605e-06,
      "loss": 1.2565,
      "step": 22670
    },
    {
      "epoch": 0.5161231595475958,
      "grad_norm": 0.004180908203125,
      "learning_rate": 9.875534088798069e-06,
      "loss": 0.4571,
      "step": 22680
    },
    {
      "epoch": 0.5163507270782605,
      "grad_norm": 156.0,
      "learning_rate": 9.870889838380086e-06,
      "loss": 0.8516,
      "step": 22690
    },
    {
      "epoch": 0.5165782946089252,
      "grad_norm": 41.5,
      "learning_rate": 9.866245587962103e-06,
      "loss": 0.6014,
      "step": 22700
    },
    {
      "epoch": 0.5168058621395899,
      "grad_norm": 1.484375,
      "learning_rate": 9.86160133754412e-06,
      "loss": 0.496,
      "step": 22710
    },
    {
      "epoch": 0.5170334296702547,
      "grad_norm": 0.00860595703125,
      "learning_rate": 9.856957087126139e-06,
      "loss": 0.739,
      "step": 22720
    },
    {
      "epoch": 0.5172609972009193,
      "grad_norm": 0.00127410888671875,
      "learning_rate": 9.852312836708156e-06,
      "loss": 0.8418,
      "step": 22730
    },
    {
      "epoch": 0.5174885647315841,
      "grad_norm": 147.0,
      "learning_rate": 9.847668586290173e-06,
      "loss": 0.8962,
      "step": 22740
    },
    {
      "epoch": 0.5177161322622488,
      "grad_norm": 1.0546875,
      "learning_rate": 9.843024335872192e-06,
      "loss": 0.1405,
      "step": 22750
    },
    {
      "epoch": 0.5179436997929135,
      "grad_norm": 426.0,
      "learning_rate": 9.838380085454209e-06,
      "loss": 1.208,
      "step": 22760
    },
    {
      "epoch": 0.5181712673235783,
      "grad_norm": 45.0,
      "learning_rate": 9.833735835036226e-06,
      "loss": 0.9229,
      "step": 22770
    },
    {
      "epoch": 0.518398834854243,
      "grad_norm": 0.001007080078125,
      "learning_rate": 9.829091584618243e-06,
      "loss": 1.0013,
      "step": 22780
    },
    {
      "epoch": 0.5186264023849078,
      "grad_norm": 628.0,
      "learning_rate": 9.824447334200262e-06,
      "loss": 0.3145,
      "step": 22790
    },
    {
      "epoch": 0.5188539699155724,
      "grad_norm": 12.375,
      "learning_rate": 9.819803083782279e-06,
      "loss": 0.3686,
      "step": 22800
    },
    {
      "epoch": 0.5190815374462372,
      "grad_norm": 0.01300048828125,
      "learning_rate": 9.815158833364296e-06,
      "loss": 0.7322,
      "step": 22810
    },
    {
      "epoch": 0.5193091049769019,
      "grad_norm": 260.0,
      "learning_rate": 9.810514582946313e-06,
      "loss": 0.8229,
      "step": 22820
    },
    {
      "epoch": 0.5195366725075666,
      "grad_norm": 378.0,
      "learning_rate": 9.805870332528332e-06,
      "loss": 0.7349,
      "step": 22830
    },
    {
      "epoch": 0.5197642400382313,
      "grad_norm": 0.05224609375,
      "learning_rate": 9.801226082110347e-06,
      "loss": 1.2588,
      "step": 22840
    },
    {
      "epoch": 0.5199918075688961,
      "grad_norm": 146.0,
      "learning_rate": 9.796581831692366e-06,
      "loss": 0.4962,
      "step": 22850
    },
    {
      "epoch": 0.5202193750995608,
      "grad_norm": 10.3125,
      "learning_rate": 9.791937581274383e-06,
      "loss": 0.7033,
      "step": 22860
    },
    {
      "epoch": 0.5204469426302255,
      "grad_norm": 64.5,
      "learning_rate": 9.787293330856402e-06,
      "loss": 1.0251,
      "step": 22870
    },
    {
      "epoch": 0.5206745101608903,
      "grad_norm": 80.0,
      "learning_rate": 9.782649080438417e-06,
      "loss": 0.5777,
      "step": 22880
    },
    {
      "epoch": 0.5209020776915549,
      "grad_norm": 74.5,
      "learning_rate": 9.778004830020436e-06,
      "loss": 0.6878,
      "step": 22890
    },
    {
      "epoch": 0.5211296452222197,
      "grad_norm": 0.0014190673828125,
      "learning_rate": 9.773360579602453e-06,
      "loss": 0.7369,
      "step": 22900
    },
    {
      "epoch": 0.5213572127528844,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.76871632918447e-06,
      "loss": 1.2096,
      "step": 22910
    },
    {
      "epoch": 0.5215847802835492,
      "grad_norm": 0.0302734375,
      "learning_rate": 9.764072078766489e-06,
      "loss": 0.7396,
      "step": 22920
    },
    {
      "epoch": 0.5218123478142138,
      "grad_norm": 56.0,
      "learning_rate": 9.759427828348506e-06,
      "loss": 0.5816,
      "step": 22930
    },
    {
      "epoch": 0.5220399153448786,
      "grad_norm": 0.000885009765625,
      "learning_rate": 9.754783577930523e-06,
      "loss": 0.9288,
      "step": 22940
    },
    {
      "epoch": 0.5222674828755434,
      "grad_norm": 136.0,
      "learning_rate": 9.75013932751254e-06,
      "loss": 0.9478,
      "step": 22950
    },
    {
      "epoch": 0.522495050406208,
      "grad_norm": 0.010986328125,
      "learning_rate": 9.745495077094558e-06,
      "loss": 0.9584,
      "step": 22960
    },
    {
      "epoch": 0.5227226179368728,
      "grad_norm": 133.0,
      "learning_rate": 9.740850826676576e-06,
      "loss": 0.3149,
      "step": 22970
    },
    {
      "epoch": 0.5229501854675375,
      "grad_norm": 154.0,
      "learning_rate": 9.736206576258593e-06,
      "loss": 1.3308,
      "step": 22980
    },
    {
      "epoch": 0.5231777529982022,
      "grad_norm": 4.34375,
      "learning_rate": 9.73156232584061e-06,
      "loss": 0.261,
      "step": 22990
    },
    {
      "epoch": 0.5234053205288669,
      "grad_norm": 68.0,
      "learning_rate": 9.726918075422628e-06,
      "loss": 0.7063,
      "step": 23000
    },
    {
      "epoch": 0.5236328880595317,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.722273825004645e-06,
      "loss": 1.5443,
      "step": 23010
    },
    {
      "epoch": 0.5238604555901963,
      "grad_norm": 520.0,
      "learning_rate": 9.717629574586662e-06,
      "loss": 1.2454,
      "step": 23020
    },
    {
      "epoch": 0.5240880231208611,
      "grad_norm": 34.5,
      "learning_rate": 9.71298532416868e-06,
      "loss": 0.4762,
      "step": 23030
    },
    {
      "epoch": 0.5243155906515259,
      "grad_norm": 80.0,
      "learning_rate": 9.708341073750698e-06,
      "loss": 1.1557,
      "step": 23040
    },
    {
      "epoch": 0.5245431581821905,
      "grad_norm": 652.0,
      "learning_rate": 9.703696823332714e-06,
      "loss": 0.8935,
      "step": 23050
    },
    {
      "epoch": 0.5247707257128553,
      "grad_norm": 109.0,
      "learning_rate": 9.699052572914732e-06,
      "loss": 1.2117,
      "step": 23060
    },
    {
      "epoch": 0.52499829324352,
      "grad_norm": 0.006683349609375,
      "learning_rate": 9.69440832249675e-06,
      "loss": 1.2226,
      "step": 23070
    },
    {
      "epoch": 0.5252258607741848,
      "grad_norm": 100.0,
      "learning_rate": 9.689764072078768e-06,
      "loss": 0.8437,
      "step": 23080
    },
    {
      "epoch": 0.5254534283048494,
      "grad_norm": 1.6875,
      "learning_rate": 9.685119821660785e-06,
      "loss": 0.6022,
      "step": 23090
    },
    {
      "epoch": 0.5256809958355142,
      "grad_norm": 0.00121307373046875,
      "learning_rate": 9.680475571242802e-06,
      "loss": 0.5025,
      "step": 23100
    },
    {
      "epoch": 0.5259085633661789,
      "grad_norm": 43.5,
      "learning_rate": 9.67583132082482e-06,
      "loss": 0.7332,
      "step": 23110
    },
    {
      "epoch": 0.5261361308968436,
      "grad_norm": 0.7421875,
      "learning_rate": 9.671187070406836e-06,
      "loss": 0.5716,
      "step": 23120
    },
    {
      "epoch": 0.5263636984275084,
      "grad_norm": 0.0008697509765625,
      "learning_rate": 9.666542819988855e-06,
      "loss": 0.5232,
      "step": 23130
    },
    {
      "epoch": 0.5265912659581731,
      "grad_norm": 152.0,
      "learning_rate": 9.661898569570872e-06,
      "loss": 1.044,
      "step": 23140
    },
    {
      "epoch": 0.5268188334888378,
      "grad_norm": 81.0,
      "learning_rate": 9.65725431915289e-06,
      "loss": 0.2003,
      "step": 23150
    },
    {
      "epoch": 0.5270464010195025,
      "grad_norm": 194.0,
      "learning_rate": 9.652610068734906e-06,
      "loss": 0.9206,
      "step": 23160
    },
    {
      "epoch": 0.5272739685501673,
      "grad_norm": 0.0027008056640625,
      "learning_rate": 9.647965818316925e-06,
      "loss": 0.9257,
      "step": 23170
    },
    {
      "epoch": 0.527501536080832,
      "grad_norm": 226.0,
      "learning_rate": 9.643321567898942e-06,
      "loss": 0.767,
      "step": 23180
    },
    {
      "epoch": 0.5277291036114967,
      "grad_norm": 151.0,
      "learning_rate": 9.63867731748096e-06,
      "loss": 0.8413,
      "step": 23190
    },
    {
      "epoch": 0.5279566711421614,
      "grad_norm": 266.0,
      "learning_rate": 9.634033067062976e-06,
      "loss": 0.6634,
      "step": 23200
    },
    {
      "epoch": 0.5281842386728262,
      "grad_norm": 17.375,
      "learning_rate": 9.629388816644995e-06,
      "loss": 0.6534,
      "step": 23210
    },
    {
      "epoch": 0.5284118062034909,
      "grad_norm": 112.0,
      "learning_rate": 9.624744566227012e-06,
      "loss": 0.6706,
      "step": 23220
    },
    {
      "epoch": 0.5286393737341556,
      "grad_norm": 744.0,
      "learning_rate": 9.620100315809029e-06,
      "loss": 0.856,
      "step": 23230
    },
    {
      "epoch": 0.5288669412648204,
      "grad_norm": 229.0,
      "learning_rate": 9.615456065391046e-06,
      "loss": 1.1824,
      "step": 23240
    },
    {
      "epoch": 0.529094508795485,
      "grad_norm": 238.0,
      "learning_rate": 9.610811814973065e-06,
      "loss": 0.8667,
      "step": 23250
    },
    {
      "epoch": 0.5293220763261498,
      "grad_norm": 278.0,
      "learning_rate": 9.606167564555082e-06,
      "loss": 0.7862,
      "step": 23260
    },
    {
      "epoch": 0.5295496438568145,
      "grad_norm": 0.00080108642578125,
      "learning_rate": 9.601523314137099e-06,
      "loss": 0.2652,
      "step": 23270
    },
    {
      "epoch": 0.5297772113874792,
      "grad_norm": 0.000629425048828125,
      "learning_rate": 9.596879063719116e-06,
      "loss": 1.2018,
      "step": 23280
    },
    {
      "epoch": 0.5300047789181439,
      "grad_norm": 180.0,
      "learning_rate": 9.592234813301135e-06,
      "loss": 2.109,
      "step": 23290
    },
    {
      "epoch": 0.5302323464488087,
      "grad_norm": 147.0,
      "learning_rate": 9.587590562883152e-06,
      "loss": 0.4448,
      "step": 23300
    },
    {
      "epoch": 0.5304599139794735,
      "grad_norm": 0.01239013671875,
      "learning_rate": 9.582946312465169e-06,
      "loss": 0.5257,
      "step": 23310
    },
    {
      "epoch": 0.5306874815101381,
      "grad_norm": 362.0,
      "learning_rate": 9.578302062047186e-06,
      "loss": 0.7007,
      "step": 23320
    },
    {
      "epoch": 0.5309150490408029,
      "grad_norm": 201.0,
      "learning_rate": 9.573657811629203e-06,
      "loss": 0.9443,
      "step": 23330
    },
    {
      "epoch": 0.5311426165714676,
      "grad_norm": 0.007476806640625,
      "learning_rate": 9.569013561211222e-06,
      "loss": 0.6834,
      "step": 23340
    },
    {
      "epoch": 0.5313701841021323,
      "grad_norm": 98.5,
      "learning_rate": 9.564369310793239e-06,
      "loss": 0.6474,
      "step": 23350
    },
    {
      "epoch": 0.531597751632797,
      "grad_norm": 3.546875,
      "learning_rate": 9.559725060375256e-06,
      "loss": 0.2305,
      "step": 23360
    },
    {
      "epoch": 0.5318253191634618,
      "grad_norm": 103.5,
      "learning_rate": 9.555080809957273e-06,
      "loss": 0.8612,
      "step": 23370
    },
    {
      "epoch": 0.5320528866941265,
      "grad_norm": 0.000835418701171875,
      "learning_rate": 9.550436559539292e-06,
      "loss": 0.4043,
      "step": 23380
    },
    {
      "epoch": 0.5322804542247912,
      "grad_norm": 0.00091552734375,
      "learning_rate": 9.545792309121309e-06,
      "loss": 0.8981,
      "step": 23390
    },
    {
      "epoch": 0.532508021755456,
      "grad_norm": 181.0,
      "learning_rate": 9.541148058703326e-06,
      "loss": 0.7449,
      "step": 23400
    },
    {
      "epoch": 0.5327355892861206,
      "grad_norm": 0.037841796875,
      "learning_rate": 9.536503808285343e-06,
      "loss": 0.8689,
      "step": 23410
    },
    {
      "epoch": 0.5329631568167854,
      "grad_norm": 128.0,
      "learning_rate": 9.531859557867362e-06,
      "loss": 0.5592,
      "step": 23420
    },
    {
      "epoch": 0.5331907243474501,
      "grad_norm": 86.5,
      "learning_rate": 9.527215307449379e-06,
      "loss": 0.5846,
      "step": 23430
    },
    {
      "epoch": 0.5334182918781148,
      "grad_norm": 0.0003910064697265625,
      "learning_rate": 9.522571057031396e-06,
      "loss": 0.344,
      "step": 23440
    },
    {
      "epoch": 0.5336458594087795,
      "grad_norm": 31.875,
      "learning_rate": 9.517926806613413e-06,
      "loss": 0.7053,
      "step": 23450
    },
    {
      "epoch": 0.5338734269394443,
      "grad_norm": 130.0,
      "learning_rate": 9.513282556195432e-06,
      "loss": 0.9396,
      "step": 23460
    },
    {
      "epoch": 0.5341009944701091,
      "grad_norm": 0.0380859375,
      "learning_rate": 9.508638305777449e-06,
      "loss": 0.8451,
      "step": 23470
    },
    {
      "epoch": 0.5343285620007737,
      "grad_norm": 362.0,
      "learning_rate": 9.503994055359466e-06,
      "loss": 0.5204,
      "step": 23480
    },
    {
      "epoch": 0.5345561295314385,
      "grad_norm": 352.0,
      "learning_rate": 9.499349804941483e-06,
      "loss": 1.3376,
      "step": 23490
    },
    {
      "epoch": 0.5347836970621032,
      "grad_norm": 65.0,
      "learning_rate": 9.494705554523501e-06,
      "loss": 0.7356,
      "step": 23500
    },
    {
      "epoch": 0.5350112645927679,
      "grad_norm": 166.0,
      "learning_rate": 9.490061304105519e-06,
      "loss": 0.5742,
      "step": 23510
    },
    {
      "epoch": 0.5352388321234326,
      "grad_norm": 82.5,
      "learning_rate": 9.485417053687536e-06,
      "loss": 0.5862,
      "step": 23520
    },
    {
      "epoch": 0.5354663996540974,
      "grad_norm": 0.01348876953125,
      "learning_rate": 9.480772803269553e-06,
      "loss": 0.9552,
      "step": 23530
    },
    {
      "epoch": 0.535693967184762,
      "grad_norm": 98.5,
      "learning_rate": 9.47612855285157e-06,
      "loss": 0.7022,
      "step": 23540
    },
    {
      "epoch": 0.5359215347154268,
      "grad_norm": 27.625,
      "learning_rate": 9.471484302433588e-06,
      "loss": 0.6364,
      "step": 23550
    },
    {
      "epoch": 0.5361491022460916,
      "grad_norm": 200.0,
      "learning_rate": 9.466840052015605e-06,
      "loss": 1.2397,
      "step": 23560
    },
    {
      "epoch": 0.5363766697767562,
      "grad_norm": 0.0013427734375,
      "learning_rate": 9.462195801597624e-06,
      "loss": 0.5858,
      "step": 23570
    },
    {
      "epoch": 0.536604237307421,
      "grad_norm": 96.0,
      "learning_rate": 9.45755155117964e-06,
      "loss": 1.6245,
      "step": 23580
    },
    {
      "epoch": 0.5368318048380857,
      "grad_norm": 0.006011962890625,
      "learning_rate": 9.452907300761658e-06,
      "loss": 1.0109,
      "step": 23590
    },
    {
      "epoch": 0.5370593723687505,
      "grad_norm": 38.25,
      "learning_rate": 9.448263050343675e-06,
      "loss": 0.8656,
      "step": 23600
    },
    {
      "epoch": 0.5372869398994151,
      "grad_norm": 0.01031494140625,
      "learning_rate": 9.443618799925692e-06,
      "loss": 0.311,
      "step": 23610
    },
    {
      "epoch": 0.5375145074300799,
      "grad_norm": 494.0,
      "learning_rate": 9.43897454950771e-06,
      "loss": 1.2741,
      "step": 23620
    },
    {
      "epoch": 0.5377420749607446,
      "grad_norm": 0.00714111328125,
      "learning_rate": 9.434330299089728e-06,
      "loss": 0.6309,
      "step": 23630
    },
    {
      "epoch": 0.5379696424914093,
      "grad_norm": 0.1162109375,
      "learning_rate": 9.429686048671745e-06,
      "loss": 0.7897,
      "step": 23640
    },
    {
      "epoch": 0.5381972100220741,
      "grad_norm": 0.056396484375,
      "learning_rate": 9.425041798253762e-06,
      "loss": 1.1082,
      "step": 23650
    },
    {
      "epoch": 0.5384247775527388,
      "grad_norm": 0.083984375,
      "learning_rate": 9.42039754783578e-06,
      "loss": 1.3362,
      "step": 23660
    },
    {
      "epoch": 0.5386523450834035,
      "grad_norm": 206.0,
      "learning_rate": 9.415753297417798e-06,
      "loss": 0.394,
      "step": 23670
    },
    {
      "epoch": 0.5388799126140682,
      "grad_norm": 109.5,
      "learning_rate": 9.411109046999815e-06,
      "loss": 0.9042,
      "step": 23680
    },
    {
      "epoch": 0.539107480144733,
      "grad_norm": 202.0,
      "learning_rate": 9.406464796581832e-06,
      "loss": 0.6342,
      "step": 23690
    },
    {
      "epoch": 0.5393350476753976,
      "grad_norm": 110.0,
      "learning_rate": 9.40182054616385e-06,
      "loss": 0.4418,
      "step": 23700
    },
    {
      "epoch": 0.5395626152060624,
      "grad_norm": 0.03076171875,
      "learning_rate": 9.397176295745868e-06,
      "loss": 0.6573,
      "step": 23710
    },
    {
      "epoch": 0.5397901827367271,
      "grad_norm": 0.00274658203125,
      "learning_rate": 9.392532045327885e-06,
      "loss": 0.6993,
      "step": 23720
    },
    {
      "epoch": 0.5400177502673918,
      "grad_norm": 0.00122833251953125,
      "learning_rate": 9.387887794909902e-06,
      "loss": 0.6165,
      "step": 23730
    },
    {
      "epoch": 0.5402453177980566,
      "grad_norm": 0.005035400390625,
      "learning_rate": 9.383243544491921e-06,
      "loss": 1.447,
      "step": 23740
    },
    {
      "epoch": 0.5404728853287213,
      "grad_norm": 204.0,
      "learning_rate": 9.378599294073936e-06,
      "loss": 0.8468,
      "step": 23750
    },
    {
      "epoch": 0.5407004528593861,
      "grad_norm": 2.15625,
      "learning_rate": 9.373955043655955e-06,
      "loss": 1.2733,
      "step": 23760
    },
    {
      "epoch": 0.5409280203900507,
      "grad_norm": 0.0004673004150390625,
      "learning_rate": 9.369310793237972e-06,
      "loss": 0.5335,
      "step": 23770
    },
    {
      "epoch": 0.5411555879207155,
      "grad_norm": 99.0,
      "learning_rate": 9.364666542819991e-06,
      "loss": 0.7429,
      "step": 23780
    },
    {
      "epoch": 0.5413831554513802,
      "grad_norm": 0.0230712890625,
      "learning_rate": 9.360022292402006e-06,
      "loss": 0.453,
      "step": 23790
    },
    {
      "epoch": 0.5416107229820449,
      "grad_norm": 85.5,
      "learning_rate": 9.355378041984025e-06,
      "loss": 1.5686,
      "step": 23800
    },
    {
      "epoch": 0.5418382905127096,
      "grad_norm": 124.5,
      "learning_rate": 9.350733791566042e-06,
      "loss": 0.4528,
      "step": 23810
    },
    {
      "epoch": 0.5420658580433744,
      "grad_norm": 0.010986328125,
      "learning_rate": 9.346089541148059e-06,
      "loss": 0.7081,
      "step": 23820
    },
    {
      "epoch": 0.5422934255740391,
      "grad_norm": 384.0,
      "learning_rate": 9.341445290730076e-06,
      "loss": 0.9133,
      "step": 23830
    },
    {
      "epoch": 0.5425209931047038,
      "grad_norm": 0.00103759765625,
      "learning_rate": 9.336801040312095e-06,
      "loss": 0.5391,
      "step": 23840
    },
    {
      "epoch": 0.5427485606353686,
      "grad_norm": 195.0,
      "learning_rate": 9.332156789894112e-06,
      "loss": 0.6821,
      "step": 23850
    },
    {
      "epoch": 0.5429761281660332,
      "grad_norm": 120.0,
      "learning_rate": 9.327512539476129e-06,
      "loss": 0.2072,
      "step": 23860
    },
    {
      "epoch": 0.543203695696698,
      "grad_norm": 0.0029754638671875,
      "learning_rate": 9.322868289058146e-06,
      "loss": 1.5059,
      "step": 23870
    },
    {
      "epoch": 0.5434312632273627,
      "grad_norm": 0.546875,
      "learning_rate": 9.318224038640165e-06,
      "loss": 1.3615,
      "step": 23880
    },
    {
      "epoch": 0.5436588307580275,
      "grad_norm": 146.0,
      "learning_rate": 9.313579788222182e-06,
      "loss": 1.215,
      "step": 23890
    },
    {
      "epoch": 0.5438863982886921,
      "grad_norm": 59.75,
      "learning_rate": 9.308935537804199e-06,
      "loss": 0.5825,
      "step": 23900
    },
    {
      "epoch": 0.5441139658193569,
      "grad_norm": 46.25,
      "learning_rate": 9.304291287386216e-06,
      "loss": 0.9045,
      "step": 23910
    },
    {
      "epoch": 0.5443415333500217,
      "grad_norm": 204.0,
      "learning_rate": 9.299647036968235e-06,
      "loss": 0.4385,
      "step": 23920
    },
    {
      "epoch": 0.5445691008806863,
      "grad_norm": 167.0,
      "learning_rate": 9.295002786550252e-06,
      "loss": 0.6527,
      "step": 23930
    },
    {
      "epoch": 0.5447966684113511,
      "grad_norm": 7.03125,
      "learning_rate": 9.290358536132269e-06,
      "loss": 0.3641,
      "step": 23940
    },
    {
      "epoch": 0.5450242359420158,
      "grad_norm": 104.5,
      "learning_rate": 9.285714285714288e-06,
      "loss": 0.8834,
      "step": 23950
    },
    {
      "epoch": 0.5452518034726805,
      "grad_norm": 161.0,
      "learning_rate": 9.281070035296303e-06,
      "loss": 1.2778,
      "step": 23960
    },
    {
      "epoch": 0.5454793710033452,
      "grad_norm": 116.5,
      "learning_rate": 9.276425784878322e-06,
      "loss": 0.8246,
      "step": 23970
    },
    {
      "epoch": 0.54570693853401,
      "grad_norm": 132.0,
      "learning_rate": 9.271781534460339e-06,
      "loss": 0.9897,
      "step": 23980
    },
    {
      "epoch": 0.5459345060646746,
      "grad_norm": 2.671875,
      "learning_rate": 9.267137284042357e-06,
      "loss": 0.9666,
      "step": 23990
    },
    {
      "epoch": 0.5461620735953394,
      "grad_norm": 176.0,
      "learning_rate": 9.262493033624373e-06,
      "loss": 1.272,
      "step": 24000
    },
    {
      "epoch": 0.5463896411260042,
      "grad_norm": 0.06640625,
      "learning_rate": 9.257848783206392e-06,
      "loss": 0.5377,
      "step": 24010
    },
    {
      "epoch": 0.5466172086566689,
      "grad_norm": 122.5,
      "learning_rate": 9.253204532788409e-06,
      "loss": 0.5626,
      "step": 24020
    },
    {
      "epoch": 0.5468447761873336,
      "grad_norm": 0.2021484375,
      "learning_rate": 9.248560282370426e-06,
      "loss": 0.5244,
      "step": 24030
    },
    {
      "epoch": 0.5470723437179983,
      "grad_norm": 13.375,
      "learning_rate": 9.243916031952443e-06,
      "loss": 0.4959,
      "step": 24040
    },
    {
      "epoch": 0.5472999112486631,
      "grad_norm": 125.0,
      "learning_rate": 9.239271781534462e-06,
      "loss": 0.9633,
      "step": 24050
    },
    {
      "epoch": 0.5475274787793277,
      "grad_norm": 0.0024871826171875,
      "learning_rate": 9.234627531116479e-06,
      "loss": 0.7292,
      "step": 24060
    },
    {
      "epoch": 0.5477550463099925,
      "grad_norm": 87.0,
      "learning_rate": 9.229983280698496e-06,
      "loss": 0.8352,
      "step": 24070
    },
    {
      "epoch": 0.5479826138406572,
      "grad_norm": 123.5,
      "learning_rate": 9.225339030280513e-06,
      "loss": 0.7272,
      "step": 24080
    },
    {
      "epoch": 0.5482101813713219,
      "grad_norm": 64.0,
      "learning_rate": 9.220694779862531e-06,
      "loss": 0.6276,
      "step": 24090
    },
    {
      "epoch": 0.5484377489019867,
      "grad_norm": 59.0,
      "learning_rate": 9.216050529444548e-06,
      "loss": 1.9153,
      "step": 24100
    },
    {
      "epoch": 0.5486653164326514,
      "grad_norm": 0.0004825592041015625,
      "learning_rate": 9.211406279026566e-06,
      "loss": 0.6773,
      "step": 24110
    },
    {
      "epoch": 0.5488928839633161,
      "grad_norm": 0.00151824951171875,
      "learning_rate": 9.206762028608584e-06,
      "loss": 1.382,
      "step": 24120
    },
    {
      "epoch": 0.5491204514939808,
      "grad_norm": 0.0101318359375,
      "learning_rate": 9.202117778190601e-06,
      "loss": 0.5747,
      "step": 24130
    },
    {
      "epoch": 0.5493480190246456,
      "grad_norm": 0.11083984375,
      "learning_rate": 9.197473527772618e-06,
      "loss": 0.4195,
      "step": 24140
    },
    {
      "epoch": 0.5495755865553102,
      "grad_norm": 171.0,
      "learning_rate": 9.192829277354635e-06,
      "loss": 0.8162,
      "step": 24150
    },
    {
      "epoch": 0.549803154085975,
      "grad_norm": 0.004058837890625,
      "learning_rate": 9.188185026936654e-06,
      "loss": 1.0668,
      "step": 24160
    },
    {
      "epoch": 0.5500307216166397,
      "grad_norm": 2.203125,
      "learning_rate": 9.18354077651867e-06,
      "loss": 0.5932,
      "step": 24170
    },
    {
      "epoch": 0.5502582891473045,
      "grad_norm": 16.125,
      "learning_rate": 9.178896526100688e-06,
      "loss": 1.4161,
      "step": 24180
    },
    {
      "epoch": 0.5504858566779692,
      "grad_norm": 234.0,
      "learning_rate": 9.174252275682705e-06,
      "loss": 0.7773,
      "step": 24190
    },
    {
      "epoch": 0.5507134242086339,
      "grad_norm": 75.5,
      "learning_rate": 9.169608025264724e-06,
      "loss": 0.4092,
      "step": 24200
    },
    {
      "epoch": 0.5509409917392987,
      "grad_norm": 0.0810546875,
      "learning_rate": 9.16496377484674e-06,
      "loss": 0.4852,
      "step": 24210
    },
    {
      "epoch": 0.5511685592699633,
      "grad_norm": 1.5,
      "learning_rate": 9.160319524428758e-06,
      "loss": 0.951,
      "step": 24220
    },
    {
      "epoch": 0.5513961268006281,
      "grad_norm": 0.00177764892578125,
      "learning_rate": 9.155675274010775e-06,
      "loss": 0.3903,
      "step": 24230
    },
    {
      "epoch": 0.5516236943312928,
      "grad_norm": 124.5,
      "learning_rate": 9.151031023592792e-06,
      "loss": 1.7245,
      "step": 24240
    },
    {
      "epoch": 0.5518512618619575,
      "grad_norm": 764.0,
      "learning_rate": 9.14638677317481e-06,
      "loss": 0.8228,
      "step": 24250
    },
    {
      "epoch": 0.5520788293926222,
      "grad_norm": 84.5,
      "learning_rate": 9.141742522756828e-06,
      "loss": 0.1098,
      "step": 24260
    },
    {
      "epoch": 0.552306396923287,
      "grad_norm": 149.0,
      "learning_rate": 9.137098272338845e-06,
      "loss": 1.3838,
      "step": 24270
    },
    {
      "epoch": 0.5525339644539518,
      "grad_norm": 120.5,
      "learning_rate": 9.132454021920862e-06,
      "loss": 1.0168,
      "step": 24280
    },
    {
      "epoch": 0.5527615319846164,
      "grad_norm": 0.003326416015625,
      "learning_rate": 9.127809771502881e-06,
      "loss": 1.1998,
      "step": 24290
    },
    {
      "epoch": 0.5529890995152812,
      "grad_norm": 98.5,
      "learning_rate": 9.123165521084898e-06,
      "loss": 0.7283,
      "step": 24300
    },
    {
      "epoch": 0.5532166670459459,
      "grad_norm": 89.5,
      "learning_rate": 9.118521270666915e-06,
      "loss": 1.6546,
      "step": 24310
    },
    {
      "epoch": 0.5534442345766106,
      "grad_norm": 235.0,
      "learning_rate": 9.113877020248932e-06,
      "loss": 0.5088,
      "step": 24320
    },
    {
      "epoch": 0.5536718021072753,
      "grad_norm": 2.15625,
      "learning_rate": 9.109232769830951e-06,
      "loss": 1.9311,
      "step": 24330
    },
    {
      "epoch": 0.5538993696379401,
      "grad_norm": 2.015625,
      "learning_rate": 9.104588519412968e-06,
      "loss": 0.8085,
      "step": 24340
    },
    {
      "epoch": 0.5541269371686048,
      "grad_norm": 92.0,
      "learning_rate": 9.099944268994985e-06,
      "loss": 1.1299,
      "step": 24350
    },
    {
      "epoch": 0.5543545046992695,
      "grad_norm": 97.0,
      "learning_rate": 9.095300018577002e-06,
      "loss": 0.6479,
      "step": 24360
    },
    {
      "epoch": 0.5545820722299343,
      "grad_norm": 101.5,
      "learning_rate": 9.09065576815902e-06,
      "loss": 1.1174,
      "step": 24370
    },
    {
      "epoch": 0.5548096397605989,
      "grad_norm": 121.0,
      "learning_rate": 9.086011517741036e-06,
      "loss": 0.2282,
      "step": 24380
    },
    {
      "epoch": 0.5550372072912637,
      "grad_norm": 48.0,
      "learning_rate": 9.081367267323055e-06,
      "loss": 0.5477,
      "step": 24390
    },
    {
      "epoch": 0.5552647748219284,
      "grad_norm": 0.486328125,
      "learning_rate": 9.076723016905072e-06,
      "loss": 0.9137,
      "step": 24400
    },
    {
      "epoch": 0.5554923423525931,
      "grad_norm": 80.0,
      "learning_rate": 9.07207876648709e-06,
      "loss": 0.8723,
      "step": 24410
    },
    {
      "epoch": 0.5557199098832578,
      "grad_norm": 101.0,
      "learning_rate": 9.067434516069106e-06,
      "loss": 0.4637,
      "step": 24420
    },
    {
      "epoch": 0.5559474774139226,
      "grad_norm": 19.625,
      "learning_rate": 9.062790265651125e-06,
      "loss": 0.2319,
      "step": 24430
    },
    {
      "epoch": 0.5561750449445874,
      "grad_norm": 260.0,
      "learning_rate": 9.058146015233142e-06,
      "loss": 0.3633,
      "step": 24440
    },
    {
      "epoch": 0.556402612475252,
      "grad_norm": 0.17578125,
      "learning_rate": 9.053501764815159e-06,
      "loss": 0.8178,
      "step": 24450
    },
    {
      "epoch": 0.5566301800059168,
      "grad_norm": 85.0,
      "learning_rate": 9.048857514397178e-06,
      "loss": 0.7126,
      "step": 24460
    },
    {
      "epoch": 0.5568577475365815,
      "grad_norm": 77.0,
      "learning_rate": 9.044213263979195e-06,
      "loss": 0.6135,
      "step": 24470
    },
    {
      "epoch": 0.5570853150672462,
      "grad_norm": 544.0,
      "learning_rate": 9.039569013561212e-06,
      "loss": 0.6845,
      "step": 24480
    },
    {
      "epoch": 0.5573128825979109,
      "grad_norm": 76.5,
      "learning_rate": 9.034924763143229e-06,
      "loss": 0.6439,
      "step": 24490
    },
    {
      "epoch": 0.5575404501285757,
      "grad_norm": 0.000858306884765625,
      "learning_rate": 9.030280512725248e-06,
      "loss": 0.845,
      "step": 24500
    },
    {
      "epoch": 0.5577680176592403,
      "grad_norm": 109.5,
      "learning_rate": 9.025636262307265e-06,
      "loss": 0.4849,
      "step": 24510
    },
    {
      "epoch": 0.5579955851899051,
      "grad_norm": 56.25,
      "learning_rate": 9.020992011889282e-06,
      "loss": 0.6319,
      "step": 24520
    },
    {
      "epoch": 0.5582231527205699,
      "grad_norm": 120.0,
      "learning_rate": 9.016347761471299e-06,
      "loss": 0.8995,
      "step": 24530
    },
    {
      "epoch": 0.5584507202512345,
      "grad_norm": 249.0,
      "learning_rate": 9.011703511053318e-06,
      "loss": 0.5883,
      "step": 24540
    },
    {
      "epoch": 0.5586782877818993,
      "grad_norm": 103.0,
      "learning_rate": 9.007059260635335e-06,
      "loss": 0.6253,
      "step": 24550
    },
    {
      "epoch": 0.558905855312564,
      "grad_norm": 0.007720947265625,
      "learning_rate": 9.002415010217352e-06,
      "loss": 0.557,
      "step": 24560
    },
    {
      "epoch": 0.5591334228432288,
      "grad_norm": 0.01806640625,
      "learning_rate": 8.997770759799369e-06,
      "loss": 0.523,
      "step": 24570
    },
    {
      "epoch": 0.5593609903738934,
      "grad_norm": 25.625,
      "learning_rate": 8.993126509381387e-06,
      "loss": 0.5487,
      "step": 24580
    },
    {
      "epoch": 0.5595885579045582,
      "grad_norm": 932.0,
      "learning_rate": 8.988482258963403e-06,
      "loss": 0.729,
      "step": 24590
    },
    {
      "epoch": 0.5598161254352229,
      "grad_norm": 0.00537109375,
      "learning_rate": 8.983838008545422e-06,
      "loss": 0.331,
      "step": 24600
    },
    {
      "epoch": 0.5600436929658876,
      "grad_norm": 255.0,
      "learning_rate": 8.979193758127439e-06,
      "loss": 0.4554,
      "step": 24610
    },
    {
      "epoch": 0.5602712604965524,
      "grad_norm": 27.625,
      "learning_rate": 8.974549507709457e-06,
      "loss": 1.5149,
      "step": 24620
    },
    {
      "epoch": 0.5604988280272171,
      "grad_norm": 224.0,
      "learning_rate": 8.969905257291473e-06,
      "loss": 0.4854,
      "step": 24630
    },
    {
      "epoch": 0.5607263955578818,
      "grad_norm": 0.0023345947265625,
      "learning_rate": 8.965261006873491e-06,
      "loss": 0.4818,
      "step": 24640
    },
    {
      "epoch": 0.5609539630885465,
      "grad_norm": 0.0234375,
      "learning_rate": 8.960616756455509e-06,
      "loss": 1.1701,
      "step": 24650
    },
    {
      "epoch": 0.5611815306192113,
      "grad_norm": 12.4375,
      "learning_rate": 8.955972506037526e-06,
      "loss": 0.7328,
      "step": 24660
    },
    {
      "epoch": 0.5614090981498759,
      "grad_norm": 18.5,
      "learning_rate": 8.951328255619544e-06,
      "loss": 0.5727,
      "step": 24670
    },
    {
      "epoch": 0.5616366656805407,
      "grad_norm": 146.0,
      "learning_rate": 8.946684005201561e-06,
      "loss": 1.06,
      "step": 24680
    },
    {
      "epoch": 0.5618642332112054,
      "grad_norm": 292.0,
      "learning_rate": 8.942039754783578e-06,
      "loss": 0.381,
      "step": 24690
    },
    {
      "epoch": 0.5620918007418702,
      "grad_norm": 50.5,
      "learning_rate": 8.937395504365595e-06,
      "loss": 1.6632,
      "step": 24700
    },
    {
      "epoch": 0.5623193682725349,
      "grad_norm": 4.40625,
      "learning_rate": 8.932751253947614e-06,
      "loss": 1.2516,
      "step": 24710
    },
    {
      "epoch": 0.5625469358031996,
      "grad_norm": 46.0,
      "learning_rate": 8.928107003529631e-06,
      "loss": 0.4654,
      "step": 24720
    },
    {
      "epoch": 0.5627745033338644,
      "grad_norm": 171.0,
      "learning_rate": 8.923462753111648e-06,
      "loss": 0.7275,
      "step": 24730
    },
    {
      "epoch": 0.563002070864529,
      "grad_norm": 11.75,
      "learning_rate": 8.918818502693665e-06,
      "loss": 0.9091,
      "step": 24740
    },
    {
      "epoch": 0.5632296383951938,
      "grad_norm": 15.0,
      "learning_rate": 8.914174252275684e-06,
      "loss": 1.0297,
      "step": 24750
    },
    {
      "epoch": 0.5634572059258585,
      "grad_norm": 51.75,
      "learning_rate": 8.909530001857701e-06,
      "loss": 0.6384,
      "step": 24760
    },
    {
      "epoch": 0.5636847734565232,
      "grad_norm": 68.5,
      "learning_rate": 8.904885751439718e-06,
      "loss": 0.3996,
      "step": 24770
    },
    {
      "epoch": 0.5639123409871879,
      "grad_norm": 113.5,
      "learning_rate": 8.900241501021735e-06,
      "loss": 0.8625,
      "step": 24780
    },
    {
      "epoch": 0.5641399085178527,
      "grad_norm": 396.0,
      "learning_rate": 8.895597250603754e-06,
      "loss": 1.0918,
      "step": 24790
    },
    {
      "epoch": 0.5643674760485174,
      "grad_norm": 137.0,
      "learning_rate": 8.89095300018577e-06,
      "loss": 0.4534,
      "step": 24800
    },
    {
      "epoch": 0.5645950435791821,
      "grad_norm": 256.0,
      "learning_rate": 8.886308749767788e-06,
      "loss": 1.3702,
      "step": 24810
    },
    {
      "epoch": 0.5648226111098469,
      "grad_norm": 1.53125,
      "learning_rate": 8.881664499349805e-06,
      "loss": 0.4997,
      "step": 24820
    },
    {
      "epoch": 0.5650501786405115,
      "grad_norm": 0.003875732421875,
      "learning_rate": 8.877020248931824e-06,
      "loss": 0.616,
      "step": 24830
    },
    {
      "epoch": 0.5652777461711763,
      "grad_norm": 0.11669921875,
      "learning_rate": 8.872375998513841e-06,
      "loss": 0.7854,
      "step": 24840
    },
    {
      "epoch": 0.565505313701841,
      "grad_norm": 112.5,
      "learning_rate": 8.867731748095858e-06,
      "loss": 1.1545,
      "step": 24850
    },
    {
      "epoch": 0.5657328812325058,
      "grad_norm": 106.0,
      "learning_rate": 8.863087497677875e-06,
      "loss": 0.6944,
      "step": 24860
    },
    {
      "epoch": 0.5659604487631704,
      "grad_norm": 112.5,
      "learning_rate": 8.858443247259892e-06,
      "loss": 0.3515,
      "step": 24870
    },
    {
      "epoch": 0.5661880162938352,
      "grad_norm": 0.000766754150390625,
      "learning_rate": 8.853798996841911e-06,
      "loss": 0.9856,
      "step": 24880
    },
    {
      "epoch": 0.5664155838245,
      "grad_norm": 251.0,
      "learning_rate": 8.849154746423928e-06,
      "loss": 0.6841,
      "step": 24890
    },
    {
      "epoch": 0.5666431513551646,
      "grad_norm": 0.0012054443359375,
      "learning_rate": 8.844510496005945e-06,
      "loss": 1.6468,
      "step": 24900
    },
    {
      "epoch": 0.5668707188858294,
      "grad_norm": 0.314453125,
      "learning_rate": 8.839866245587962e-06,
      "loss": 0.5117,
      "step": 24910
    },
    {
      "epoch": 0.5670982864164941,
      "grad_norm": 189.0,
      "learning_rate": 8.835221995169981e-06,
      "loss": 1.0986,
      "step": 24920
    },
    {
      "epoch": 0.5673258539471588,
      "grad_norm": 87.0,
      "learning_rate": 8.830577744751998e-06,
      "loss": 0.4972,
      "step": 24930
    },
    {
      "epoch": 0.5675534214778235,
      "grad_norm": 298.0,
      "learning_rate": 8.825933494334015e-06,
      "loss": 1.4745,
      "step": 24940
    },
    {
      "epoch": 0.5677809890084883,
      "grad_norm": 179.0,
      "learning_rate": 8.821289243916032e-06,
      "loss": 0.4699,
      "step": 24950
    },
    {
      "epoch": 0.5680085565391529,
      "grad_norm": 129.0,
      "learning_rate": 8.81664499349805e-06,
      "loss": 0.6677,
      "step": 24960
    },
    {
      "epoch": 0.5682361240698177,
      "grad_norm": 286.0,
      "learning_rate": 8.812000743080068e-06,
      "loss": 0.4975,
      "step": 24970
    },
    {
      "epoch": 0.5684636916004825,
      "grad_norm": 0.000675201416015625,
      "learning_rate": 8.807356492662085e-06,
      "loss": 0.575,
      "step": 24980
    },
    {
      "epoch": 0.5686912591311472,
      "grad_norm": 96.0,
      "learning_rate": 8.802712242244102e-06,
      "loss": 1.0276,
      "step": 24990
    },
    {
      "epoch": 0.5689188266618119,
      "grad_norm": 83.0,
      "learning_rate": 8.79806799182612e-06,
      "loss": 0.6041,
      "step": 25000
    },
    {
      "epoch": 0.5691463941924766,
      "grad_norm": 59.25,
      "learning_rate": 8.793423741408138e-06,
      "loss": 0.3358,
      "step": 25010
    },
    {
      "epoch": 0.5693739617231414,
      "grad_norm": 117.0,
      "learning_rate": 8.788779490990155e-06,
      "loss": 1.1294,
      "step": 25020
    },
    {
      "epoch": 0.569601529253806,
      "grad_norm": 90.5,
      "learning_rate": 8.784135240572172e-06,
      "loss": 0.6709,
      "step": 25030
    },
    {
      "epoch": 0.5698290967844708,
      "grad_norm": 0.0001735687255859375,
      "learning_rate": 8.77949099015419e-06,
      "loss": 0.5855,
      "step": 25040
    },
    {
      "epoch": 0.5700566643151355,
      "grad_norm": 165.0,
      "learning_rate": 8.774846739736208e-06,
      "loss": 0.5593,
      "step": 25050
    },
    {
      "epoch": 0.5702842318458002,
      "grad_norm": 1.859375,
      "learning_rate": 8.770202489318225e-06,
      "loss": 0.4913,
      "step": 25060
    },
    {
      "epoch": 0.570511799376465,
      "grad_norm": 712.0,
      "learning_rate": 8.765558238900242e-06,
      "loss": 0.5018,
      "step": 25070
    },
    {
      "epoch": 0.5707393669071297,
      "grad_norm": 33.5,
      "learning_rate": 8.760913988482259e-06,
      "loss": 0.5816,
      "step": 25080
    },
    {
      "epoch": 0.5709669344377944,
      "grad_norm": 4.71875,
      "learning_rate": 8.756269738064278e-06,
      "loss": 0.6243,
      "step": 25090
    },
    {
      "epoch": 0.5711945019684591,
      "grad_norm": 0.236328125,
      "learning_rate": 8.751625487646295e-06,
      "loss": 0.5794,
      "step": 25100
    },
    {
      "epoch": 0.5714220694991239,
      "grad_norm": 1.1171875,
      "learning_rate": 8.746981237228312e-06,
      "loss": 0.4193,
      "step": 25110
    },
    {
      "epoch": 0.5716496370297885,
      "grad_norm": 60.75,
      "learning_rate": 8.742336986810329e-06,
      "loss": 0.5261,
      "step": 25120
    },
    {
      "epoch": 0.5718772045604533,
      "grad_norm": 0.12109375,
      "learning_rate": 8.737692736392347e-06,
      "loss": 1.2418,
      "step": 25130
    },
    {
      "epoch": 0.572104772091118,
      "grad_norm": 0.002044677734375,
      "learning_rate": 8.733048485974365e-06,
      "loss": 0.8983,
      "step": 25140
    },
    {
      "epoch": 0.5723323396217828,
      "grad_norm": 0.0004425048828125,
      "learning_rate": 8.728404235556382e-06,
      "loss": 0.9055,
      "step": 25150
    },
    {
      "epoch": 0.5725599071524475,
      "grad_norm": 157.0,
      "learning_rate": 8.723759985138399e-06,
      "loss": 0.7146,
      "step": 25160
    },
    {
      "epoch": 0.5727874746831122,
      "grad_norm": 0.000621795654296875,
      "learning_rate": 8.719115734720417e-06,
      "loss": 0.6849,
      "step": 25170
    },
    {
      "epoch": 0.573015042213777,
      "grad_norm": 131.0,
      "learning_rate": 8.714471484302434e-06,
      "loss": 0.8447,
      "step": 25180
    },
    {
      "epoch": 0.5732426097444416,
      "grad_norm": 75.0,
      "learning_rate": 8.709827233884452e-06,
      "loss": 0.5961,
      "step": 25190
    },
    {
      "epoch": 0.5734701772751064,
      "grad_norm": 43.0,
      "learning_rate": 8.705182983466469e-06,
      "loss": 0.8061,
      "step": 25200
    },
    {
      "epoch": 0.5736977448057711,
      "grad_norm": 210.0,
      "learning_rate": 8.700538733048487e-06,
      "loss": 0.8985,
      "step": 25210
    },
    {
      "epoch": 0.5739253123364358,
      "grad_norm": 1416.0,
      "learning_rate": 8.695894482630504e-06,
      "loss": 1.0445,
      "step": 25220
    },
    {
      "epoch": 0.5741528798671005,
      "grad_norm": 0.263671875,
      "learning_rate": 8.691250232212521e-06,
      "loss": 0.4159,
      "step": 25230
    },
    {
      "epoch": 0.5743804473977653,
      "grad_norm": 23.125,
      "learning_rate": 8.686605981794538e-06,
      "loss": 1.5902,
      "step": 25240
    },
    {
      "epoch": 0.57460801492843,
      "grad_norm": 932.0,
      "learning_rate": 8.681961731376557e-06,
      "loss": 1.8505,
      "step": 25250
    },
    {
      "epoch": 0.5748355824590947,
      "grad_norm": 137.0,
      "learning_rate": 8.677317480958574e-06,
      "loss": 0.3337,
      "step": 25260
    },
    {
      "epoch": 0.5750631499897595,
      "grad_norm": 27.5,
      "learning_rate": 8.672673230540591e-06,
      "loss": 1.3583,
      "step": 25270
    },
    {
      "epoch": 0.5752907175204242,
      "grad_norm": 0.2197265625,
      "learning_rate": 8.668028980122608e-06,
      "loss": 0.8984,
      "step": 25280
    },
    {
      "epoch": 0.5755182850510889,
      "grad_norm": 0.03271484375,
      "learning_rate": 8.663384729704625e-06,
      "loss": 1.136,
      "step": 25290
    },
    {
      "epoch": 0.5757458525817536,
      "grad_norm": 2.8371810913085938e-05,
      "learning_rate": 8.658740479286644e-06,
      "loss": 0.405,
      "step": 25300
    },
    {
      "epoch": 0.5759734201124184,
      "grad_norm": 229.0,
      "learning_rate": 8.654096228868661e-06,
      "loss": 1.217,
      "step": 25310
    },
    {
      "epoch": 0.5762009876430831,
      "grad_norm": 121.0,
      "learning_rate": 8.64945197845068e-06,
      "loss": 1.1756,
      "step": 25320
    },
    {
      "epoch": 0.5764285551737478,
      "grad_norm": 17.375,
      "learning_rate": 8.644807728032695e-06,
      "loss": 0.5437,
      "step": 25330
    },
    {
      "epoch": 0.5766561227044126,
      "grad_norm": 0.0027923583984375,
      "learning_rate": 8.640163477614714e-06,
      "loss": 0.7501,
      "step": 25340
    },
    {
      "epoch": 0.5768836902350772,
      "grad_norm": 0.005584716796875,
      "learning_rate": 8.635519227196731e-06,
      "loss": 0.3539,
      "step": 25350
    },
    {
      "epoch": 0.577111257765742,
      "grad_norm": 48.25,
      "learning_rate": 8.630874976778748e-06,
      "loss": 2.057,
      "step": 25360
    },
    {
      "epoch": 0.5773388252964067,
      "grad_norm": 157.0,
      "learning_rate": 8.626230726360765e-06,
      "loss": 1.3398,
      "step": 25370
    },
    {
      "epoch": 0.5775663928270715,
      "grad_norm": 73.5,
      "learning_rate": 8.621586475942784e-06,
      "loss": 1.3099,
      "step": 25380
    },
    {
      "epoch": 0.5777939603577361,
      "grad_norm": 308.0,
      "learning_rate": 8.616942225524801e-06,
      "loss": 0.6419,
      "step": 25390
    },
    {
      "epoch": 0.5780215278884009,
      "grad_norm": 0.06640625,
      "learning_rate": 8.612297975106818e-06,
      "loss": 0.297,
      "step": 25400
    },
    {
      "epoch": 0.5782490954190657,
      "grad_norm": 99.5,
      "learning_rate": 8.607653724688835e-06,
      "loss": 0.757,
      "step": 25410
    },
    {
      "epoch": 0.5784766629497303,
      "grad_norm": 220.0,
      "learning_rate": 8.603009474270854e-06,
      "loss": 1.2253,
      "step": 25420
    },
    {
      "epoch": 0.5787042304803951,
      "grad_norm": 0.09228515625,
      "learning_rate": 8.598365223852871e-06,
      "loss": 0.9792,
      "step": 25430
    },
    {
      "epoch": 0.5789317980110598,
      "grad_norm": 956.0,
      "learning_rate": 8.593720973434888e-06,
      "loss": 0.6888,
      "step": 25440
    },
    {
      "epoch": 0.5791593655417245,
      "grad_norm": 350.0,
      "learning_rate": 8.589076723016905e-06,
      "loss": 1.1452,
      "step": 25450
    },
    {
      "epoch": 0.5793869330723892,
      "grad_norm": 0.000278472900390625,
      "learning_rate": 8.584432472598924e-06,
      "loss": 0.5728,
      "step": 25460
    },
    {
      "epoch": 0.579614500603054,
      "grad_norm": 386.0,
      "learning_rate": 8.579788222180941e-06,
      "loss": 0.5878,
      "step": 25470
    },
    {
      "epoch": 0.5798420681337186,
      "grad_norm": 109.5,
      "learning_rate": 8.575143971762958e-06,
      "loss": 0.8491,
      "step": 25480
    },
    {
      "epoch": 0.5800696356643834,
      "grad_norm": 280.0,
      "learning_rate": 8.570499721344977e-06,
      "loss": 0.8904,
      "step": 25490
    },
    {
      "epoch": 0.5802972031950482,
      "grad_norm": 79.0,
      "learning_rate": 8.565855470926992e-06,
      "loss": 0.7012,
      "step": 25500
    },
    {
      "epoch": 0.5805247707257128,
      "grad_norm": 0.71875,
      "learning_rate": 8.56121122050901e-06,
      "loss": 0.169,
      "step": 25510
    },
    {
      "epoch": 0.5807523382563776,
      "grad_norm": 0.0191650390625,
      "learning_rate": 8.556566970091028e-06,
      "loss": 1.1218,
      "step": 25520
    },
    {
      "epoch": 0.5809799057870423,
      "grad_norm": 180.0,
      "learning_rate": 8.551922719673047e-06,
      "loss": 0.4839,
      "step": 25530
    },
    {
      "epoch": 0.5812074733177071,
      "grad_norm": 83.0,
      "learning_rate": 8.547278469255062e-06,
      "loss": 0.744,
      "step": 25540
    },
    {
      "epoch": 0.5814350408483717,
      "grad_norm": 62.0,
      "learning_rate": 8.54263421883708e-06,
      "loss": 1.1455,
      "step": 25550
    },
    {
      "epoch": 0.5816626083790365,
      "grad_norm": 3.5625,
      "learning_rate": 8.537989968419098e-06,
      "loss": 0.9934,
      "step": 25560
    },
    {
      "epoch": 0.5818901759097012,
      "grad_norm": 0.00080108642578125,
      "learning_rate": 8.533345718001115e-06,
      "loss": 0.8347,
      "step": 25570
    },
    {
      "epoch": 0.5821177434403659,
      "grad_norm": 100.0,
      "learning_rate": 8.528701467583132e-06,
      "loss": 2.2822,
      "step": 25580
    },
    {
      "epoch": 0.5823453109710307,
      "grad_norm": 74.5,
      "learning_rate": 8.52405721716515e-06,
      "loss": 0.6918,
      "step": 25590
    },
    {
      "epoch": 0.5825728785016954,
      "grad_norm": 147.0,
      "learning_rate": 8.519412966747168e-06,
      "loss": 1.4254,
      "step": 25600
    },
    {
      "epoch": 0.5828004460323601,
      "grad_norm": 0.44921875,
      "learning_rate": 8.514768716329185e-06,
      "loss": 0.1297,
      "step": 25610
    },
    {
      "epoch": 0.5830280135630248,
      "grad_norm": 40.75,
      "learning_rate": 8.510124465911202e-06,
      "loss": 1.1366,
      "step": 25620
    },
    {
      "epoch": 0.5832555810936896,
      "grad_norm": 167.0,
      "learning_rate": 8.50548021549322e-06,
      "loss": 0.3697,
      "step": 25630
    },
    {
      "epoch": 0.5834831486243542,
      "grad_norm": 166.0,
      "learning_rate": 8.500835965075238e-06,
      "loss": 1.2442,
      "step": 25640
    },
    {
      "epoch": 0.583710716155019,
      "grad_norm": 44.75,
      "learning_rate": 8.496191714657255e-06,
      "loss": 0.7065,
      "step": 25650
    },
    {
      "epoch": 0.5839382836856837,
      "grad_norm": 107.5,
      "learning_rate": 8.491547464239272e-06,
      "loss": 1.1518,
      "step": 25660
    },
    {
      "epoch": 0.5841658512163485,
      "grad_norm": 0.04541015625,
      "learning_rate": 8.48690321382129e-06,
      "loss": 0.2402,
      "step": 25670
    },
    {
      "epoch": 0.5843934187470132,
      "grad_norm": 78.5,
      "learning_rate": 8.482258963403308e-06,
      "loss": 1.0734,
      "step": 25680
    },
    {
      "epoch": 0.5846209862776779,
      "grad_norm": 70.5,
      "learning_rate": 8.477614712985325e-06,
      "loss": 0.5783,
      "step": 25690
    },
    {
      "epoch": 0.5848485538083427,
      "grad_norm": 3.765625,
      "learning_rate": 8.472970462567343e-06,
      "loss": 0.4745,
      "step": 25700
    },
    {
      "epoch": 0.5850761213390073,
      "grad_norm": 142.0,
      "learning_rate": 8.468326212149359e-06,
      "loss": 1.2366,
      "step": 25710
    },
    {
      "epoch": 0.5853036888696721,
      "grad_norm": 0.0035552978515625,
      "learning_rate": 8.463681961731377e-06,
      "loss": 0.561,
      "step": 25720
    },
    {
      "epoch": 0.5855312564003368,
      "grad_norm": 0.00244140625,
      "learning_rate": 8.459037711313394e-06,
      "loss": 0.9686,
      "step": 25730
    },
    {
      "epoch": 0.5857588239310015,
      "grad_norm": 6.375,
      "learning_rate": 8.454393460895413e-06,
      "loss": 1.2211,
      "step": 25740
    },
    {
      "epoch": 0.5859863914616662,
      "grad_norm": 48.0,
      "learning_rate": 8.449749210477429e-06,
      "loss": 1.0324,
      "step": 25750
    },
    {
      "epoch": 0.586213958992331,
      "grad_norm": 68.5,
      "learning_rate": 8.445104960059447e-06,
      "loss": 0.3552,
      "step": 25760
    },
    {
      "epoch": 0.5864415265229957,
      "grad_norm": 466.0,
      "learning_rate": 8.440460709641464e-06,
      "loss": 0.7302,
      "step": 25770
    },
    {
      "epoch": 0.5866690940536604,
      "grad_norm": 108.0,
      "learning_rate": 8.435816459223481e-06,
      "loss": 0.7708,
      "step": 25780
    },
    {
      "epoch": 0.5868966615843252,
      "grad_norm": 150.0,
      "learning_rate": 8.431172208805499e-06,
      "loss": 1.062,
      "step": 25790
    },
    {
      "epoch": 0.5871242291149898,
      "grad_norm": 241.0,
      "learning_rate": 8.426527958387517e-06,
      "loss": 0.975,
      "step": 25800
    },
    {
      "epoch": 0.5873517966456546,
      "grad_norm": 0.0137939453125,
      "learning_rate": 8.421883707969534e-06,
      "loss": 1.0863,
      "step": 25810
    },
    {
      "epoch": 0.5875793641763193,
      "grad_norm": 185.0,
      "learning_rate": 8.417239457551551e-06,
      "loss": 0.5952,
      "step": 25820
    },
    {
      "epoch": 0.5878069317069841,
      "grad_norm": 0.0028533935546875,
      "learning_rate": 8.412595207133568e-06,
      "loss": 0.7066,
      "step": 25830
    },
    {
      "epoch": 0.5880344992376487,
      "grad_norm": 42.25,
      "learning_rate": 8.407950956715587e-06,
      "loss": 0.5089,
      "step": 25840
    },
    {
      "epoch": 0.5882620667683135,
      "grad_norm": 676.0,
      "learning_rate": 8.403306706297604e-06,
      "loss": 1.4851,
      "step": 25850
    },
    {
      "epoch": 0.5884896342989783,
      "grad_norm": 159.0,
      "learning_rate": 8.398662455879621e-06,
      "loss": 1.3547,
      "step": 25860
    },
    {
      "epoch": 0.5887172018296429,
      "grad_norm": 0.29296875,
      "learning_rate": 8.39401820546164e-06,
      "loss": 0.1918,
      "step": 25870
    },
    {
      "epoch": 0.5889447693603077,
      "grad_norm": 14.8125,
      "learning_rate": 8.389373955043657e-06,
      "loss": 1.2163,
      "step": 25880
    },
    {
      "epoch": 0.5891723368909724,
      "grad_norm": 91.5,
      "learning_rate": 8.384729704625674e-06,
      "loss": 1.3602,
      "step": 25890
    },
    {
      "epoch": 0.5893999044216371,
      "grad_norm": 101.0,
      "learning_rate": 8.380085454207691e-06,
      "loss": 0.5255,
      "step": 25900
    },
    {
      "epoch": 0.5896274719523018,
      "grad_norm": 67.0,
      "learning_rate": 8.37544120378971e-06,
      "loss": 0.2264,
      "step": 25910
    },
    {
      "epoch": 0.5898550394829666,
      "grad_norm": 109.5,
      "learning_rate": 8.370796953371727e-06,
      "loss": 0.3729,
      "step": 25920
    },
    {
      "epoch": 0.5900826070136312,
      "grad_norm": 22.375,
      "learning_rate": 8.366152702953744e-06,
      "loss": 0.5509,
      "step": 25930
    },
    {
      "epoch": 0.590310174544296,
      "grad_norm": 63.5,
      "learning_rate": 8.361508452535761e-06,
      "loss": 0.797,
      "step": 25940
    },
    {
      "epoch": 0.5905377420749608,
      "grad_norm": 68.5,
      "learning_rate": 8.35686420211778e-06,
      "loss": 1.002,
      "step": 25950
    },
    {
      "epoch": 0.5907653096056255,
      "grad_norm": 152.0,
      "learning_rate": 8.352219951699795e-06,
      "loss": 0.6576,
      "step": 25960
    },
    {
      "epoch": 0.5909928771362902,
      "grad_norm": 26.625,
      "learning_rate": 8.347575701281814e-06,
      "loss": 0.8116,
      "step": 25970
    },
    {
      "epoch": 0.5912204446669549,
      "grad_norm": 45.75,
      "learning_rate": 8.342931450863831e-06,
      "loss": 0.7567,
      "step": 25980
    },
    {
      "epoch": 0.5914480121976197,
      "grad_norm": 272.0,
      "learning_rate": 8.33828720044585e-06,
      "loss": 0.6953,
      "step": 25990
    },
    {
      "epoch": 0.5916755797282843,
      "grad_norm": 0.0026092529296875,
      "learning_rate": 8.333642950027865e-06,
      "loss": 1.1816,
      "step": 26000
    },
    {
      "epoch": 0.5919031472589491,
      "grad_norm": 132.0,
      "learning_rate": 8.328998699609884e-06,
      "loss": 0.12,
      "step": 26010
    },
    {
      "epoch": 0.5921307147896138,
      "grad_norm": 62.75,
      "learning_rate": 8.324354449191901e-06,
      "loss": 0.5977,
      "step": 26020
    },
    {
      "epoch": 0.5923582823202785,
      "grad_norm": 648.0,
      "learning_rate": 8.319710198773918e-06,
      "loss": 0.9235,
      "step": 26030
    },
    {
      "epoch": 0.5925858498509433,
      "grad_norm": 1.4765625,
      "learning_rate": 8.315065948355937e-06,
      "loss": 0.7423,
      "step": 26040
    },
    {
      "epoch": 0.592813417381608,
      "grad_norm": 0.0004673004150390625,
      "learning_rate": 8.310421697937954e-06,
      "loss": 0.4735,
      "step": 26050
    },
    {
      "epoch": 0.5930409849122728,
      "grad_norm": 2.859375,
      "learning_rate": 8.305777447519971e-06,
      "loss": 0.5857,
      "step": 26060
    },
    {
      "epoch": 0.5932685524429374,
      "grad_norm": 0.01031494140625,
      "learning_rate": 8.301133197101988e-06,
      "loss": 1.1062,
      "step": 26070
    },
    {
      "epoch": 0.5934961199736022,
      "grad_norm": 0.0025787353515625,
      "learning_rate": 8.296488946684007e-06,
      "loss": 0.9363,
      "step": 26080
    },
    {
      "epoch": 0.5937236875042669,
      "grad_norm": 0.006134033203125,
      "learning_rate": 8.291844696266024e-06,
      "loss": 0.8896,
      "step": 26090
    },
    {
      "epoch": 0.5939512550349316,
      "grad_norm": 0.002288818359375,
      "learning_rate": 8.28720044584804e-06,
      "loss": 0.4354,
      "step": 26100
    },
    {
      "epoch": 0.5941788225655963,
      "grad_norm": 322.0,
      "learning_rate": 8.282556195430058e-06,
      "loss": 0.6099,
      "step": 26110
    },
    {
      "epoch": 0.5944063900962611,
      "grad_norm": 87.5,
      "learning_rate": 8.277911945012077e-06,
      "loss": 0.5458,
      "step": 26120
    },
    {
      "epoch": 0.5946339576269258,
      "grad_norm": 19.375,
      "learning_rate": 8.273267694594094e-06,
      "loss": 0.9316,
      "step": 26130
    },
    {
      "epoch": 0.5948615251575905,
      "grad_norm": 0.494140625,
      "learning_rate": 8.26862344417611e-06,
      "loss": 0.4042,
      "step": 26140
    },
    {
      "epoch": 0.5950890926882553,
      "grad_norm": 106.5,
      "learning_rate": 8.263979193758128e-06,
      "loss": 1.4761,
      "step": 26150
    },
    {
      "epoch": 0.5953166602189199,
      "grad_norm": 80.5,
      "learning_rate": 8.259334943340146e-06,
      "loss": 0.6948,
      "step": 26160
    },
    {
      "epoch": 0.5955442277495847,
      "grad_norm": 0.00019741058349609375,
      "learning_rate": 8.254690692922162e-06,
      "loss": 1.0418,
      "step": 26170
    },
    {
      "epoch": 0.5957717952802494,
      "grad_norm": 0.01129150390625,
      "learning_rate": 8.25004644250418e-06,
      "loss": 0.2494,
      "step": 26180
    },
    {
      "epoch": 0.5959993628109141,
      "grad_norm": 185.0,
      "learning_rate": 8.245402192086198e-06,
      "loss": 0.7833,
      "step": 26190
    },
    {
      "epoch": 0.5962269303415788,
      "grad_norm": 4.34375,
      "learning_rate": 8.240757941668216e-06,
      "loss": 0.9126,
      "step": 26200
    },
    {
      "epoch": 0.5964544978722436,
      "grad_norm": 0.007171630859375,
      "learning_rate": 8.236113691250233e-06,
      "loss": 0.6116,
      "step": 26210
    },
    {
      "epoch": 0.5966820654029084,
      "grad_norm": 0.002166748046875,
      "learning_rate": 8.23146944083225e-06,
      "loss": 0.5816,
      "step": 26220
    },
    {
      "epoch": 0.596909632933573,
      "grad_norm": 96.0,
      "learning_rate": 8.226825190414268e-06,
      "loss": 0.4859,
      "step": 26230
    },
    {
      "epoch": 0.5971372004642378,
      "grad_norm": 2.90625,
      "learning_rate": 8.222180939996285e-06,
      "loss": 1.0017,
      "step": 26240
    },
    {
      "epoch": 0.5973647679949025,
      "grad_norm": 86.5,
      "learning_rate": 8.217536689578303e-06,
      "loss": 0.6071,
      "step": 26250
    },
    {
      "epoch": 0.5975923355255672,
      "grad_norm": 262.0,
      "learning_rate": 8.21289243916032e-06,
      "loss": 0.6,
      "step": 26260
    },
    {
      "epoch": 0.5978199030562319,
      "grad_norm": 107.5,
      "learning_rate": 8.208248188742337e-06,
      "loss": 0.7266,
      "step": 26270
    },
    {
      "epoch": 0.5980474705868967,
      "grad_norm": 0.001556396484375,
      "learning_rate": 8.203603938324355e-06,
      "loss": 0.501,
      "step": 26280
    },
    {
      "epoch": 0.5982750381175614,
      "grad_norm": 3.3125,
      "learning_rate": 8.198959687906373e-06,
      "loss": 0.6813,
      "step": 26290
    },
    {
      "epoch": 0.5985026056482261,
      "grad_norm": 0.32421875,
      "learning_rate": 8.19431543748839e-06,
      "loss": 1.4521,
      "step": 26300
    },
    {
      "epoch": 0.5987301731788909,
      "grad_norm": 0.00118255615234375,
      "learning_rate": 8.189671187070407e-06,
      "loss": 0.7628,
      "step": 26310
    },
    {
      "epoch": 0.5989577407095555,
      "grad_norm": 150.0,
      "learning_rate": 8.185026936652424e-06,
      "loss": 1.2952,
      "step": 26320
    },
    {
      "epoch": 0.5991853082402203,
      "grad_norm": 159.0,
      "learning_rate": 8.180382686234443e-06,
      "loss": 1.0989,
      "step": 26330
    },
    {
      "epoch": 0.599412875770885,
      "grad_norm": 0.00168609619140625,
      "learning_rate": 8.17573843581646e-06,
      "loss": 0.6091,
      "step": 26340
    },
    {
      "epoch": 0.5996404433015498,
      "grad_norm": 84.0,
      "learning_rate": 8.171094185398477e-06,
      "loss": 0.52,
      "step": 26350
    },
    {
      "epoch": 0.5998680108322144,
      "grad_norm": 30.875,
      "learning_rate": 8.166449934980494e-06,
      "loss": 1.7106,
      "step": 26360
    },
    {
      "epoch": 0.6000955783628792,
      "grad_norm": 54.75,
      "learning_rate": 8.161805684562513e-06,
      "loss": 0.7871,
      "step": 26370
    },
    {
      "epoch": 0.600323145893544,
      "grad_norm": 167.0,
      "learning_rate": 8.15716143414453e-06,
      "loss": 1.188,
      "step": 26380
    },
    {
      "epoch": 0.6005507134242086,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.152517183726547e-06,
      "loss": 1.3993,
      "step": 26390
    },
    {
      "epoch": 0.6007782809548734,
      "grad_norm": 0.000614166259765625,
      "learning_rate": 8.147872933308564e-06,
      "loss": 1.2213,
      "step": 26400
    },
    {
      "epoch": 0.6010058484855381,
      "grad_norm": 268.0,
      "learning_rate": 8.143228682890583e-06,
      "loss": 0.7938,
      "step": 26410
    },
    {
      "epoch": 0.6012334160162028,
      "grad_norm": 130.0,
      "learning_rate": 8.1385844324726e-06,
      "loss": 1.0251,
      "step": 26420
    },
    {
      "epoch": 0.6014609835468675,
      "grad_norm": 252.0,
      "learning_rate": 8.133940182054617e-06,
      "loss": 1.5981,
      "step": 26430
    },
    {
      "epoch": 0.6016885510775323,
      "grad_norm": 0.004791259765625,
      "learning_rate": 8.129295931636634e-06,
      "loss": 0.8626,
      "step": 26440
    },
    {
      "epoch": 0.6019161186081969,
      "grad_norm": 104.5,
      "learning_rate": 8.124651681218651e-06,
      "loss": 1.0539,
      "step": 26450
    },
    {
      "epoch": 0.6021436861388617,
      "grad_norm": 132.0,
      "learning_rate": 8.12000743080067e-06,
      "loss": 0.3388,
      "step": 26460
    },
    {
      "epoch": 0.6023712536695265,
      "grad_norm": 0.0177001953125,
      "learning_rate": 8.115363180382687e-06,
      "loss": 1.4597,
      "step": 26470
    },
    {
      "epoch": 0.6025988212001911,
      "grad_norm": 105.5,
      "learning_rate": 8.110718929964704e-06,
      "loss": 0.9583,
      "step": 26480
    },
    {
      "epoch": 0.6028263887308559,
      "grad_norm": 1.1328125,
      "learning_rate": 8.106074679546721e-06,
      "loss": 1.4814,
      "step": 26490
    },
    {
      "epoch": 0.6030539562615206,
      "grad_norm": 18.5,
      "learning_rate": 8.10143042912874e-06,
      "loss": 0.6068,
      "step": 26500
    },
    {
      "epoch": 0.6032815237921854,
      "grad_norm": 0.515625,
      "learning_rate": 8.096786178710757e-06,
      "loss": 0.5982,
      "step": 26510
    },
    {
      "epoch": 0.60350909132285,
      "grad_norm": 0.0012054443359375,
      "learning_rate": 8.092141928292774e-06,
      "loss": 1.2551,
      "step": 26520
    },
    {
      "epoch": 0.6037366588535148,
      "grad_norm": 53.75,
      "learning_rate": 8.087497677874791e-06,
      "loss": 0.6468,
      "step": 26530
    },
    {
      "epoch": 0.6039642263841795,
      "grad_norm": 27.125,
      "learning_rate": 8.08285342745681e-06,
      "loss": 0.4572,
      "step": 26540
    },
    {
      "epoch": 0.6041917939148442,
      "grad_norm": 194.0,
      "learning_rate": 8.078209177038827e-06,
      "loss": 0.6716,
      "step": 26550
    },
    {
      "epoch": 0.604419361445509,
      "grad_norm": 220.0,
      "learning_rate": 8.073564926620844e-06,
      "loss": 0.7758,
      "step": 26560
    },
    {
      "epoch": 0.6046469289761737,
      "grad_norm": 0.90234375,
      "learning_rate": 8.068920676202861e-06,
      "loss": 0.708,
      "step": 26570
    },
    {
      "epoch": 0.6048744965068384,
      "grad_norm": 52.25,
      "learning_rate": 8.06427642578488e-06,
      "loss": 0.8855,
      "step": 26580
    },
    {
      "epoch": 0.6051020640375031,
      "grad_norm": 0.00390625,
      "learning_rate": 8.059632175366897e-06,
      "loss": 0.6922,
      "step": 26590
    },
    {
      "epoch": 0.6053296315681679,
      "grad_norm": 96.0,
      "learning_rate": 8.054987924948914e-06,
      "loss": 0.6089,
      "step": 26600
    },
    {
      "epoch": 0.6055571990988325,
      "grad_norm": 60.25,
      "learning_rate": 8.050343674530931e-06,
      "loss": 0.2634,
      "step": 26610
    },
    {
      "epoch": 0.6057847666294973,
      "grad_norm": 171.0,
      "learning_rate": 8.04569942411295e-06,
      "loss": 0.6496,
      "step": 26620
    },
    {
      "epoch": 0.606012334160162,
      "grad_norm": 0.04736328125,
      "learning_rate": 8.041055173694967e-06,
      "loss": 0.3781,
      "step": 26630
    },
    {
      "epoch": 0.6062399016908268,
      "grad_norm": 185.0,
      "learning_rate": 8.036410923276984e-06,
      "loss": 0.6565,
      "step": 26640
    },
    {
      "epoch": 0.6064674692214915,
      "grad_norm": 0.0012054443359375,
      "learning_rate": 8.031766672859e-06,
      "loss": 0.5764,
      "step": 26650
    },
    {
      "epoch": 0.6066950367521562,
      "grad_norm": 93.0,
      "learning_rate": 8.027122422441018e-06,
      "loss": 0.3594,
      "step": 26660
    },
    {
      "epoch": 0.606922604282821,
      "grad_norm": 496.0,
      "learning_rate": 8.022478172023037e-06,
      "loss": 0.9532,
      "step": 26670
    },
    {
      "epoch": 0.6071501718134856,
      "grad_norm": 166.0,
      "learning_rate": 8.017833921605054e-06,
      "loss": 1.0604,
      "step": 26680
    },
    {
      "epoch": 0.6073777393441504,
      "grad_norm": 0.015380859375,
      "learning_rate": 8.013189671187072e-06,
      "loss": 0.5362,
      "step": 26690
    },
    {
      "epoch": 0.6076053068748151,
      "grad_norm": 0.019287109375,
      "learning_rate": 8.008545420769088e-06,
      "loss": 0.8211,
      "step": 26700
    },
    {
      "epoch": 0.6078328744054798,
      "grad_norm": 270.0,
      "learning_rate": 8.003901170351107e-06,
      "loss": 0.5774,
      "step": 26710
    },
    {
      "epoch": 0.6080604419361445,
      "grad_norm": 0.3984375,
      "learning_rate": 7.999256919933124e-06,
      "loss": 0.5301,
      "step": 26720
    },
    {
      "epoch": 0.6082880094668093,
      "grad_norm": 61.0,
      "learning_rate": 7.99461266951514e-06,
      "loss": 0.5963,
      "step": 26730
    },
    {
      "epoch": 0.608515576997474,
      "grad_norm": 0.341796875,
      "learning_rate": 7.989968419097158e-06,
      "loss": 0.8017,
      "step": 26740
    },
    {
      "epoch": 0.6087431445281387,
      "grad_norm": 83.0,
      "learning_rate": 7.985324168679176e-06,
      "loss": 1.4949,
      "step": 26750
    },
    {
      "epoch": 0.6089707120588035,
      "grad_norm": 0.00579833984375,
      "learning_rate": 7.980679918261194e-06,
      "loss": 0.6625,
      "step": 26760
    },
    {
      "epoch": 0.6091982795894682,
      "grad_norm": 0.000946044921875,
      "learning_rate": 7.97603566784321e-06,
      "loss": 1.6605,
      "step": 26770
    },
    {
      "epoch": 0.6094258471201329,
      "grad_norm": 119.5,
      "learning_rate": 7.971391417425228e-06,
      "loss": 1.5189,
      "step": 26780
    },
    {
      "epoch": 0.6096534146507976,
      "grad_norm": 0.003204345703125,
      "learning_rate": 7.966747167007246e-06,
      "loss": 1.4303,
      "step": 26790
    },
    {
      "epoch": 0.6098809821814624,
      "grad_norm": 112.0,
      "learning_rate": 7.962102916589263e-06,
      "loss": 1.1434,
      "step": 26800
    },
    {
      "epoch": 0.610108549712127,
      "grad_norm": 6.75,
      "learning_rate": 7.95745866617128e-06,
      "loss": 0.5335,
      "step": 26810
    },
    {
      "epoch": 0.6103361172427918,
      "grad_norm": 0.9453125,
      "learning_rate": 7.952814415753298e-06,
      "loss": 1.1048,
      "step": 26820
    },
    {
      "epoch": 0.6105636847734566,
      "grad_norm": 222.0,
      "learning_rate": 7.948170165335316e-06,
      "loss": 0.8536,
      "step": 26830
    },
    {
      "epoch": 0.6107912523041212,
      "grad_norm": 63.75,
      "learning_rate": 7.943525914917333e-06,
      "loss": 0.445,
      "step": 26840
    },
    {
      "epoch": 0.611018819834786,
      "grad_norm": 104.0,
      "learning_rate": 7.93888166449935e-06,
      "loss": 0.6573,
      "step": 26850
    },
    {
      "epoch": 0.6112463873654507,
      "grad_norm": 38.5,
      "learning_rate": 7.934237414081367e-06,
      "loss": 0.6572,
      "step": 26860
    },
    {
      "epoch": 0.6114739548961154,
      "grad_norm": 216.0,
      "learning_rate": 7.929593163663384e-06,
      "loss": 1.1735,
      "step": 26870
    },
    {
      "epoch": 0.6117015224267801,
      "grad_norm": 21.5,
      "learning_rate": 7.924948913245403e-06,
      "loss": 0.8914,
      "step": 26880
    },
    {
      "epoch": 0.6119290899574449,
      "grad_norm": 160.0,
      "learning_rate": 7.92030466282742e-06,
      "loss": 0.4202,
      "step": 26890
    },
    {
      "epoch": 0.6121566574881095,
      "grad_norm": 227.0,
      "learning_rate": 7.915660412409439e-06,
      "loss": 0.626,
      "step": 26900
    },
    {
      "epoch": 0.6123842250187743,
      "grad_norm": 0.00048828125,
      "learning_rate": 7.911016161991454e-06,
      "loss": 0.4403,
      "step": 26910
    },
    {
      "epoch": 0.6126117925494391,
      "grad_norm": 0.369140625,
      "learning_rate": 7.906371911573473e-06,
      "loss": 0.96,
      "step": 26920
    },
    {
      "epoch": 0.6128393600801038,
      "grad_norm": 0.091796875,
      "learning_rate": 7.90172766115549e-06,
      "loss": 0.9964,
      "step": 26930
    },
    {
      "epoch": 0.6130669276107685,
      "grad_norm": 144.0,
      "learning_rate": 7.897083410737507e-06,
      "loss": 0.7726,
      "step": 26940
    },
    {
      "epoch": 0.6132944951414332,
      "grad_norm": 0.00286865234375,
      "learning_rate": 7.892439160319524e-06,
      "loss": 0.2591,
      "step": 26950
    },
    {
      "epoch": 0.613522062672098,
      "grad_norm": 95.0,
      "learning_rate": 7.887794909901543e-06,
      "loss": 0.5821,
      "step": 26960
    },
    {
      "epoch": 0.6137496302027626,
      "grad_norm": 312.0,
      "learning_rate": 7.88315065948356e-06,
      "loss": 1.7937,
      "step": 26970
    },
    {
      "epoch": 0.6139771977334274,
      "grad_norm": 5.9375,
      "learning_rate": 7.878506409065577e-06,
      "loss": 0.4598,
      "step": 26980
    },
    {
      "epoch": 0.6142047652640921,
      "grad_norm": 102.5,
      "learning_rate": 7.873862158647594e-06,
      "loss": 0.5729,
      "step": 26990
    },
    {
      "epoch": 0.6144323327947568,
      "grad_norm": 57.5,
      "learning_rate": 7.869217908229613e-06,
      "loss": 1.1091,
      "step": 27000
    },
    {
      "epoch": 0.6146599003254216,
      "grad_norm": 0.031982421875,
      "learning_rate": 7.86457365781163e-06,
      "loss": 0.596,
      "step": 27010
    },
    {
      "epoch": 0.6148874678560863,
      "grad_norm": 0.0029144287109375,
      "learning_rate": 7.859929407393647e-06,
      "loss": 0.7869,
      "step": 27020
    },
    {
      "epoch": 0.615115035386751,
      "grad_norm": 102.5,
      "learning_rate": 7.855285156975664e-06,
      "loss": 1.0048,
      "step": 27030
    },
    {
      "epoch": 0.6153426029174157,
      "grad_norm": 29.875,
      "learning_rate": 7.850640906557683e-06,
      "loss": 0.707,
      "step": 27040
    },
    {
      "epoch": 0.6155701704480805,
      "grad_norm": 129.0,
      "learning_rate": 7.8459966561397e-06,
      "loss": 0.6457,
      "step": 27050
    },
    {
      "epoch": 0.6157977379787452,
      "grad_norm": 96.0,
      "learning_rate": 7.841352405721717e-06,
      "loss": 0.5181,
      "step": 27060
    },
    {
      "epoch": 0.6160253055094099,
      "grad_norm": 510.0,
      "learning_rate": 7.836708155303736e-06,
      "loss": 0.947,
      "step": 27070
    },
    {
      "epoch": 0.6162528730400746,
      "grad_norm": 0.0001468658447265625,
      "learning_rate": 7.832063904885751e-06,
      "loss": 0.7888,
      "step": 27080
    },
    {
      "epoch": 0.6164804405707394,
      "grad_norm": 1.5078125,
      "learning_rate": 7.82741965446777e-06,
      "loss": 0.1453,
      "step": 27090
    },
    {
      "epoch": 0.6167080081014041,
      "grad_norm": 0.9609375,
      "learning_rate": 7.822775404049787e-06,
      "loss": 0.8685,
      "step": 27100
    },
    {
      "epoch": 0.6169355756320688,
      "grad_norm": 392.0,
      "learning_rate": 7.818131153631806e-06,
      "loss": 1.1487,
      "step": 27110
    },
    {
      "epoch": 0.6171631431627336,
      "grad_norm": 0.0439453125,
      "learning_rate": 7.813486903213821e-06,
      "loss": 0.7535,
      "step": 27120
    },
    {
      "epoch": 0.6173907106933982,
      "grad_norm": 60.5,
      "learning_rate": 7.80884265279584e-06,
      "loss": 0.736,
      "step": 27130
    },
    {
      "epoch": 0.617618278224063,
      "grad_norm": 266.0,
      "learning_rate": 7.804198402377857e-06,
      "loss": 0.8922,
      "step": 27140
    },
    {
      "epoch": 0.6178458457547277,
      "grad_norm": 83.5,
      "learning_rate": 7.799554151959874e-06,
      "loss": 1.2095,
      "step": 27150
    },
    {
      "epoch": 0.6180734132853924,
      "grad_norm": 0.67578125,
      "learning_rate": 7.794909901541891e-06,
      "loss": 0.8165,
      "step": 27160
    },
    {
      "epoch": 0.6183009808160571,
      "grad_norm": 422.0,
      "learning_rate": 7.79026565112391e-06,
      "loss": 1.531,
      "step": 27170
    },
    {
      "epoch": 0.6185285483467219,
      "grad_norm": 70.5,
      "learning_rate": 7.785621400705927e-06,
      "loss": 0.2206,
      "step": 27180
    },
    {
      "epoch": 0.6187561158773867,
      "grad_norm": 326.0,
      "learning_rate": 7.780977150287944e-06,
      "loss": 0.8157,
      "step": 27190
    },
    {
      "epoch": 0.6189836834080513,
      "grad_norm": 280.0,
      "learning_rate": 7.776332899869961e-06,
      "loss": 0.7868,
      "step": 27200
    },
    {
      "epoch": 0.6192112509387161,
      "grad_norm": 0.000957489013671875,
      "learning_rate": 7.77168864945198e-06,
      "loss": 0.7919,
      "step": 27210
    },
    {
      "epoch": 0.6194388184693808,
      "grad_norm": 29.875,
      "learning_rate": 7.767044399033997e-06,
      "loss": 0.5761,
      "step": 27220
    },
    {
      "epoch": 0.6196663860000455,
      "grad_norm": 143.0,
      "learning_rate": 7.762400148616014e-06,
      "loss": 1.1154,
      "step": 27230
    },
    {
      "epoch": 0.6198939535307102,
      "grad_norm": 260.0,
      "learning_rate": 7.757755898198032e-06,
      "loss": 0.3777,
      "step": 27240
    },
    {
      "epoch": 0.620121521061375,
      "grad_norm": 1.453125,
      "learning_rate": 7.75311164778005e-06,
      "loss": 1.1455,
      "step": 27250
    },
    {
      "epoch": 0.6203490885920396,
      "grad_norm": 0.03173828125,
      "learning_rate": 7.748467397362067e-06,
      "loss": 1.0765,
      "step": 27260
    },
    {
      "epoch": 0.6205766561227044,
      "grad_norm": 0.01220703125,
      "learning_rate": 7.743823146944084e-06,
      "loss": 0.9248,
      "step": 27270
    },
    {
      "epoch": 0.6208042236533692,
      "grad_norm": 167.0,
      "learning_rate": 7.739178896526102e-06,
      "loss": 0.5301,
      "step": 27280
    },
    {
      "epoch": 0.6210317911840338,
      "grad_norm": 251.0,
      "learning_rate": 7.734534646108118e-06,
      "loss": 1.9197,
      "step": 27290
    },
    {
      "epoch": 0.6212593587146986,
      "grad_norm": 494.0,
      "learning_rate": 7.729890395690136e-06,
      "loss": 0.4547,
      "step": 27300
    },
    {
      "epoch": 0.6214869262453633,
      "grad_norm": 95.0,
      "learning_rate": 7.725246145272154e-06,
      "loss": 0.5393,
      "step": 27310
    },
    {
      "epoch": 0.621714493776028,
      "grad_norm": 116.5,
      "learning_rate": 7.720601894854172e-06,
      "loss": 1.5258,
      "step": 27320
    },
    {
      "epoch": 0.6219420613066927,
      "grad_norm": 0.003387451171875,
      "learning_rate": 7.715957644436188e-06,
      "loss": 0.5219,
      "step": 27330
    },
    {
      "epoch": 0.6221696288373575,
      "grad_norm": 71.0,
      "learning_rate": 7.711313394018206e-06,
      "loss": 1.0385,
      "step": 27340
    },
    {
      "epoch": 0.6223971963680223,
      "grad_norm": 37.75,
      "learning_rate": 7.706669143600223e-06,
      "loss": 0.8267,
      "step": 27350
    },
    {
      "epoch": 0.6226247638986869,
      "grad_norm": 12.4375,
      "learning_rate": 7.70202489318224e-06,
      "loss": 0.1927,
      "step": 27360
    },
    {
      "epoch": 0.6228523314293517,
      "grad_norm": 116.0,
      "learning_rate": 7.697380642764258e-06,
      "loss": 0.8197,
      "step": 27370
    },
    {
      "epoch": 0.6230798989600164,
      "grad_norm": 73.0,
      "learning_rate": 7.692736392346276e-06,
      "loss": 0.3026,
      "step": 27380
    },
    {
      "epoch": 0.6233074664906811,
      "grad_norm": 55.0,
      "learning_rate": 7.688092141928293e-06,
      "loss": 0.8836,
      "step": 27390
    },
    {
      "epoch": 0.6235350340213458,
      "grad_norm": 0.12451171875,
      "learning_rate": 7.68344789151031e-06,
      "loss": 0.3186,
      "step": 27400
    },
    {
      "epoch": 0.6237626015520106,
      "grad_norm": 0.123046875,
      "learning_rate": 7.67880364109233e-06,
      "loss": 0.8208,
      "step": 27410
    },
    {
      "epoch": 0.6239901690826752,
      "grad_norm": 520.0,
      "learning_rate": 7.674159390674346e-06,
      "loss": 0.9229,
      "step": 27420
    },
    {
      "epoch": 0.62421773661334,
      "grad_norm": 912.0,
      "learning_rate": 7.669515140256363e-06,
      "loss": 0.8222,
      "step": 27430
    },
    {
      "epoch": 0.6244453041440048,
      "grad_norm": 100.5,
      "learning_rate": 7.66487088983838e-06,
      "loss": 0.5477,
      "step": 27440
    },
    {
      "epoch": 0.6246728716746694,
      "grad_norm": 139.0,
      "learning_rate": 7.660226639420399e-06,
      "loss": 0.7356,
      "step": 27450
    },
    {
      "epoch": 0.6249004392053342,
      "grad_norm": 19.375,
      "learning_rate": 7.655582389002416e-06,
      "loss": 0.4427,
      "step": 27460
    },
    {
      "epoch": 0.6251280067359989,
      "grad_norm": 187.0,
      "learning_rate": 7.650938138584433e-06,
      "loss": 0.2093,
      "step": 27470
    },
    {
      "epoch": 0.6253555742666637,
      "grad_norm": 0.0030670166015625,
      "learning_rate": 7.64629388816645e-06,
      "loss": 0.4935,
      "step": 27480
    },
    {
      "epoch": 0.6255831417973283,
      "grad_norm": 0.09130859375,
      "learning_rate": 7.641649637748469e-06,
      "loss": 0.3875,
      "step": 27490
    },
    {
      "epoch": 0.6258107093279931,
      "grad_norm": 143.0,
      "learning_rate": 7.637005387330484e-06,
      "loss": 0.9687,
      "step": 27500
    },
    {
      "epoch": 0.6260382768586578,
      "grad_norm": 138.0,
      "learning_rate": 7.632361136912503e-06,
      "loss": 1.1736,
      "step": 27510
    },
    {
      "epoch": 0.6262658443893225,
      "grad_norm": 0.76953125,
      "learning_rate": 7.62771688649452e-06,
      "loss": 0.8334,
      "step": 27520
    },
    {
      "epoch": 0.6264934119199873,
      "grad_norm": 0.0771484375,
      "learning_rate": 7.623072636076538e-06,
      "loss": 0.9366,
      "step": 27530
    },
    {
      "epoch": 0.626720979450652,
      "grad_norm": 16.25,
      "learning_rate": 7.618428385658555e-06,
      "loss": 0.6488,
      "step": 27540
    },
    {
      "epoch": 0.6269485469813167,
      "grad_norm": 46.75,
      "learning_rate": 7.613784135240573e-06,
      "loss": 0.5018,
      "step": 27550
    },
    {
      "epoch": 0.6271761145119814,
      "grad_norm": 0.00634765625,
      "learning_rate": 7.609139884822591e-06,
      "loss": 0.478,
      "step": 27560
    },
    {
      "epoch": 0.6274036820426462,
      "grad_norm": 28.625,
      "learning_rate": 7.604495634404607e-06,
      "loss": 0.7136,
      "step": 27570
    },
    {
      "epoch": 0.6276312495733108,
      "grad_norm": 150.0,
      "learning_rate": 7.599851383986625e-06,
      "loss": 1.0769,
      "step": 27580
    },
    {
      "epoch": 0.6278588171039756,
      "grad_norm": 160.0,
      "learning_rate": 7.595207133568643e-06,
      "loss": 1.7043,
      "step": 27590
    },
    {
      "epoch": 0.6280863846346403,
      "grad_norm": 143.0,
      "learning_rate": 7.590562883150661e-06,
      "loss": 2.028,
      "step": 27600
    },
    {
      "epoch": 0.6283139521653051,
      "grad_norm": 88.0,
      "learning_rate": 7.585918632732677e-06,
      "loss": 0.3123,
      "step": 27610
    },
    {
      "epoch": 0.6285415196959698,
      "grad_norm": 78.5,
      "learning_rate": 7.581274382314695e-06,
      "loss": 2.3333,
      "step": 27620
    },
    {
      "epoch": 0.6287690872266345,
      "grad_norm": 120.5,
      "learning_rate": 7.576630131896713e-06,
      "loss": 0.5039,
      "step": 27630
    },
    {
      "epoch": 0.6289966547572993,
      "grad_norm": 0.0002994537353515625,
      "learning_rate": 7.571985881478729e-06,
      "loss": 1.1203,
      "step": 27640
    },
    {
      "epoch": 0.6292242222879639,
      "grad_norm": 0.0021820068359375,
      "learning_rate": 7.567341631060747e-06,
      "loss": 1.0932,
      "step": 27650
    },
    {
      "epoch": 0.6294517898186287,
      "grad_norm": 185.0,
      "learning_rate": 7.562697380642765e-06,
      "loss": 0.8792,
      "step": 27660
    },
    {
      "epoch": 0.6296793573492934,
      "grad_norm": 213.0,
      "learning_rate": 7.558053130224783e-06,
      "loss": 0.643,
      "step": 27670
    },
    {
      "epoch": 0.6299069248799581,
      "grad_norm": 50.75,
      "learning_rate": 7.5534088798068e-06,
      "loss": 0.6564,
      "step": 27680
    },
    {
      "epoch": 0.6301344924106228,
      "grad_norm": 0.0057373046875,
      "learning_rate": 7.548764629388817e-06,
      "loss": 0.9343,
      "step": 27690
    },
    {
      "epoch": 0.6303620599412876,
      "grad_norm": 91.5,
      "learning_rate": 7.544120378970835e-06,
      "loss": 1.4397,
      "step": 27700
    },
    {
      "epoch": 0.6305896274719524,
      "grad_norm": 0.0086669921875,
      "learning_rate": 7.539476128552852e-06,
      "loss": 0.734,
      "step": 27710
    },
    {
      "epoch": 0.630817195002617,
      "grad_norm": 118.5,
      "learning_rate": 7.53483187813487e-06,
      "loss": 0.8688,
      "step": 27720
    },
    {
      "epoch": 0.6310447625332818,
      "grad_norm": 27.25,
      "learning_rate": 7.530187627716888e-06,
      "loss": 0.5161,
      "step": 27730
    },
    {
      "epoch": 0.6312723300639465,
      "grad_norm": 146.0,
      "learning_rate": 7.525543377298905e-06,
      "loss": 0.4097,
      "step": 27740
    },
    {
      "epoch": 0.6314998975946112,
      "grad_norm": 0.0115966796875,
      "learning_rate": 7.520899126880922e-06,
      "loss": 0.5369,
      "step": 27750
    },
    {
      "epoch": 0.6317274651252759,
      "grad_norm": 1.4140625,
      "learning_rate": 7.51625487646294e-06,
      "loss": 0.7905,
      "step": 27760
    },
    {
      "epoch": 0.6319550326559407,
      "grad_norm": 0.0263671875,
      "learning_rate": 7.5116106260449576e-06,
      "loss": 0.8762,
      "step": 27770
    },
    {
      "epoch": 0.6321826001866053,
      "grad_norm": 0.0181884765625,
      "learning_rate": 7.506966375626974e-06,
      "loss": 2.1919,
      "step": 27780
    },
    {
      "epoch": 0.6324101677172701,
      "grad_norm": 1.8046875,
      "learning_rate": 7.502322125208992e-06,
      "loss": 0.288,
      "step": 27790
    },
    {
      "epoch": 0.6326377352479349,
      "grad_norm": 193.0,
      "learning_rate": 7.4976778747910096e-06,
      "loss": 0.2423,
      "step": 27800
    },
    {
      "epoch": 0.6328653027785995,
      "grad_norm": 11.25,
      "learning_rate": 7.4930336243730275e-06,
      "loss": 0.424,
      "step": 27810
    },
    {
      "epoch": 0.6330928703092643,
      "grad_norm": 0.0126953125,
      "learning_rate": 7.488389373955044e-06,
      "loss": 0.2471,
      "step": 27820
    },
    {
      "epoch": 0.633320437839929,
      "grad_norm": 0.043701171875,
      "learning_rate": 7.4837451235370616e-06,
      "loss": 0.5126,
      "step": 27830
    },
    {
      "epoch": 0.6335480053705937,
      "grad_norm": 134.0,
      "learning_rate": 7.4791008731190795e-06,
      "loss": 1.4243,
      "step": 27840
    },
    {
      "epoch": 0.6337755729012584,
      "grad_norm": 46.5,
      "learning_rate": 7.4744566227010965e-06,
      "loss": 0.3606,
      "step": 27850
    },
    {
      "epoch": 0.6340031404319232,
      "grad_norm": 187.0,
      "learning_rate": 7.469812372283114e-06,
      "loss": 0.9013,
      "step": 27860
    },
    {
      "epoch": 0.6342307079625878,
      "grad_norm": 398.0,
      "learning_rate": 7.4651681218651315e-06,
      "loss": 0.925,
      "step": 27870
    },
    {
      "epoch": 0.6344582754932526,
      "grad_norm": 65.5,
      "learning_rate": 7.460523871447149e-06,
      "loss": 1.8576,
      "step": 27880
    },
    {
      "epoch": 0.6346858430239174,
      "grad_norm": 0.1640625,
      "learning_rate": 7.4558796210291664e-06,
      "loss": 0.9459,
      "step": 27890
    },
    {
      "epoch": 0.6349134105545821,
      "grad_norm": 0.62890625,
      "learning_rate": 7.451235370611184e-06,
      "loss": 0.7991,
      "step": 27900
    },
    {
      "epoch": 0.6351409780852468,
      "grad_norm": 68.5,
      "learning_rate": 7.446591120193201e-06,
      "loss": 1.2029,
      "step": 27910
    },
    {
      "epoch": 0.6353685456159115,
      "grad_norm": 0.003997802734375,
      "learning_rate": 7.4419468697752185e-06,
      "loss": 0.8635,
      "step": 27920
    },
    {
      "epoch": 0.6355961131465763,
      "grad_norm": 2.40625,
      "learning_rate": 7.437302619357236e-06,
      "loss": 1.5138,
      "step": 27930
    },
    {
      "epoch": 0.6358236806772409,
      "grad_norm": 135.0,
      "learning_rate": 7.432658368939254e-06,
      "loss": 1.0647,
      "step": 27940
    },
    {
      "epoch": 0.6360512482079057,
      "grad_norm": 147.0,
      "learning_rate": 7.428014118521272e-06,
      "loss": 1.0918,
      "step": 27950
    },
    {
      "epoch": 0.6362788157385704,
      "grad_norm": 10.375,
      "learning_rate": 7.423369868103288e-06,
      "loss": 0.837,
      "step": 27960
    },
    {
      "epoch": 0.6365063832692351,
      "grad_norm": 79.0,
      "learning_rate": 7.418725617685306e-06,
      "loss": 1.362,
      "step": 27970
    },
    {
      "epoch": 0.6367339507998999,
      "grad_norm": 86.5,
      "learning_rate": 7.414081367267324e-06,
      "loss": 0.5621,
      "step": 27980
    },
    {
      "epoch": 0.6369615183305646,
      "grad_norm": 91.0,
      "learning_rate": 7.40943711684934e-06,
      "loss": 1.5771,
      "step": 27990
    },
    {
      "epoch": 0.6371890858612294,
      "grad_norm": 101.0,
      "learning_rate": 7.404792866431358e-06,
      "loss": 1.1156,
      "step": 28000
    },
    {
      "epoch": 0.637416653391894,
      "grad_norm": 0.01336669921875,
      "learning_rate": 7.400148616013376e-06,
      "loss": 0.8309,
      "step": 28010
    },
    {
      "epoch": 0.6376442209225588,
      "grad_norm": 398.0,
      "learning_rate": 7.395504365595394e-06,
      "loss": 0.9556,
      "step": 28020
    },
    {
      "epoch": 0.6378717884532235,
      "grad_norm": 84.0,
      "learning_rate": 7.39086011517741e-06,
      "loss": 1.2638,
      "step": 28030
    },
    {
      "epoch": 0.6380993559838882,
      "grad_norm": 264.0,
      "learning_rate": 7.386215864759428e-06,
      "loss": 0.9675,
      "step": 28040
    },
    {
      "epoch": 0.6383269235145529,
      "grad_norm": 6.90625,
      "learning_rate": 7.381571614341446e-06,
      "loss": 1.3657,
      "step": 28050
    },
    {
      "epoch": 0.6385544910452177,
      "grad_norm": 0.0164794921875,
      "learning_rate": 7.376927363923463e-06,
      "loss": 0.6357,
      "step": 28060
    },
    {
      "epoch": 0.6387820585758824,
      "grad_norm": 0.033447265625,
      "learning_rate": 7.372283113505481e-06,
      "loss": 0.7065,
      "step": 28070
    },
    {
      "epoch": 0.6390096261065471,
      "grad_norm": 0.83984375,
      "learning_rate": 7.367638863087498e-06,
      "loss": 1.306,
      "step": 28080
    },
    {
      "epoch": 0.6392371936372119,
      "grad_norm": 183.0,
      "learning_rate": 7.362994612669516e-06,
      "loss": 1.5312,
      "step": 28090
    },
    {
      "epoch": 0.6394647611678765,
      "grad_norm": 238.0,
      "learning_rate": 7.358350362251533e-06,
      "loss": 1.1962,
      "step": 28100
    },
    {
      "epoch": 0.6396923286985413,
      "grad_norm": 174.0,
      "learning_rate": 7.353706111833551e-06,
      "loss": 1.8168,
      "step": 28110
    },
    {
      "epoch": 0.639919896229206,
      "grad_norm": 40.25,
      "learning_rate": 7.349061861415568e-06,
      "loss": 1.7208,
      "step": 28120
    },
    {
      "epoch": 0.6401474637598707,
      "grad_norm": 53.0,
      "learning_rate": 7.344417610997585e-06,
      "loss": 0.3827,
      "step": 28130
    },
    {
      "epoch": 0.6403750312905354,
      "grad_norm": 0.3046875,
      "learning_rate": 7.339773360579603e-06,
      "loss": 0.6288,
      "step": 28140
    },
    {
      "epoch": 0.6406025988212002,
      "grad_norm": 0.00543212890625,
      "learning_rate": 7.335129110161621e-06,
      "loss": 0.69,
      "step": 28150
    },
    {
      "epoch": 0.640830166351865,
      "grad_norm": 83.0,
      "learning_rate": 7.330484859743639e-06,
      "loss": 1.0247,
      "step": 28160
    },
    {
      "epoch": 0.6410577338825296,
      "grad_norm": 384.0,
      "learning_rate": 7.325840609325655e-06,
      "loss": 0.4132,
      "step": 28170
    },
    {
      "epoch": 0.6412853014131944,
      "grad_norm": 0.12158203125,
      "learning_rate": 7.321196358907673e-06,
      "loss": 0.6866,
      "step": 28180
    },
    {
      "epoch": 0.6415128689438591,
      "grad_norm": 55.5,
      "learning_rate": 7.316552108489691e-06,
      "loss": 0.7304,
      "step": 28190
    },
    {
      "epoch": 0.6417404364745238,
      "grad_norm": 5.59375,
      "learning_rate": 7.311907858071707e-06,
      "loss": 0.7675,
      "step": 28200
    },
    {
      "epoch": 0.6419680040051885,
      "grad_norm": 4.78125,
      "learning_rate": 7.307263607653725e-06,
      "loss": 0.6377,
      "step": 28210
    },
    {
      "epoch": 0.6421955715358533,
      "grad_norm": 6.40625,
      "learning_rate": 7.302619357235743e-06,
      "loss": 0.2851,
      "step": 28220
    },
    {
      "epoch": 0.6424231390665179,
      "grad_norm": 175.0,
      "learning_rate": 7.297975106817761e-06,
      "loss": 1.0224,
      "step": 28230
    },
    {
      "epoch": 0.6426507065971827,
      "grad_norm": 65.0,
      "learning_rate": 7.293330856399777e-06,
      "loss": 0.8281,
      "step": 28240
    },
    {
      "epoch": 0.6428782741278475,
      "grad_norm": 0.427734375,
      "learning_rate": 7.288686605981795e-06,
      "loss": 1.3506,
      "step": 28250
    },
    {
      "epoch": 0.6431058416585121,
      "grad_norm": 231.0,
      "learning_rate": 7.284042355563813e-06,
      "loss": 1.6154,
      "step": 28260
    },
    {
      "epoch": 0.6433334091891769,
      "grad_norm": 202.0,
      "learning_rate": 7.27939810514583e-06,
      "loss": 0.2823,
      "step": 28270
    },
    {
      "epoch": 0.6435609767198416,
      "grad_norm": 0.51953125,
      "learning_rate": 7.274753854727848e-06,
      "loss": 0.7322,
      "step": 28280
    },
    {
      "epoch": 0.6437885442505064,
      "grad_norm": 218.0,
      "learning_rate": 7.270109604309865e-06,
      "loss": 1.9364,
      "step": 28290
    },
    {
      "epoch": 0.644016111781171,
      "grad_norm": 103.5,
      "learning_rate": 7.265465353891883e-06,
      "loss": 1.0703,
      "step": 28300
    },
    {
      "epoch": 0.6442436793118358,
      "grad_norm": 108.5,
      "learning_rate": 7.2608211034739e-06,
      "loss": 1.0986,
      "step": 28310
    },
    {
      "epoch": 0.6444712468425006,
      "grad_norm": 0.26953125,
      "learning_rate": 7.256176853055918e-06,
      "loss": 0.8981,
      "step": 28320
    },
    {
      "epoch": 0.6446988143731652,
      "grad_norm": 58.0,
      "learning_rate": 7.2515326026379355e-06,
      "loss": 0.6079,
      "step": 28330
    },
    {
      "epoch": 0.64492638190383,
      "grad_norm": 112.0,
      "learning_rate": 7.246888352219952e-06,
      "loss": 0.5997,
      "step": 28340
    },
    {
      "epoch": 0.6451539494344947,
      "grad_norm": 196.0,
      "learning_rate": 7.24224410180197e-06,
      "loss": 0.9189,
      "step": 28350
    },
    {
      "epoch": 0.6453815169651594,
      "grad_norm": 0.0091552734375,
      "learning_rate": 7.2375998513839875e-06,
      "loss": 0.5351,
      "step": 28360
    },
    {
      "epoch": 0.6456090844958241,
      "grad_norm": 148.0,
      "learning_rate": 7.232955600966005e-06,
      "loss": 0.576,
      "step": 28370
    },
    {
      "epoch": 0.6458366520264889,
      "grad_norm": 102.5,
      "learning_rate": 7.228311350548022e-06,
      "loss": 1.0215,
      "step": 28380
    },
    {
      "epoch": 0.6460642195571535,
      "grad_norm": 42.0,
      "learning_rate": 7.2236671001300395e-06,
      "loss": 0.994,
      "step": 28390
    },
    {
      "epoch": 0.6462917870878183,
      "grad_norm": 95.5,
      "learning_rate": 7.2190228497120574e-06,
      "loss": 0.5555,
      "step": 28400
    },
    {
      "epoch": 0.6465193546184831,
      "grad_norm": 0.02001953125,
      "learning_rate": 7.214378599294074e-06,
      "loss": 0.2731,
      "step": 28410
    },
    {
      "epoch": 0.6467469221491478,
      "grad_norm": 3.15625,
      "learning_rate": 7.2097343488760915e-06,
      "loss": 0.8538,
      "step": 28420
    },
    {
      "epoch": 0.6469744896798125,
      "grad_norm": 137.0,
      "learning_rate": 7.2050900984581094e-06,
      "loss": 1.1068,
      "step": 28430
    },
    {
      "epoch": 0.6472020572104772,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.200445848040127e-06,
      "loss": 0.4599,
      "step": 28440
    },
    {
      "epoch": 0.647429624741142,
      "grad_norm": 159.0,
      "learning_rate": 7.195801597622144e-06,
      "loss": 0.6411,
      "step": 28450
    },
    {
      "epoch": 0.6476571922718066,
      "grad_norm": 0.00023746490478515625,
      "learning_rate": 7.1911573472041614e-06,
      "loss": 0.5034,
      "step": 28460
    },
    {
      "epoch": 0.6478847598024714,
      "grad_norm": 7.875,
      "learning_rate": 7.186513096786179e-06,
      "loss": 0.8581,
      "step": 28470
    },
    {
      "epoch": 0.6481123273331361,
      "grad_norm": 0.10595703125,
      "learning_rate": 7.181868846368196e-06,
      "loss": 0.7156,
      "step": 28480
    },
    {
      "epoch": 0.6483398948638008,
      "grad_norm": 89.5,
      "learning_rate": 7.177224595950214e-06,
      "loss": 0.3234,
      "step": 28490
    },
    {
      "epoch": 0.6485674623944656,
      "grad_norm": 190.0,
      "learning_rate": 7.172580345532232e-06,
      "loss": 1.0109,
      "step": 28500
    },
    {
      "epoch": 0.6487950299251303,
      "grad_norm": 89.0,
      "learning_rate": 7.167936095114249e-06,
      "loss": 0.5118,
      "step": 28510
    },
    {
      "epoch": 0.649022597455795,
      "grad_norm": 1.40625,
      "learning_rate": 7.163291844696266e-06,
      "loss": 0.5127,
      "step": 28520
    },
    {
      "epoch": 0.6492501649864597,
      "grad_norm": 84.5,
      "learning_rate": 7.158647594278284e-06,
      "loss": 0.3938,
      "step": 28530
    },
    {
      "epoch": 0.6494777325171245,
      "grad_norm": 1.15625,
      "learning_rate": 7.154003343860302e-06,
      "loss": 0.8762,
      "step": 28540
    },
    {
      "epoch": 0.6497053000477891,
      "grad_norm": 112.5,
      "learning_rate": 7.149359093442318e-06,
      "loss": 0.7791,
      "step": 28550
    },
    {
      "epoch": 0.6499328675784539,
      "grad_norm": 0.0027313232421875,
      "learning_rate": 7.144714843024336e-06,
      "loss": 0.7079,
      "step": 28560
    },
    {
      "epoch": 0.6501604351091186,
      "grad_norm": 138.0,
      "learning_rate": 7.140070592606354e-06,
      "loss": 1.0715,
      "step": 28570
    },
    {
      "epoch": 0.6503880026397834,
      "grad_norm": 88.5,
      "learning_rate": 7.135426342188372e-06,
      "loss": 1.14,
      "step": 28580
    },
    {
      "epoch": 0.6506155701704481,
      "grad_norm": 0.002227783203125,
      "learning_rate": 7.130782091770388e-06,
      "loss": 1.1638,
      "step": 28590
    },
    {
      "epoch": 0.6508431377011128,
      "grad_norm": 193.0,
      "learning_rate": 7.126137841352406e-06,
      "loss": 0.5161,
      "step": 28600
    },
    {
      "epoch": 0.6510707052317776,
      "grad_norm": 22.625,
      "learning_rate": 7.121493590934424e-06,
      "loss": 0.858,
      "step": 28610
    },
    {
      "epoch": 0.6512982727624422,
      "grad_norm": 88.0,
      "learning_rate": 7.116849340516441e-06,
      "loss": 0.6675,
      "step": 28620
    },
    {
      "epoch": 0.651525840293107,
      "grad_norm": 69.5,
      "learning_rate": 7.112205090098458e-06,
      "loss": 0.7692,
      "step": 28630
    },
    {
      "epoch": 0.6517534078237717,
      "grad_norm": 140.0,
      "learning_rate": 7.107560839680476e-06,
      "loss": 0.6866,
      "step": 28640
    },
    {
      "epoch": 0.6519809753544364,
      "grad_norm": 207.0,
      "learning_rate": 7.102916589262494e-06,
      "loss": 0.6844,
      "step": 28650
    },
    {
      "epoch": 0.6522085428851011,
      "grad_norm": 10.6875,
      "learning_rate": 7.098272338844511e-06,
      "loss": 0.5126,
      "step": 28660
    },
    {
      "epoch": 0.6524361104157659,
      "grad_norm": 0.004150390625,
      "learning_rate": 7.093628088426529e-06,
      "loss": 0.4495,
      "step": 28670
    },
    {
      "epoch": 0.6526636779464307,
      "grad_norm": 91.0,
      "learning_rate": 7.088983838008546e-06,
      "loss": 1.1466,
      "step": 28680
    },
    {
      "epoch": 0.6528912454770953,
      "grad_norm": 109.5,
      "learning_rate": 7.084339587590563e-06,
      "loss": 0.8738,
      "step": 28690
    },
    {
      "epoch": 0.6531188130077601,
      "grad_norm": 28.375,
      "learning_rate": 7.079695337172581e-06,
      "loss": 0.5707,
      "step": 28700
    },
    {
      "epoch": 0.6533463805384248,
      "grad_norm": 536.0,
      "learning_rate": 7.075051086754599e-06,
      "loss": 1.0992,
      "step": 28710
    },
    {
      "epoch": 0.6535739480690895,
      "grad_norm": 8.375,
      "learning_rate": 7.070406836336616e-06,
      "loss": 0.4904,
      "step": 28720
    },
    {
      "epoch": 0.6538015155997542,
      "grad_norm": 274.0,
      "learning_rate": 7.065762585918633e-06,
      "loss": 0.9922,
      "step": 28730
    },
    {
      "epoch": 0.654029083130419,
      "grad_norm": 132.0,
      "learning_rate": 7.061118335500651e-06,
      "loss": 0.4915,
      "step": 28740
    },
    {
      "epoch": 0.6542566506610836,
      "grad_norm": 0.036865234375,
      "learning_rate": 7.056474085082669e-06,
      "loss": 0.1973,
      "step": 28750
    },
    {
      "epoch": 0.6544842181917484,
      "grad_norm": 76.0,
      "learning_rate": 7.051829834664685e-06,
      "loss": 0.4562,
      "step": 28760
    },
    {
      "epoch": 0.6547117857224132,
      "grad_norm": 69.0,
      "learning_rate": 7.047185584246703e-06,
      "loss": 0.337,
      "step": 28770
    },
    {
      "epoch": 0.6549393532530778,
      "grad_norm": 169.0,
      "learning_rate": 7.042541333828721e-06,
      "loss": 0.735,
      "step": 28780
    },
    {
      "epoch": 0.6551669207837426,
      "grad_norm": 0.6328125,
      "learning_rate": 7.037897083410739e-06,
      "loss": 1.1803,
      "step": 28790
    },
    {
      "epoch": 0.6553944883144073,
      "grad_norm": 212.0,
      "learning_rate": 7.033252832992755e-06,
      "loss": 0.4718,
      "step": 28800
    },
    {
      "epoch": 0.655622055845072,
      "grad_norm": 314.0,
      "learning_rate": 7.028608582574773e-06,
      "loss": 0.7963,
      "step": 28810
    },
    {
      "epoch": 0.6558496233757367,
      "grad_norm": 156.0,
      "learning_rate": 7.023964332156791e-06,
      "loss": 0.6631,
      "step": 28820
    },
    {
      "epoch": 0.6560771909064015,
      "grad_norm": 302.0,
      "learning_rate": 7.019320081738808e-06,
      "loss": 0.4793,
      "step": 28830
    },
    {
      "epoch": 0.6563047584370661,
      "grad_norm": 0.00677490234375,
      "learning_rate": 7.014675831320825e-06,
      "loss": 0.3534,
      "step": 28840
    },
    {
      "epoch": 0.6565323259677309,
      "grad_norm": 0.02001953125,
      "learning_rate": 7.010031580902843e-06,
      "loss": 1.1589,
      "step": 28850
    },
    {
      "epoch": 0.6567598934983957,
      "grad_norm": 79.0,
      "learning_rate": 7.005387330484861e-06,
      "loss": 1.3525,
      "step": 28860
    },
    {
      "epoch": 0.6569874610290604,
      "grad_norm": 0.00396728515625,
      "learning_rate": 7.000743080066878e-06,
      "loss": 0.8992,
      "step": 28870
    },
    {
      "epoch": 0.6572150285597251,
      "grad_norm": 87.0,
      "learning_rate": 6.9960988296488955e-06,
      "loss": 0.8453,
      "step": 28880
    },
    {
      "epoch": 0.6574425960903898,
      "grad_norm": 0.0189208984375,
      "learning_rate": 6.991454579230913e-06,
      "loss": 0.4191,
      "step": 28890
    },
    {
      "epoch": 0.6576701636210546,
      "grad_norm": 163.0,
      "learning_rate": 6.98681032881293e-06,
      "loss": 0.2026,
      "step": 28900
    },
    {
      "epoch": 0.6578977311517192,
      "grad_norm": 152.0,
      "learning_rate": 6.9821660783949476e-06,
      "loss": 0.6287,
      "step": 28910
    },
    {
      "epoch": 0.658125298682384,
      "grad_norm": 158.0,
      "learning_rate": 6.9775218279769655e-06,
      "loss": 1.5176,
      "step": 28920
    },
    {
      "epoch": 0.6583528662130487,
      "grad_norm": 15.4375,
      "learning_rate": 6.972877577558983e-06,
      "loss": 0.5684,
      "step": 28930
    },
    {
      "epoch": 0.6585804337437134,
      "grad_norm": 27.375,
      "learning_rate": 6.9682333271409996e-06,
      "loss": 0.7891,
      "step": 28940
    },
    {
      "epoch": 0.6588080012743782,
      "grad_norm": 239.0,
      "learning_rate": 6.9635890767230175e-06,
      "loss": 0.6726,
      "step": 28950
    },
    {
      "epoch": 0.6590355688050429,
      "grad_norm": 51.25,
      "learning_rate": 6.958944826305035e-06,
      "loss": 0.9748,
      "step": 28960
    },
    {
      "epoch": 0.6592631363357077,
      "grad_norm": 31.5,
      "learning_rate": 6.954300575887052e-06,
      "loss": 0.8454,
      "step": 28970
    },
    {
      "epoch": 0.6594907038663723,
      "grad_norm": 107.0,
      "learning_rate": 6.9496563254690695e-06,
      "loss": 0.8428,
      "step": 28980
    },
    {
      "epoch": 0.6597182713970371,
      "grad_norm": 2.875,
      "learning_rate": 6.945012075051087e-06,
      "loss": 0.8919,
      "step": 28990
    },
    {
      "epoch": 0.6599458389277018,
      "grad_norm": 207.0,
      "learning_rate": 6.940367824633105e-06,
      "loss": 0.6434,
      "step": 29000
    },
    {
      "epoch": 0.6601734064583665,
      "grad_norm": 253.0,
      "learning_rate": 6.9357235742151215e-06,
      "loss": 1.5361,
      "step": 29010
    },
    {
      "epoch": 0.6604009739890312,
      "grad_norm": 79.0,
      "learning_rate": 6.931079323797139e-06,
      "loss": 1.1939,
      "step": 29020
    },
    {
      "epoch": 0.660628541519696,
      "grad_norm": 89.5,
      "learning_rate": 6.926435073379157e-06,
      "loss": 1.1751,
      "step": 29030
    },
    {
      "epoch": 0.6608561090503607,
      "grad_norm": 175.0,
      "learning_rate": 6.921790822961174e-06,
      "loss": 1.3992,
      "step": 29040
    },
    {
      "epoch": 0.6610836765810254,
      "grad_norm": 1.578125,
      "learning_rate": 6.917146572543192e-06,
      "loss": 0.8139,
      "step": 29050
    },
    {
      "epoch": 0.6613112441116902,
      "grad_norm": 99.5,
      "learning_rate": 6.912502322125209e-06,
      "loss": 0.7943,
      "step": 29060
    },
    {
      "epoch": 0.6615388116423548,
      "grad_norm": 2.921875,
      "learning_rate": 6.907858071707227e-06,
      "loss": 0.5736,
      "step": 29070
    },
    {
      "epoch": 0.6617663791730196,
      "grad_norm": 0.005645751953125,
      "learning_rate": 6.903213821289244e-06,
      "loss": 0.1383,
      "step": 29080
    },
    {
      "epoch": 0.6619939467036843,
      "grad_norm": 67.5,
      "learning_rate": 6.898569570871262e-06,
      "loss": 0.6006,
      "step": 29090
    },
    {
      "epoch": 0.662221514234349,
      "grad_norm": 38.0,
      "learning_rate": 6.89392532045328e-06,
      "loss": 1.2577,
      "step": 29100
    },
    {
      "epoch": 0.6624490817650137,
      "grad_norm": 99.5,
      "learning_rate": 6.889281070035296e-06,
      "loss": 0.7868,
      "step": 29110
    },
    {
      "epoch": 0.6626766492956785,
      "grad_norm": 248.0,
      "learning_rate": 6.884636819617314e-06,
      "loss": 0.9668,
      "step": 29120
    },
    {
      "epoch": 0.6629042168263433,
      "grad_norm": 191.0,
      "learning_rate": 6.879992569199332e-06,
      "loss": 1.1018,
      "step": 29130
    },
    {
      "epoch": 0.6631317843570079,
      "grad_norm": 10.875,
      "learning_rate": 6.87534831878135e-06,
      "loss": 0.7908,
      "step": 29140
    },
    {
      "epoch": 0.6633593518876727,
      "grad_norm": 144.0,
      "learning_rate": 6.870704068363366e-06,
      "loss": 1.52,
      "step": 29150
    },
    {
      "epoch": 0.6635869194183374,
      "grad_norm": 192.0,
      "learning_rate": 6.866059817945384e-06,
      "loss": 1.2771,
      "step": 29160
    },
    {
      "epoch": 0.6638144869490021,
      "grad_norm": 94.5,
      "learning_rate": 6.861415567527402e-06,
      "loss": 0.5152,
      "step": 29170
    },
    {
      "epoch": 0.6640420544796668,
      "grad_norm": 0.003936767578125,
      "learning_rate": 6.856771317109418e-06,
      "loss": 0.6822,
      "step": 29180
    },
    {
      "epoch": 0.6642696220103316,
      "grad_norm": 21.625,
      "learning_rate": 6.852127066691436e-06,
      "loss": 0.6673,
      "step": 29190
    },
    {
      "epoch": 0.6644971895409962,
      "grad_norm": 158.0,
      "learning_rate": 6.847482816273454e-06,
      "loss": 1.1962,
      "step": 29200
    },
    {
      "epoch": 0.664724757071661,
      "grad_norm": 62.25,
      "learning_rate": 6.842838565855472e-06,
      "loss": 0.5138,
      "step": 29210
    },
    {
      "epoch": 0.6649523246023258,
      "grad_norm": 284.0,
      "learning_rate": 6.838194315437489e-06,
      "loss": 1.3191,
      "step": 29220
    },
    {
      "epoch": 0.6651798921329904,
      "grad_norm": 63.25,
      "learning_rate": 6.833550065019506e-06,
      "loss": 0.3344,
      "step": 29230
    },
    {
      "epoch": 0.6654074596636552,
      "grad_norm": 215.0,
      "learning_rate": 6.828905814601524e-06,
      "loss": 1.3105,
      "step": 29240
    },
    {
      "epoch": 0.6656350271943199,
      "grad_norm": 342.0,
      "learning_rate": 6.824261564183541e-06,
      "loss": 1.1083,
      "step": 29250
    },
    {
      "epoch": 0.6658625947249847,
      "grad_norm": 232.0,
      "learning_rate": 6.819617313765559e-06,
      "loss": 0.7288,
      "step": 29260
    },
    {
      "epoch": 0.6660901622556493,
      "grad_norm": 274.0,
      "learning_rate": 6.814973063347577e-06,
      "loss": 1.2467,
      "step": 29270
    },
    {
      "epoch": 0.6663177297863141,
      "grad_norm": 274.0,
      "learning_rate": 6.810328812929594e-06,
      "loss": 1.363,
      "step": 29280
    },
    {
      "epoch": 0.6665452973169789,
      "grad_norm": 0.6953125,
      "learning_rate": 6.805684562511611e-06,
      "loss": 0.7181,
      "step": 29290
    },
    {
      "epoch": 0.6667728648476435,
      "grad_norm": 253.0,
      "learning_rate": 6.801040312093629e-06,
      "loss": 0.8906,
      "step": 29300
    },
    {
      "epoch": 0.6670004323783083,
      "grad_norm": 124.5,
      "learning_rate": 6.796396061675647e-06,
      "loss": 0.5481,
      "step": 29310
    },
    {
      "epoch": 0.667227999908973,
      "grad_norm": 0.002685546875,
      "learning_rate": 6.791751811257663e-06,
      "loss": 0.6924,
      "step": 29320
    },
    {
      "epoch": 0.6674555674396377,
      "grad_norm": 164.0,
      "learning_rate": 6.787107560839681e-06,
      "loss": 0.7624,
      "step": 29330
    },
    {
      "epoch": 0.6676831349703024,
      "grad_norm": 105.5,
      "learning_rate": 6.782463310421699e-06,
      "loss": 0.8895,
      "step": 29340
    },
    {
      "epoch": 0.6679107025009672,
      "grad_norm": 0.00136566162109375,
      "learning_rate": 6.777819060003717e-06,
      "loss": 0.8149,
      "step": 29350
    },
    {
      "epoch": 0.6681382700316318,
      "grad_norm": 137.0,
      "learning_rate": 6.773174809585733e-06,
      "loss": 1.0462,
      "step": 29360
    },
    {
      "epoch": 0.6683658375622966,
      "grad_norm": 76.5,
      "learning_rate": 6.768530559167751e-06,
      "loss": 0.5308,
      "step": 29370
    },
    {
      "epoch": 0.6685934050929614,
      "grad_norm": 0.036865234375,
      "learning_rate": 6.763886308749769e-06,
      "loss": 0.4774,
      "step": 29380
    },
    {
      "epoch": 0.668820972623626,
      "grad_norm": 142.0,
      "learning_rate": 6.759242058331786e-06,
      "loss": 0.6906,
      "step": 29390
    },
    {
      "epoch": 0.6690485401542908,
      "grad_norm": 152.0,
      "learning_rate": 6.754597807913803e-06,
      "loss": 0.708,
      "step": 29400
    },
    {
      "epoch": 0.6692761076849555,
      "grad_norm": 0.005035400390625,
      "learning_rate": 6.749953557495821e-06,
      "loss": 1.7272,
      "step": 29410
    },
    {
      "epoch": 0.6695036752156203,
      "grad_norm": 50.25,
      "learning_rate": 6.7453093070778385e-06,
      "loss": 0.4084,
      "step": 29420
    },
    {
      "epoch": 0.6697312427462849,
      "grad_norm": 0.435546875,
      "learning_rate": 6.740665056659856e-06,
      "loss": 0.8288,
      "step": 29430
    },
    {
      "epoch": 0.6699588102769497,
      "grad_norm": 0.0013885498046875,
      "learning_rate": 6.736020806241873e-06,
      "loss": 0.9777,
      "step": 29440
    },
    {
      "epoch": 0.6701863778076144,
      "grad_norm": 1.3671875,
      "learning_rate": 6.7313765558238905e-06,
      "loss": 0.5848,
      "step": 29450
    },
    {
      "epoch": 0.6704139453382791,
      "grad_norm": 162.0,
      "learning_rate": 6.726732305405908e-06,
      "loss": 1.2781,
      "step": 29460
    },
    {
      "epoch": 0.6706415128689439,
      "grad_norm": 11.0625,
      "learning_rate": 6.7220880549879255e-06,
      "loss": 0.6443,
      "step": 29470
    },
    {
      "epoch": 0.6708690803996086,
      "grad_norm": 276.0,
      "learning_rate": 6.717443804569943e-06,
      "loss": 0.6079,
      "step": 29480
    },
    {
      "epoch": 0.6710966479302733,
      "grad_norm": 0.130859375,
      "learning_rate": 6.7127995541519605e-06,
      "loss": 1.5336,
      "step": 29490
    },
    {
      "epoch": 0.671324215460938,
      "grad_norm": 0.00982666015625,
      "learning_rate": 6.7081553037339775e-06,
      "loss": 0.739,
      "step": 29500
    },
    {
      "epoch": 0.6715517829916028,
      "grad_norm": 1.484375,
      "learning_rate": 6.703511053315995e-06,
      "loss": 0.1794,
      "step": 29510
    },
    {
      "epoch": 0.6717793505222674,
      "grad_norm": 0.0751953125,
      "learning_rate": 6.698866802898013e-06,
      "loss": 1.0048,
      "step": 29520
    },
    {
      "epoch": 0.6720069180529322,
      "grad_norm": 110.5,
      "learning_rate": 6.6942225524800295e-06,
      "loss": 0.5569,
      "step": 29530
    },
    {
      "epoch": 0.6722344855835969,
      "grad_norm": 110.0,
      "learning_rate": 6.6895783020620474e-06,
      "loss": 0.4082,
      "step": 29540
    },
    {
      "epoch": 0.6724620531142617,
      "grad_norm": 0.0517578125,
      "learning_rate": 6.684934051644065e-06,
      "loss": 1.1431,
      "step": 29550
    },
    {
      "epoch": 0.6726896206449264,
      "grad_norm": 61.25,
      "learning_rate": 6.680289801226083e-06,
      "loss": 0.5949,
      "step": 29560
    },
    {
      "epoch": 0.6729171881755911,
      "grad_norm": 23.5,
      "learning_rate": 6.6756455508080994e-06,
      "loss": 0.4307,
      "step": 29570
    },
    {
      "epoch": 0.6731447557062559,
      "grad_norm": 127.0,
      "learning_rate": 6.671001300390117e-06,
      "loss": 0.6261,
      "step": 29580
    },
    {
      "epoch": 0.6733723232369205,
      "grad_norm": 0.20703125,
      "learning_rate": 6.666357049972135e-06,
      "loss": 0.8326,
      "step": 29590
    },
    {
      "epoch": 0.6735998907675853,
      "grad_norm": 0.00213623046875,
      "learning_rate": 6.661712799554152e-06,
      "loss": 0.6776,
      "step": 29600
    },
    {
      "epoch": 0.67382745829825,
      "grad_norm": 0.53515625,
      "learning_rate": 6.657068549136169e-06,
      "loss": 0.1373,
      "step": 29610
    },
    {
      "epoch": 0.6740550258289147,
      "grad_norm": 230.0,
      "learning_rate": 6.652424298718187e-06,
      "loss": 0.6001,
      "step": 29620
    },
    {
      "epoch": 0.6742825933595794,
      "grad_norm": 243.0,
      "learning_rate": 6.647780048300205e-06,
      "loss": 0.5196,
      "step": 29630
    },
    {
      "epoch": 0.6745101608902442,
      "grad_norm": 0.0145263671875,
      "learning_rate": 6.643135797882222e-06,
      "loss": 0.293,
      "step": 29640
    },
    {
      "epoch": 0.674737728420909,
      "grad_norm": 334.0,
      "learning_rate": 6.63849154746424e-06,
      "loss": 1.9418,
      "step": 29650
    },
    {
      "epoch": 0.6749652959515736,
      "grad_norm": 66.0,
      "learning_rate": 6.633847297046257e-06,
      "loss": 1.1006,
      "step": 29660
    },
    {
      "epoch": 0.6751928634822384,
      "grad_norm": 59.75,
      "learning_rate": 6.629203046628274e-06,
      "loss": 0.2896,
      "step": 29670
    },
    {
      "epoch": 0.675420431012903,
      "grad_norm": 0.06591796875,
      "learning_rate": 6.624558796210292e-06,
      "loss": 0.6353,
      "step": 29680
    },
    {
      "epoch": 0.6756479985435678,
      "grad_norm": 73.5,
      "learning_rate": 6.61991454579231e-06,
      "loss": 0.6565,
      "step": 29690
    },
    {
      "epoch": 0.6758755660742325,
      "grad_norm": 260.0,
      "learning_rate": 6.615270295374328e-06,
      "loss": 1.1819,
      "step": 29700
    },
    {
      "epoch": 0.6761031336048973,
      "grad_norm": 74.5,
      "learning_rate": 6.610626044956344e-06,
      "loss": 1.4757,
      "step": 29710
    },
    {
      "epoch": 0.6763307011355619,
      "grad_norm": 0.29296875,
      "learning_rate": 6.605981794538362e-06,
      "loss": 0.4499,
      "step": 29720
    },
    {
      "epoch": 0.6765582686662267,
      "grad_norm": 138.0,
      "learning_rate": 6.60133754412038e-06,
      "loss": 0.5811,
      "step": 29730
    },
    {
      "epoch": 0.6767858361968915,
      "grad_norm": 410.0,
      "learning_rate": 6.596693293702398e-06,
      "loss": 1.5365,
      "step": 29740
    },
    {
      "epoch": 0.6770134037275561,
      "grad_norm": 113.5,
      "learning_rate": 6.592049043284414e-06,
      "loss": 0.5053,
      "step": 29750
    },
    {
      "epoch": 0.6772409712582209,
      "grad_norm": 306.0,
      "learning_rate": 6.587404792866432e-06,
      "loss": 0.7087,
      "step": 29760
    },
    {
      "epoch": 0.6774685387888856,
      "grad_norm": 94.0,
      "learning_rate": 6.58276054244845e-06,
      "loss": 0.659,
      "step": 29770
    },
    {
      "epoch": 0.6776961063195504,
      "grad_norm": 220.0,
      "learning_rate": 6.578116292030466e-06,
      "loss": 1.0416,
      "step": 29780
    },
    {
      "epoch": 0.677923673850215,
      "grad_norm": 264.0,
      "learning_rate": 6.573472041612484e-06,
      "loss": 0.4819,
      "step": 29790
    },
    {
      "epoch": 0.6781512413808798,
      "grad_norm": 165.0,
      "learning_rate": 6.568827791194502e-06,
      "loss": 0.8764,
      "step": 29800
    },
    {
      "epoch": 0.6783788089115445,
      "grad_norm": 458.0,
      "learning_rate": 6.56418354077652e-06,
      "loss": 0.9902,
      "step": 29810
    },
    {
      "epoch": 0.6786063764422092,
      "grad_norm": 38.25,
      "learning_rate": 6.559539290358537e-06,
      "loss": 1.1581,
      "step": 29820
    },
    {
      "epoch": 0.678833943972874,
      "grad_norm": 370.0,
      "learning_rate": 6.554895039940554e-06,
      "loss": 0.8237,
      "step": 29830
    },
    {
      "epoch": 0.6790615115035387,
      "grad_norm": 163.0,
      "learning_rate": 6.550250789522572e-06,
      "loss": 0.188,
      "step": 29840
    },
    {
      "epoch": 0.6792890790342034,
      "grad_norm": 69.5,
      "learning_rate": 6.545606539104589e-06,
      "loss": 0.6225,
      "step": 29850
    },
    {
      "epoch": 0.6795166465648681,
      "grad_norm": 1.015625,
      "learning_rate": 6.540962288686607e-06,
      "loss": 0.3744,
      "step": 29860
    },
    {
      "epoch": 0.6797442140955329,
      "grad_norm": 110.5,
      "learning_rate": 6.536318038268624e-06,
      "loss": 1.0366,
      "step": 29870
    },
    {
      "epoch": 0.6799717816261975,
      "grad_norm": 132.0,
      "learning_rate": 6.531673787850642e-06,
      "loss": 0.8929,
      "step": 29880
    },
    {
      "epoch": 0.6801993491568623,
      "grad_norm": 3.25,
      "learning_rate": 6.527029537432659e-06,
      "loss": 0.6263,
      "step": 29890
    },
    {
      "epoch": 0.680426916687527,
      "grad_norm": 145.0,
      "learning_rate": 6.522385287014677e-06,
      "loss": 0.8416,
      "step": 29900
    },
    {
      "epoch": 0.6806544842181917,
      "grad_norm": 167.0,
      "learning_rate": 6.5177410365966946e-06,
      "loss": 1.3066,
      "step": 29910
    },
    {
      "epoch": 0.6808820517488565,
      "grad_norm": 42.25,
      "learning_rate": 6.513096786178711e-06,
      "loss": 0.8588,
      "step": 29920
    },
    {
      "epoch": 0.6811096192795212,
      "grad_norm": 3.625,
      "learning_rate": 6.508452535760729e-06,
      "loss": 0.5016,
      "step": 29930
    },
    {
      "epoch": 0.681337186810186,
      "grad_norm": 0.00164794921875,
      "learning_rate": 6.5038082853427466e-06,
      "loss": 0.6076,
      "step": 29940
    },
    {
      "epoch": 0.6815647543408506,
      "grad_norm": 314.0,
      "learning_rate": 6.4991640349247645e-06,
      "loss": 0.8688,
      "step": 29950
    },
    {
      "epoch": 0.6817923218715154,
      "grad_norm": 150.0,
      "learning_rate": 6.494519784506781e-06,
      "loss": 0.6828,
      "step": 29960
    },
    {
      "epoch": 0.6820198894021801,
      "grad_norm": 0.10791015625,
      "learning_rate": 6.4898755340887986e-06,
      "loss": 1.1902,
      "step": 29970
    },
    {
      "epoch": 0.6822474569328448,
      "grad_norm": 135.0,
      "learning_rate": 6.4852312836708165e-06,
      "loss": 0.6441,
      "step": 29980
    },
    {
      "epoch": 0.6824750244635095,
      "grad_norm": 0.5390625,
      "learning_rate": 6.4805870332528335e-06,
      "loss": 0.6523,
      "step": 29990
    },
    {
      "epoch": 0.6827025919941743,
      "grad_norm": 0.00860595703125,
      "learning_rate": 6.475942782834851e-06,
      "loss": 0.3291,
      "step": 30000
    },
    {
      "epoch": 0.682930159524839,
      "grad_norm": 33.25,
      "learning_rate": 6.4712985324168685e-06,
      "loss": 0.4691,
      "step": 30010
    },
    {
      "epoch": 0.6831577270555037,
      "grad_norm": 294.0,
      "learning_rate": 6.466654281998886e-06,
      "loss": 1.0303,
      "step": 30020
    },
    {
      "epoch": 0.6833852945861685,
      "grad_norm": 0.001007080078125,
      "learning_rate": 6.4620100315809034e-06,
      "loss": 0.0157,
      "step": 30030
    },
    {
      "epoch": 0.6836128621168331,
      "grad_norm": 132.0,
      "learning_rate": 6.4573657811629205e-06,
      "loss": 0.5983,
      "step": 30040
    },
    {
      "epoch": 0.6838404296474979,
      "grad_norm": 0.00118255615234375,
      "learning_rate": 6.452721530744938e-06,
      "loss": 0.6903,
      "step": 30050
    },
    {
      "epoch": 0.6840679971781626,
      "grad_norm": 89.0,
      "learning_rate": 6.4480772803269555e-06,
      "loss": 0.6733,
      "step": 30060
    },
    {
      "epoch": 0.6842955647088274,
      "grad_norm": 0.015869140625,
      "learning_rate": 6.443433029908973e-06,
      "loss": 0.48,
      "step": 30070
    },
    {
      "epoch": 0.684523132239492,
      "grad_norm": 8.0625,
      "learning_rate": 6.438788779490991e-06,
      "loss": 1.8604,
      "step": 30080
    },
    {
      "epoch": 0.6847506997701568,
      "grad_norm": 0.0034637451171875,
      "learning_rate": 6.434144529073008e-06,
      "loss": 0.2588,
      "step": 30090
    },
    {
      "epoch": 0.6849782673008216,
      "grad_norm": 0.00665283203125,
      "learning_rate": 6.429500278655025e-06,
      "loss": 0.9941,
      "step": 30100
    },
    {
      "epoch": 0.6852058348314862,
      "grad_norm": 304.0,
      "learning_rate": 6.424856028237043e-06,
      "loss": 0.9489,
      "step": 30110
    },
    {
      "epoch": 0.685433402362151,
      "grad_norm": 98.0,
      "learning_rate": 6.420211777819061e-06,
      "loss": 1.0994,
      "step": 30120
    },
    {
      "epoch": 0.6856609698928157,
      "grad_norm": 328.0,
      "learning_rate": 6.415567527401077e-06,
      "loss": 1.1752,
      "step": 30130
    },
    {
      "epoch": 0.6858885374234804,
      "grad_norm": 80.5,
      "learning_rate": 6.410923276983095e-06,
      "loss": 0.5964,
      "step": 30140
    },
    {
      "epoch": 0.6861161049541451,
      "grad_norm": 219.0,
      "learning_rate": 6.406279026565113e-06,
      "loss": 1.1011,
      "step": 30150
    },
    {
      "epoch": 0.6863436724848099,
      "grad_norm": 107.0,
      "learning_rate": 6.401634776147131e-06,
      "loss": 0.3195,
      "step": 30160
    },
    {
      "epoch": 0.6865712400154745,
      "grad_norm": 138.0,
      "learning_rate": 6.396990525729147e-06,
      "loss": 0.55,
      "step": 30170
    },
    {
      "epoch": 0.6867988075461393,
      "grad_norm": 0.625,
      "learning_rate": 6.392346275311165e-06,
      "loss": 0.2786,
      "step": 30180
    },
    {
      "epoch": 0.6870263750768041,
      "grad_norm": 34.25,
      "learning_rate": 6.387702024893183e-06,
      "loss": 0.835,
      "step": 30190
    },
    {
      "epoch": 0.6872539426074687,
      "grad_norm": 0.0011444091796875,
      "learning_rate": 6.3830577744752e-06,
      "loss": 0.9243,
      "step": 30200
    },
    {
      "epoch": 0.6874815101381335,
      "grad_norm": 0.006011962890625,
      "learning_rate": 6.378413524057217e-06,
      "loss": 1.5308,
      "step": 30210
    },
    {
      "epoch": 0.6877090776687982,
      "grad_norm": 0.2099609375,
      "learning_rate": 6.373769273639235e-06,
      "loss": 0.4434,
      "step": 30220
    },
    {
      "epoch": 0.687936645199463,
      "grad_norm": 0.51171875,
      "learning_rate": 6.369125023221253e-06,
      "loss": 0.951,
      "step": 30230
    },
    {
      "epoch": 0.6881642127301276,
      "grad_norm": 0.00104522705078125,
      "learning_rate": 6.36448077280327e-06,
      "loss": 1.0154,
      "step": 30240
    },
    {
      "epoch": 0.6883917802607924,
      "grad_norm": 324.0,
      "learning_rate": 6.359836522385288e-06,
      "loss": 0.5114,
      "step": 30250
    },
    {
      "epoch": 0.6886193477914572,
      "grad_norm": 0.431640625,
      "learning_rate": 6.355192271967305e-06,
      "loss": 0.5994,
      "step": 30260
    },
    {
      "epoch": 0.6888469153221218,
      "grad_norm": 178.0,
      "learning_rate": 6.350548021549322e-06,
      "loss": 0.9954,
      "step": 30270
    },
    {
      "epoch": 0.6890744828527866,
      "grad_norm": 55.25,
      "learning_rate": 6.34590377113134e-06,
      "loss": 0.2583,
      "step": 30280
    },
    {
      "epoch": 0.6893020503834513,
      "grad_norm": 0.003448486328125,
      "learning_rate": 6.341259520713358e-06,
      "loss": 1.5482,
      "step": 30290
    },
    {
      "epoch": 0.689529617914116,
      "grad_norm": 58.5,
      "learning_rate": 6.336615270295376e-06,
      "loss": 0.2008,
      "step": 30300
    },
    {
      "epoch": 0.6897571854447807,
      "grad_norm": 67.0,
      "learning_rate": 6.331971019877392e-06,
      "loss": 0.5176,
      "step": 30310
    },
    {
      "epoch": 0.6899847529754455,
      "grad_norm": 150.0,
      "learning_rate": 6.32732676945941e-06,
      "loss": 1.0199,
      "step": 30320
    },
    {
      "epoch": 0.6902123205061101,
      "grad_norm": 182.0,
      "learning_rate": 6.322682519041428e-06,
      "loss": 1.6286,
      "step": 30330
    },
    {
      "epoch": 0.6904398880367749,
      "grad_norm": 181.0,
      "learning_rate": 6.318038268623444e-06,
      "loss": 0.6565,
      "step": 30340
    },
    {
      "epoch": 0.6906674555674397,
      "grad_norm": 0.003082275390625,
      "learning_rate": 6.313394018205462e-06,
      "loss": 1.4196,
      "step": 30350
    },
    {
      "epoch": 0.6908950230981044,
      "grad_norm": 0.046630859375,
      "learning_rate": 6.30874976778748e-06,
      "loss": 1.3721,
      "step": 30360
    },
    {
      "epoch": 0.6911225906287691,
      "grad_norm": 110.5,
      "learning_rate": 6.304105517369498e-06,
      "loss": 0.9599,
      "step": 30370
    },
    {
      "epoch": 0.6913501581594338,
      "grad_norm": 0.005584716796875,
      "learning_rate": 6.299461266951514e-06,
      "loss": 0.5141,
      "step": 30380
    },
    {
      "epoch": 0.6915777256900986,
      "grad_norm": 472.0,
      "learning_rate": 6.294817016533532e-06,
      "loss": 0.603,
      "step": 30390
    },
    {
      "epoch": 0.6918052932207632,
      "grad_norm": 48.0,
      "learning_rate": 6.29017276611555e-06,
      "loss": 0.706,
      "step": 30400
    },
    {
      "epoch": 0.692032860751428,
      "grad_norm": 0.00176239013671875,
      "learning_rate": 6.285528515697567e-06,
      "loss": 0.2772,
      "step": 30410
    },
    {
      "epoch": 0.6922604282820927,
      "grad_norm": 138.0,
      "learning_rate": 6.280884265279585e-06,
      "loss": 1.1302,
      "step": 30420
    },
    {
      "epoch": 0.6924879958127574,
      "grad_norm": 398.0,
      "learning_rate": 6.276240014861602e-06,
      "loss": 0.9018,
      "step": 30430
    },
    {
      "epoch": 0.6927155633434222,
      "grad_norm": 201.0,
      "learning_rate": 6.27159576444362e-06,
      "loss": 1.1761,
      "step": 30440
    },
    {
      "epoch": 0.6929431308740869,
      "grad_norm": 66.0,
      "learning_rate": 6.266951514025637e-06,
      "loss": 1.1284,
      "step": 30450
    },
    {
      "epoch": 0.6931706984047517,
      "grad_norm": 0.00018787384033203125,
      "learning_rate": 6.262307263607655e-06,
      "loss": 0.2196,
      "step": 30460
    },
    {
      "epoch": 0.6933982659354163,
      "grad_norm": 0.000396728515625,
      "learning_rate": 6.257663013189672e-06,
      "loss": 0.837,
      "step": 30470
    },
    {
      "epoch": 0.6936258334660811,
      "grad_norm": 0.012939453125,
      "learning_rate": 6.253018762771689e-06,
      "loss": 1.163,
      "step": 30480
    },
    {
      "epoch": 0.6938534009967458,
      "grad_norm": 227.0,
      "learning_rate": 6.248374512353707e-06,
      "loss": 1.0812,
      "step": 30490
    },
    {
      "epoch": 0.6940809685274105,
      "grad_norm": 0.0026092529296875,
      "learning_rate": 6.2437302619357245e-06,
      "loss": 0.4629,
      "step": 30500
    },
    {
      "epoch": 0.6943085360580752,
      "grad_norm": 0.0004291534423828125,
      "learning_rate": 6.239086011517742e-06,
      "loss": 1.4385,
      "step": 30510
    },
    {
      "epoch": 0.69453610358874,
      "grad_norm": 161.0,
      "learning_rate": 6.234441761099759e-06,
      "loss": 0.5595,
      "step": 30520
    },
    {
      "epoch": 0.6947636711194047,
      "grad_norm": 181.0,
      "learning_rate": 6.2297975106817765e-06,
      "loss": 0.2773,
      "step": 30530
    },
    {
      "epoch": 0.6949912386500694,
      "grad_norm": 52.0,
      "learning_rate": 6.225153260263794e-06,
      "loss": 1.1629,
      "step": 30540
    },
    {
      "epoch": 0.6952188061807342,
      "grad_norm": 0.921875,
      "learning_rate": 6.220509009845811e-06,
      "loss": 0.9512,
      "step": 30550
    },
    {
      "epoch": 0.6954463737113988,
      "grad_norm": 68.5,
      "learning_rate": 6.2158647594278285e-06,
      "loss": 0.8913,
      "step": 30560
    },
    {
      "epoch": 0.6956739412420636,
      "grad_norm": 103.0,
      "learning_rate": 6.2112205090098464e-06,
      "loss": 0.5669,
      "step": 30570
    },
    {
      "epoch": 0.6959015087727283,
      "grad_norm": 0.1552734375,
      "learning_rate": 6.206576258591864e-06,
      "loss": 1.1229,
      "step": 30580
    },
    {
      "epoch": 0.696129076303393,
      "grad_norm": 34.5,
      "learning_rate": 6.201932008173881e-06,
      "loss": 0.775,
      "step": 30590
    },
    {
      "epoch": 0.6963566438340577,
      "grad_norm": 164.0,
      "learning_rate": 6.1972877577558984e-06,
      "loss": 0.8508,
      "step": 30600
    },
    {
      "epoch": 0.6965842113647225,
      "grad_norm": 39.5,
      "learning_rate": 6.192643507337916e-06,
      "loss": 0.4268,
      "step": 30610
    },
    {
      "epoch": 0.6968117788953873,
      "grad_norm": 232.0,
      "learning_rate": 6.187999256919933e-06,
      "loss": 1.3293,
      "step": 30620
    },
    {
      "epoch": 0.6970393464260519,
      "grad_norm": 220.0,
      "learning_rate": 6.183355006501951e-06,
      "loss": 0.8973,
      "step": 30630
    },
    {
      "epoch": 0.6972669139567167,
      "grad_norm": 0.01416015625,
      "learning_rate": 6.178710756083968e-06,
      "loss": 0.1674,
      "step": 30640
    },
    {
      "epoch": 0.6974944814873814,
      "grad_norm": 149.0,
      "learning_rate": 6.174066505665986e-06,
      "loss": 0.3363,
      "step": 30650
    },
    {
      "epoch": 0.6977220490180461,
      "grad_norm": 140.0,
      "learning_rate": 6.169422255248003e-06,
      "loss": 1.1134,
      "step": 30660
    },
    {
      "epoch": 0.6979496165487108,
      "grad_norm": 0.0009765625,
      "learning_rate": 6.164778004830021e-06,
      "loss": 0.8335,
      "step": 30670
    },
    {
      "epoch": 0.6981771840793756,
      "grad_norm": 290.0,
      "learning_rate": 6.160133754412039e-06,
      "loss": 0.6245,
      "step": 30680
    },
    {
      "epoch": 0.6984047516100402,
      "grad_norm": 4.625,
      "learning_rate": 6.155489503994055e-06,
      "loss": 1.3146,
      "step": 30690
    },
    {
      "epoch": 0.698632319140705,
      "grad_norm": 16.625,
      "learning_rate": 6.150845253576073e-06,
      "loss": 0.3938,
      "step": 30700
    },
    {
      "epoch": 0.6988598866713698,
      "grad_norm": 0.00372314453125,
      "learning_rate": 6.146201003158091e-06,
      "loss": 1.1865,
      "step": 30710
    },
    {
      "epoch": 0.6990874542020344,
      "grad_norm": 0.51953125,
      "learning_rate": 6.141556752740109e-06,
      "loss": 0.7717,
      "step": 30720
    },
    {
      "epoch": 0.6993150217326992,
      "grad_norm": 73.0,
      "learning_rate": 6.136912502322125e-06,
      "loss": 0.9626,
      "step": 30730
    },
    {
      "epoch": 0.6995425892633639,
      "grad_norm": 16.5,
      "learning_rate": 6.132268251904143e-06,
      "loss": 1.1349,
      "step": 30740
    },
    {
      "epoch": 0.6997701567940287,
      "grad_norm": 338.0,
      "learning_rate": 6.127624001486161e-06,
      "loss": 1.0597,
      "step": 30750
    },
    {
      "epoch": 0.6999977243246933,
      "grad_norm": 16.375,
      "learning_rate": 6.122979751068177e-06,
      "loss": 0.5158,
      "step": 30760
    },
    {
      "epoch": 0.7002252918553581,
      "grad_norm": 0.115234375,
      "learning_rate": 6.118335500650195e-06,
      "loss": 0.7253,
      "step": 30770
    },
    {
      "epoch": 0.7004528593860228,
      "grad_norm": 290.0,
      "learning_rate": 6.113691250232213e-06,
      "loss": 0.7678,
      "step": 30780
    },
    {
      "epoch": 0.7006804269166875,
      "grad_norm": 146.0,
      "learning_rate": 6.109046999814231e-06,
      "loss": 0.3002,
      "step": 30790
    },
    {
      "epoch": 0.7009079944473523,
      "grad_norm": 2.34375,
      "learning_rate": 6.104402749396248e-06,
      "loss": 1.1036,
      "step": 30800
    },
    {
      "epoch": 0.701135561978017,
      "grad_norm": 148.0,
      "learning_rate": 6.099758498978265e-06,
      "loss": 0.801,
      "step": 30810
    },
    {
      "epoch": 0.7013631295086817,
      "grad_norm": 0.478515625,
      "learning_rate": 6.095114248560283e-06,
      "loss": 0.4756,
      "step": 30820
    },
    {
      "epoch": 0.7015906970393464,
      "grad_norm": 31.75,
      "learning_rate": 6.0904699981423e-06,
      "loss": 1.7041,
      "step": 30830
    },
    {
      "epoch": 0.7018182645700112,
      "grad_norm": 0.0003337860107421875,
      "learning_rate": 6.085825747724318e-06,
      "loss": 0.7673,
      "step": 30840
    },
    {
      "epoch": 0.7020458321006758,
      "grad_norm": 0.0014495849609375,
      "learning_rate": 6.081181497306336e-06,
      "loss": 1.1488,
      "step": 30850
    },
    {
      "epoch": 0.7022733996313406,
      "grad_norm": 0.002838134765625,
      "learning_rate": 6.076537246888353e-06,
      "loss": 0.2106,
      "step": 30860
    },
    {
      "epoch": 0.7025009671620053,
      "grad_norm": 245.0,
      "learning_rate": 6.07189299647037e-06,
      "loss": 0.8626,
      "step": 30870
    },
    {
      "epoch": 0.70272853469267,
      "grad_norm": 96.0,
      "learning_rate": 6.067248746052388e-06,
      "loss": 0.9122,
      "step": 30880
    },
    {
      "epoch": 0.7029561022233348,
      "grad_norm": 0.00213623046875,
      "learning_rate": 6.062604495634406e-06,
      "loss": 0.7019,
      "step": 30890
    },
    {
      "epoch": 0.7031836697539995,
      "grad_norm": 81.5,
      "learning_rate": 6.057960245216422e-06,
      "loss": 0.4238,
      "step": 30900
    },
    {
      "epoch": 0.7034112372846643,
      "grad_norm": 0.0184326171875,
      "learning_rate": 6.05331599479844e-06,
      "loss": 0.3054,
      "step": 30910
    },
    {
      "epoch": 0.7036388048153289,
      "grad_norm": 88.5,
      "learning_rate": 6.048671744380458e-06,
      "loss": 1.6876,
      "step": 30920
    },
    {
      "epoch": 0.7038663723459937,
      "grad_norm": 38.75,
      "learning_rate": 6.044027493962476e-06,
      "loss": 0.3755,
      "step": 30930
    },
    {
      "epoch": 0.7040939398766584,
      "grad_norm": 72.5,
      "learning_rate": 6.039383243544492e-06,
      "loss": 1.0602,
      "step": 30940
    },
    {
      "epoch": 0.7043215074073231,
      "grad_norm": 138.0,
      "learning_rate": 6.03473899312651e-06,
      "loss": 1.0472,
      "step": 30950
    },
    {
      "epoch": 0.7045490749379878,
      "grad_norm": 0.02099609375,
      "learning_rate": 6.030094742708528e-06,
      "loss": 0.3505,
      "step": 30960
    },
    {
      "epoch": 0.7047766424686526,
      "grad_norm": 121.5,
      "learning_rate": 6.025450492290545e-06,
      "loss": 0.9165,
      "step": 30970
    },
    {
      "epoch": 0.7050042099993173,
      "grad_norm": 0.00130462646484375,
      "learning_rate": 6.020806241872562e-06,
      "loss": 0.6073,
      "step": 30980
    },
    {
      "epoch": 0.705231777529982,
      "grad_norm": 2.03125,
      "learning_rate": 6.01616199145458e-06,
      "loss": 1.0795,
      "step": 30990
    },
    {
      "epoch": 0.7054593450606468,
      "grad_norm": 139.0,
      "learning_rate": 6.011517741036598e-06,
      "loss": 1.1143,
      "step": 31000
    },
    {
      "epoch": 0.7056869125913114,
      "grad_norm": 0.36328125,
      "learning_rate": 6.006873490618615e-06,
      "loss": 0.0494,
      "step": 31010
    },
    {
      "epoch": 0.7059144801219762,
      "grad_norm": 21.25,
      "learning_rate": 6.0022292402006325e-06,
      "loss": 1.1029,
      "step": 31020
    },
    {
      "epoch": 0.7061420476526409,
      "grad_norm": 680.0,
      "learning_rate": 5.99758498978265e-06,
      "loss": 0.5868,
      "step": 31030
    },
    {
      "epoch": 0.7063696151833057,
      "grad_norm": 109.5,
      "learning_rate": 5.992940739364667e-06,
      "loss": 1.2648,
      "step": 31040
    },
    {
      "epoch": 0.7065971827139703,
      "grad_norm": 115.5,
      "learning_rate": 5.9882964889466846e-06,
      "loss": 0.6155,
      "step": 31050
    },
    {
      "epoch": 0.7068247502446351,
      "grad_norm": 73.5,
      "learning_rate": 5.9836522385287025e-06,
      "loss": 0.7303,
      "step": 31060
    },
    {
      "epoch": 0.7070523177752999,
      "grad_norm": 163.0,
      "learning_rate": 5.9790079881107195e-06,
      "loss": 0.5064,
      "step": 31070
    },
    {
      "epoch": 0.7072798853059645,
      "grad_norm": 159.0,
      "learning_rate": 5.9743637376927366e-06,
      "loss": 1.2467,
      "step": 31080
    },
    {
      "epoch": 0.7075074528366293,
      "grad_norm": 228.0,
      "learning_rate": 5.9697194872747545e-06,
      "loss": 1.5694,
      "step": 31090
    },
    {
      "epoch": 0.707735020367294,
      "grad_norm": 4.71875,
      "learning_rate": 5.965075236856772e-06,
      "loss": 0.4212,
      "step": 31100
    },
    {
      "epoch": 0.7079625878979587,
      "grad_norm": 174.0,
      "learning_rate": 5.960430986438789e-06,
      "loss": 0.4361,
      "step": 31110
    },
    {
      "epoch": 0.7081901554286234,
      "grad_norm": 18.0,
      "learning_rate": 5.9557867360208065e-06,
      "loss": 1.3537,
      "step": 31120
    },
    {
      "epoch": 0.7084177229592882,
      "grad_norm": 185.0,
      "learning_rate": 5.951142485602824e-06,
      "loss": 0.4701,
      "step": 31130
    },
    {
      "epoch": 0.7086452904899528,
      "grad_norm": 59.25,
      "learning_rate": 5.946498235184842e-06,
      "loss": 0.8827,
      "step": 31140
    },
    {
      "epoch": 0.7088728580206176,
      "grad_norm": 1.0859375,
      "learning_rate": 5.9418539847668585e-06,
      "loss": 0.7444,
      "step": 31150
    },
    {
      "epoch": 0.7091004255512824,
      "grad_norm": 98.5,
      "learning_rate": 5.937209734348876e-06,
      "loss": 0.9574,
      "step": 31160
    },
    {
      "epoch": 0.709327993081947,
      "grad_norm": 76.0,
      "learning_rate": 5.932565483930894e-06,
      "loss": 0.5572,
      "step": 31170
    },
    {
      "epoch": 0.7095555606126118,
      "grad_norm": 47.25,
      "learning_rate": 5.927921233512911e-06,
      "loss": 0.5339,
      "step": 31180
    },
    {
      "epoch": 0.7097831281432765,
      "grad_norm": 60.5,
      "learning_rate": 5.923276983094928e-06,
      "loss": 0.6078,
      "step": 31190
    },
    {
      "epoch": 0.7100106956739413,
      "grad_norm": 217.0,
      "learning_rate": 5.918632732676946e-06,
      "loss": 0.435,
      "step": 31200
    },
    {
      "epoch": 0.7102382632046059,
      "grad_norm": 24.375,
      "learning_rate": 5.913988482258964e-06,
      "loss": 0.721,
      "step": 31210
    },
    {
      "epoch": 0.7104658307352707,
      "grad_norm": 78.0,
      "learning_rate": 5.909344231840981e-06,
      "loss": 0.2426,
      "step": 31220
    },
    {
      "epoch": 0.7106933982659354,
      "grad_norm": 146.0,
      "learning_rate": 5.904699981422999e-06,
      "loss": 0.5741,
      "step": 31230
    },
    {
      "epoch": 0.7109209657966001,
      "grad_norm": 0.09814453125,
      "learning_rate": 5.900055731005016e-06,
      "loss": 0.6006,
      "step": 31240
    },
    {
      "epoch": 0.7111485333272649,
      "grad_norm": 3.296875,
      "learning_rate": 5.895411480587033e-06,
      "loss": 0.7801,
      "step": 31250
    },
    {
      "epoch": 0.7113761008579296,
      "grad_norm": 100.5,
      "learning_rate": 5.890767230169051e-06,
      "loss": 0.7449,
      "step": 31260
    },
    {
      "epoch": 0.7116036683885943,
      "grad_norm": 161.0,
      "learning_rate": 5.886122979751069e-06,
      "loss": 0.9067,
      "step": 31270
    },
    {
      "epoch": 0.711831235919259,
      "grad_norm": 77.0,
      "learning_rate": 5.881478729333087e-06,
      "loss": 1.1141,
      "step": 31280
    },
    {
      "epoch": 0.7120588034499238,
      "grad_norm": 8.25,
      "learning_rate": 5.876834478915103e-06,
      "loss": 0.5541,
      "step": 31290
    },
    {
      "epoch": 0.7122863709805884,
      "grad_norm": 81.5,
      "learning_rate": 5.872190228497121e-06,
      "loss": 0.3083,
      "step": 31300
    },
    {
      "epoch": 0.7125139385112532,
      "grad_norm": 0.050048828125,
      "learning_rate": 5.867545978079139e-06,
      "loss": 0.4452,
      "step": 31310
    },
    {
      "epoch": 0.712741506041918,
      "grad_norm": 171.0,
      "learning_rate": 5.862901727661155e-06,
      "loss": 0.6841,
      "step": 31320
    },
    {
      "epoch": 0.7129690735725827,
      "grad_norm": 0.0036773681640625,
      "learning_rate": 5.858257477243173e-06,
      "loss": 1.1581,
      "step": 31330
    },
    {
      "epoch": 0.7131966411032474,
      "grad_norm": 43.5,
      "learning_rate": 5.853613226825191e-06,
      "loss": 1.094,
      "step": 31340
    },
    {
      "epoch": 0.7134242086339121,
      "grad_norm": 0.71875,
      "learning_rate": 5.848968976407209e-06,
      "loss": 0.7785,
      "step": 31350
    },
    {
      "epoch": 0.7136517761645769,
      "grad_norm": 113.0,
      "learning_rate": 5.844324725989225e-06,
      "loss": 1.1154,
      "step": 31360
    },
    {
      "epoch": 0.7138793436952415,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.839680475571243e-06,
      "loss": 0.8071,
      "step": 31370
    },
    {
      "epoch": 0.7141069112259063,
      "grad_norm": 374.0,
      "learning_rate": 5.835036225153261e-06,
      "loss": 1.7172,
      "step": 31380
    },
    {
      "epoch": 0.714334478756571,
      "grad_norm": 41.0,
      "learning_rate": 5.830391974735278e-06,
      "loss": 0.2336,
      "step": 31390
    },
    {
      "epoch": 0.7145620462872357,
      "grad_norm": 294.0,
      "learning_rate": 5.825747724317296e-06,
      "loss": 2.3095,
      "step": 31400
    },
    {
      "epoch": 0.7147896138179005,
      "grad_norm": 145.0,
      "learning_rate": 5.821103473899313e-06,
      "loss": 0.9765,
      "step": 31410
    },
    {
      "epoch": 0.7150171813485652,
      "grad_norm": 72.5,
      "learning_rate": 5.816459223481331e-06,
      "loss": 0.4695,
      "step": 31420
    },
    {
      "epoch": 0.71524474887923,
      "grad_norm": 0.0025177001953125,
      "learning_rate": 5.811814973063348e-06,
      "loss": 0.3084,
      "step": 31430
    },
    {
      "epoch": 0.7154723164098946,
      "grad_norm": 0.1396484375,
      "learning_rate": 5.807170722645366e-06,
      "loss": 0.7108,
      "step": 31440
    },
    {
      "epoch": 0.7156998839405594,
      "grad_norm": 161.0,
      "learning_rate": 5.802526472227384e-06,
      "loss": 0.707,
      "step": 31450
    },
    {
      "epoch": 0.715927451471224,
      "grad_norm": 228.0,
      "learning_rate": 5.7978822218094e-06,
      "loss": 0.9455,
      "step": 31460
    },
    {
      "epoch": 0.7161550190018888,
      "grad_norm": 0.01336669921875,
      "learning_rate": 5.793237971391418e-06,
      "loss": 0.4935,
      "step": 31470
    },
    {
      "epoch": 0.7163825865325535,
      "grad_norm": 145.0,
      "learning_rate": 5.788593720973436e-06,
      "loss": 1.5777,
      "step": 31480
    },
    {
      "epoch": 0.7166101540632183,
      "grad_norm": 9.9375,
      "learning_rate": 5.783949470555454e-06,
      "loss": 0.4491,
      "step": 31490
    },
    {
      "epoch": 0.716837721593883,
      "grad_norm": 116.5,
      "learning_rate": 5.77930522013747e-06,
      "loss": 1.0652,
      "step": 31500
    },
    {
      "epoch": 0.7170652891245477,
      "grad_norm": 44.25,
      "learning_rate": 5.774660969719488e-06,
      "loss": 0.9927,
      "step": 31510
    },
    {
      "epoch": 0.7172928566552125,
      "grad_norm": 0.0025482177734375,
      "learning_rate": 5.770016719301506e-06,
      "loss": 0.1149,
      "step": 31520
    },
    {
      "epoch": 0.7175204241858771,
      "grad_norm": 173.0,
      "learning_rate": 5.765372468883522e-06,
      "loss": 1.3964,
      "step": 31530
    },
    {
      "epoch": 0.7177479917165419,
      "grad_norm": 0.034912109375,
      "learning_rate": 5.76072821846554e-06,
      "loss": 0.9465,
      "step": 31540
    },
    {
      "epoch": 0.7179755592472066,
      "grad_norm": 115.5,
      "learning_rate": 5.756083968047558e-06,
      "loss": 0.6977,
      "step": 31550
    },
    {
      "epoch": 0.7182031267778713,
      "grad_norm": 0.376953125,
      "learning_rate": 5.7514397176295755e-06,
      "loss": 0.3787,
      "step": 31560
    },
    {
      "epoch": 0.718430694308536,
      "grad_norm": 37.0,
      "learning_rate": 5.746795467211593e-06,
      "loss": 0.9214,
      "step": 31570
    },
    {
      "epoch": 0.7186582618392008,
      "grad_norm": 87.0,
      "learning_rate": 5.74215121679361e-06,
      "loss": 0.5925,
      "step": 31580
    },
    {
      "epoch": 0.7188858293698656,
      "grad_norm": 75.5,
      "learning_rate": 5.7375069663756275e-06,
      "loss": 0.4311,
      "step": 31590
    },
    {
      "epoch": 0.7191133969005302,
      "grad_norm": 131.0,
      "learning_rate": 5.732862715957645e-06,
      "loss": 1.9153,
      "step": 31600
    },
    {
      "epoch": 0.719340964431195,
      "grad_norm": 0.007659912109375,
      "learning_rate": 5.7282184655396625e-06,
      "loss": 0.8574,
      "step": 31610
    },
    {
      "epoch": 0.7195685319618597,
      "grad_norm": 242.0,
      "learning_rate": 5.72357421512168e-06,
      "loss": 1.5322,
      "step": 31620
    },
    {
      "epoch": 0.7197960994925244,
      "grad_norm": 340.0,
      "learning_rate": 5.7189299647036975e-06,
      "loss": 1.0505,
      "step": 31630
    },
    {
      "epoch": 0.7200236670231891,
      "grad_norm": 103.5,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.8903,
      "step": 31640
    },
    {
      "epoch": 0.7202512345538539,
      "grad_norm": 0.00347900390625,
      "learning_rate": 5.709641463867732e-06,
      "loss": 1.0776,
      "step": 31650
    },
    {
      "epoch": 0.7204788020845185,
      "grad_norm": 0.00213623046875,
      "learning_rate": 5.70499721344975e-06,
      "loss": 0.3684,
      "step": 31660
    },
    {
      "epoch": 0.7207063696151833,
      "grad_norm": 52.75,
      "learning_rate": 5.7003529630317665e-06,
      "loss": 0.3727,
      "step": 31670
    },
    {
      "epoch": 0.7209339371458481,
      "grad_norm": 39.5,
      "learning_rate": 5.6957087126137844e-06,
      "loss": 1.0637,
      "step": 31680
    },
    {
      "epoch": 0.7211615046765127,
      "grad_norm": 0.146484375,
      "learning_rate": 5.691064462195802e-06,
      "loss": 1.0417,
      "step": 31690
    },
    {
      "epoch": 0.7213890722071775,
      "grad_norm": 76.0,
      "learning_rate": 5.68642021177782e-06,
      "loss": 0.9001,
      "step": 31700
    },
    {
      "epoch": 0.7216166397378422,
      "grad_norm": 164.0,
      "learning_rate": 5.6817759613598364e-06,
      "loss": 0.4436,
      "step": 31710
    },
    {
      "epoch": 0.721844207268507,
      "grad_norm": 0.2470703125,
      "learning_rate": 5.677131710941854e-06,
      "loss": 1.2977,
      "step": 31720
    },
    {
      "epoch": 0.7220717747991716,
      "grad_norm": 253.0,
      "learning_rate": 5.672487460523872e-06,
      "loss": 1.0016,
      "step": 31730
    },
    {
      "epoch": 0.7222993423298364,
      "grad_norm": 0.003204345703125,
      "learning_rate": 5.667843210105889e-06,
      "loss": 1.0249,
      "step": 31740
    },
    {
      "epoch": 0.722526909860501,
      "grad_norm": 201.0,
      "learning_rate": 5.663198959687906e-06,
      "loss": 0.7059,
      "step": 31750
    },
    {
      "epoch": 0.7227544773911658,
      "grad_norm": 728.0,
      "learning_rate": 5.658554709269924e-06,
      "loss": 1.3763,
      "step": 31760
    },
    {
      "epoch": 0.7229820449218306,
      "grad_norm": 0.062255859375,
      "learning_rate": 5.653910458851942e-06,
      "loss": 1.6671,
      "step": 31770
    },
    {
      "epoch": 0.7232096124524953,
      "grad_norm": 0.00040435791015625,
      "learning_rate": 5.649266208433959e-06,
      "loss": 0.4813,
      "step": 31780
    },
    {
      "epoch": 0.72343717998316,
      "grad_norm": 23.5,
      "learning_rate": 5.644621958015976e-06,
      "loss": 0.474,
      "step": 31790
    },
    {
      "epoch": 0.7236647475138247,
      "grad_norm": 0.99609375,
      "learning_rate": 5.639977707597994e-06,
      "loss": 1.1278,
      "step": 31800
    },
    {
      "epoch": 0.7238923150444895,
      "grad_norm": 221.0,
      "learning_rate": 5.635333457180011e-06,
      "loss": 1.5057,
      "step": 31810
    },
    {
      "epoch": 0.7241198825751541,
      "grad_norm": 48.25,
      "learning_rate": 5.630689206762029e-06,
      "loss": 1.008,
      "step": 31820
    },
    {
      "epoch": 0.7243474501058189,
      "grad_norm": 116.5,
      "learning_rate": 5.626044956344047e-06,
      "loss": 0.5513,
      "step": 31830
    },
    {
      "epoch": 0.7245750176364836,
      "grad_norm": 0.000400543212890625,
      "learning_rate": 5.621400705926064e-06,
      "loss": 0.3141,
      "step": 31840
    },
    {
      "epoch": 0.7248025851671483,
      "grad_norm": 64.5,
      "learning_rate": 5.616756455508081e-06,
      "loss": 1.5741,
      "step": 31850
    },
    {
      "epoch": 0.7250301526978131,
      "grad_norm": 104.0,
      "learning_rate": 5.612112205090099e-06,
      "loss": 0.8305,
      "step": 31860
    },
    {
      "epoch": 0.7252577202284778,
      "grad_norm": 0.021728515625,
      "learning_rate": 5.607467954672117e-06,
      "loss": 0.254,
      "step": 31870
    },
    {
      "epoch": 0.7254852877591426,
      "grad_norm": 308.0,
      "learning_rate": 5.602823704254133e-06,
      "loss": 0.692,
      "step": 31880
    },
    {
      "epoch": 0.7257128552898072,
      "grad_norm": 105.0,
      "learning_rate": 5.598179453836151e-06,
      "loss": 0.6012,
      "step": 31890
    },
    {
      "epoch": 0.725940422820472,
      "grad_norm": 33.0,
      "learning_rate": 5.593535203418169e-06,
      "loss": 0.9286,
      "step": 31900
    },
    {
      "epoch": 0.7261679903511367,
      "grad_norm": 0.0155029296875,
      "learning_rate": 5.588890953000187e-06,
      "loss": 0.8108,
      "step": 31910
    },
    {
      "epoch": 0.7263955578818014,
      "grad_norm": 0.201171875,
      "learning_rate": 5.584246702582203e-06,
      "loss": 0.8751,
      "step": 31920
    },
    {
      "epoch": 0.7266231254124661,
      "grad_norm": 0.107421875,
      "learning_rate": 5.579602452164221e-06,
      "loss": 2.6684,
      "step": 31930
    },
    {
      "epoch": 0.7268506929431309,
      "grad_norm": 0.007659912109375,
      "learning_rate": 5.574958201746239e-06,
      "loss": 0.6196,
      "step": 31940
    },
    {
      "epoch": 0.7270782604737956,
      "grad_norm": 247.0,
      "learning_rate": 5.570313951328256e-06,
      "loss": 1.0708,
      "step": 31950
    },
    {
      "epoch": 0.7273058280044603,
      "grad_norm": 0.0040283203125,
      "learning_rate": 5.565669700910273e-06,
      "loss": 0.2569,
      "step": 31960
    },
    {
      "epoch": 0.7275333955351251,
      "grad_norm": 71.5,
      "learning_rate": 5.561025450492291e-06,
      "loss": 0.4322,
      "step": 31970
    },
    {
      "epoch": 0.7277609630657897,
      "grad_norm": 183.0,
      "learning_rate": 5.556381200074309e-06,
      "loss": 0.7901,
      "step": 31980
    },
    {
      "epoch": 0.7279885305964545,
      "grad_norm": 368.0,
      "learning_rate": 5.551736949656326e-06,
      "loss": 0.9497,
      "step": 31990
    },
    {
      "epoch": 0.7282160981271192,
      "grad_norm": 166.0,
      "learning_rate": 5.547092699238344e-06,
      "loss": 1.1271,
      "step": 32000
    },
    {
      "epoch": 0.728443665657784,
      "grad_norm": 13.3125,
      "learning_rate": 5.542448448820361e-06,
      "loss": 0.2628,
      "step": 32010
    },
    {
      "epoch": 0.7286712331884486,
      "grad_norm": 98.5,
      "learning_rate": 5.537804198402378e-06,
      "loss": 1.0449,
      "step": 32020
    },
    {
      "epoch": 0.7288988007191134,
      "grad_norm": 2.375,
      "learning_rate": 5.533159947984396e-06,
      "loss": 0.9997,
      "step": 32030
    },
    {
      "epoch": 0.7291263682497782,
      "grad_norm": 94.5,
      "learning_rate": 5.528515697566414e-06,
      "loss": 0.9473,
      "step": 32040
    },
    {
      "epoch": 0.7293539357804428,
      "grad_norm": 154.0,
      "learning_rate": 5.5238714471484316e-06,
      "loss": 1.2952,
      "step": 32050
    },
    {
      "epoch": 0.7295815033111076,
      "grad_norm": 0.010498046875,
      "learning_rate": 5.519227196730448e-06,
      "loss": 1.7574,
      "step": 32060
    },
    {
      "epoch": 0.7298090708417723,
      "grad_norm": 1.1484375,
      "learning_rate": 5.514582946312466e-06,
      "loss": 1.4025,
      "step": 32070
    },
    {
      "epoch": 0.730036638372437,
      "grad_norm": 224.0,
      "learning_rate": 5.5099386958944836e-06,
      "loss": 1.054,
      "step": 32080
    },
    {
      "epoch": 0.7302642059031017,
      "grad_norm": 89.5,
      "learning_rate": 5.5052944454765e-06,
      "loss": 0.471,
      "step": 32090
    },
    {
      "epoch": 0.7304917734337665,
      "grad_norm": 51.5,
      "learning_rate": 5.500650195058518e-06,
      "loss": 0.9985,
      "step": 32100
    },
    {
      "epoch": 0.7307193409644311,
      "grad_norm": 8.5,
      "learning_rate": 5.4960059446405356e-06,
      "loss": 0.9002,
      "step": 32110
    },
    {
      "epoch": 0.7309469084950959,
      "grad_norm": 3.859375,
      "learning_rate": 5.4913616942225535e-06,
      "loss": 0.9282,
      "step": 32120
    },
    {
      "epoch": 0.7311744760257607,
      "grad_norm": 232.0,
      "learning_rate": 5.48671744380457e-06,
      "loss": 0.916,
      "step": 32130
    },
    {
      "epoch": 0.7314020435564254,
      "grad_norm": 108.0,
      "learning_rate": 5.482073193386588e-06,
      "loss": 0.5656,
      "step": 32140
    },
    {
      "epoch": 0.7316296110870901,
      "grad_norm": 46.75,
      "learning_rate": 5.4774289429686055e-06,
      "loss": 1.0033,
      "step": 32150
    },
    {
      "epoch": 0.7318571786177548,
      "grad_norm": 61.5,
      "learning_rate": 5.4727846925506225e-06,
      "loss": 0.5867,
      "step": 32160
    },
    {
      "epoch": 0.7320847461484196,
      "grad_norm": 0.00469970703125,
      "learning_rate": 5.4681404421326404e-06,
      "loss": 0.4891,
      "step": 32170
    },
    {
      "epoch": 0.7323123136790842,
      "grad_norm": 51.5,
      "learning_rate": 5.4634961917146575e-06,
      "loss": 0.7372,
      "step": 32180
    },
    {
      "epoch": 0.732539881209749,
      "grad_norm": 185.0,
      "learning_rate": 5.458851941296675e-06,
      "loss": 0.9012,
      "step": 32190
    },
    {
      "epoch": 0.7327674487404137,
      "grad_norm": 179.0,
      "learning_rate": 5.4542076908786925e-06,
      "loss": 0.9028,
      "step": 32200
    },
    {
      "epoch": 0.7329950162710784,
      "grad_norm": 0.0007171630859375,
      "learning_rate": 5.44956344046071e-06,
      "loss": 0.7397,
      "step": 32210
    },
    {
      "epoch": 0.7332225838017432,
      "grad_norm": 87.5,
      "learning_rate": 5.444919190042728e-06,
      "loss": 0.8098,
      "step": 32220
    },
    {
      "epoch": 0.7334501513324079,
      "grad_norm": 0.306640625,
      "learning_rate": 5.4402749396247445e-06,
      "loss": 0.9793,
      "step": 32230
    },
    {
      "epoch": 0.7336777188630726,
      "grad_norm": 0.0087890625,
      "learning_rate": 5.435630689206762e-06,
      "loss": 0.4178,
      "step": 32240
    },
    {
      "epoch": 0.7339052863937373,
      "grad_norm": 0.00958251953125,
      "learning_rate": 5.43098643878878e-06,
      "loss": 0.3515,
      "step": 32250
    },
    {
      "epoch": 0.7341328539244021,
      "grad_norm": 262.0,
      "learning_rate": 5.426342188370798e-06,
      "loss": 1.0374,
      "step": 32260
    },
    {
      "epoch": 0.7343604214550667,
      "grad_norm": 1.59375,
      "learning_rate": 5.421697937952814e-06,
      "loss": 0.4201,
      "step": 32270
    },
    {
      "epoch": 0.7345879889857315,
      "grad_norm": 164.0,
      "learning_rate": 5.417053687534832e-06,
      "loss": 1.2203,
      "step": 32280
    },
    {
      "epoch": 0.7348155565163963,
      "grad_norm": 428.0,
      "learning_rate": 5.41240943711685e-06,
      "loss": 1.3076,
      "step": 32290
    },
    {
      "epoch": 0.735043124047061,
      "grad_norm": 0.1201171875,
      "learning_rate": 5.407765186698866e-06,
      "loss": 0.6184,
      "step": 32300
    },
    {
      "epoch": 0.7352706915777257,
      "grad_norm": 0.006195068359375,
      "learning_rate": 5.403120936280884e-06,
      "loss": 0.5561,
      "step": 32310
    },
    {
      "epoch": 0.7354982591083904,
      "grad_norm": 119.0,
      "learning_rate": 5.398476685862902e-06,
      "loss": 0.403,
      "step": 32320
    },
    {
      "epoch": 0.7357258266390552,
      "grad_norm": 33.75,
      "learning_rate": 5.39383243544492e-06,
      "loss": 0.548,
      "step": 32330
    },
    {
      "epoch": 0.7359533941697198,
      "grad_norm": 0.103515625,
      "learning_rate": 5.389188185026937e-06,
      "loss": 0.4353,
      "step": 32340
    },
    {
      "epoch": 0.7361809617003846,
      "grad_norm": 82.5,
      "learning_rate": 5.384543934608954e-06,
      "loss": 0.5967,
      "step": 32350
    },
    {
      "epoch": 0.7364085292310493,
      "grad_norm": 0.00921630859375,
      "learning_rate": 5.379899684190972e-06,
      "loss": 0.8002,
      "step": 32360
    },
    {
      "epoch": 0.736636096761714,
      "grad_norm": 276.0,
      "learning_rate": 5.375255433772989e-06,
      "loss": 0.8929,
      "step": 32370
    },
    {
      "epoch": 0.7368636642923788,
      "grad_norm": 6.34375,
      "learning_rate": 5.370611183355007e-06,
      "loss": 0.6049,
      "step": 32380
    },
    {
      "epoch": 0.7370912318230435,
      "grad_norm": 162.0,
      "learning_rate": 5.365966932937024e-06,
      "loss": 0.4219,
      "step": 32390
    },
    {
      "epoch": 0.7373187993537083,
      "grad_norm": 52.5,
      "learning_rate": 5.361322682519042e-06,
      "loss": 0.5342,
      "step": 32400
    },
    {
      "epoch": 0.7375463668843729,
      "grad_norm": 0.0155029296875,
      "learning_rate": 5.356678432101059e-06,
      "loss": 0.0915,
      "step": 32410
    },
    {
      "epoch": 0.7377739344150377,
      "grad_norm": 302.0,
      "learning_rate": 5.352034181683077e-06,
      "loss": 1.2751,
      "step": 32420
    },
    {
      "epoch": 0.7380015019457024,
      "grad_norm": 0.000629425048828125,
      "learning_rate": 5.347389931265095e-06,
      "loss": 1.1275,
      "step": 32430
    },
    {
      "epoch": 0.7382290694763671,
      "grad_norm": 40.75,
      "learning_rate": 5.342745680847111e-06,
      "loss": 0.5124,
      "step": 32440
    },
    {
      "epoch": 0.7384566370070318,
      "grad_norm": 134.0,
      "learning_rate": 5.338101430429129e-06,
      "loss": 0.8746,
      "step": 32450
    },
    {
      "epoch": 0.7386842045376966,
      "grad_norm": 200.0,
      "learning_rate": 5.333457180011147e-06,
      "loss": 0.8715,
      "step": 32460
    },
    {
      "epoch": 0.7389117720683613,
      "grad_norm": 292.0,
      "learning_rate": 5.328812929593165e-06,
      "loss": 0.9141,
      "step": 32470
    },
    {
      "epoch": 0.739139339599026,
      "grad_norm": 100.0,
      "learning_rate": 5.324168679175181e-06,
      "loss": 1.0076,
      "step": 32480
    },
    {
      "epoch": 0.7393669071296908,
      "grad_norm": 278.0,
      "learning_rate": 5.319524428757199e-06,
      "loss": 0.5397,
      "step": 32490
    },
    {
      "epoch": 0.7395944746603554,
      "grad_norm": 55.75,
      "learning_rate": 5.314880178339217e-06,
      "loss": 0.8196,
      "step": 32500
    },
    {
      "epoch": 0.7398220421910202,
      "grad_norm": 0.00677490234375,
      "learning_rate": 5.310235927921233e-06,
      "loss": 0.9053,
      "step": 32510
    },
    {
      "epoch": 0.7400496097216849,
      "grad_norm": 122.5,
      "learning_rate": 5.305591677503251e-06,
      "loss": 0.3075,
      "step": 32520
    },
    {
      "epoch": 0.7402771772523496,
      "grad_norm": 374.0,
      "learning_rate": 5.300947427085269e-06,
      "loss": 0.8844,
      "step": 32530
    },
    {
      "epoch": 0.7405047447830143,
      "grad_norm": 0.007568359375,
      "learning_rate": 5.296303176667287e-06,
      "loss": 0.5985,
      "step": 32540
    },
    {
      "epoch": 0.7407323123136791,
      "grad_norm": 0.003936767578125,
      "learning_rate": 5.291658926249304e-06,
      "loss": 0.7109,
      "step": 32550
    },
    {
      "epoch": 0.7409598798443439,
      "grad_norm": 0.0033721923828125,
      "learning_rate": 5.287014675831321e-06,
      "loss": 0.3126,
      "step": 32560
    },
    {
      "epoch": 0.7411874473750085,
      "grad_norm": 1.53125,
      "learning_rate": 5.282370425413339e-06,
      "loss": 0.5435,
      "step": 32570
    },
    {
      "epoch": 0.7414150149056733,
      "grad_norm": 0.00421142578125,
      "learning_rate": 5.277726174995356e-06,
      "loss": 0.4986,
      "step": 32580
    },
    {
      "epoch": 0.741642582436338,
      "grad_norm": 0.0517578125,
      "learning_rate": 5.273081924577374e-06,
      "loss": 1.4821,
      "step": 32590
    },
    {
      "epoch": 0.7418701499670027,
      "grad_norm": 12.75,
      "learning_rate": 5.268437674159392e-06,
      "loss": 0.1919,
      "step": 32600
    },
    {
      "epoch": 0.7420977174976674,
      "grad_norm": 139.0,
      "learning_rate": 5.263793423741409e-06,
      "loss": 0.9076,
      "step": 32610
    },
    {
      "epoch": 0.7423252850283322,
      "grad_norm": 19.875,
      "learning_rate": 5.259149173323426e-06,
      "loss": 0.8514,
      "step": 32620
    },
    {
      "epoch": 0.7425528525589968,
      "grad_norm": 96.5,
      "learning_rate": 5.254504922905444e-06,
      "loss": 0.6062,
      "step": 32630
    },
    {
      "epoch": 0.7427804200896616,
      "grad_norm": 9.0,
      "learning_rate": 5.2498606724874615e-06,
      "loss": 0.6512,
      "step": 32640
    },
    {
      "epoch": 0.7430079876203264,
      "grad_norm": 0.03662109375,
      "learning_rate": 5.245216422069478e-06,
      "loss": 0.3303,
      "step": 32650
    },
    {
      "epoch": 0.743235555150991,
      "grad_norm": 104.5,
      "learning_rate": 5.240572171651496e-06,
      "loss": 1.6047,
      "step": 32660
    },
    {
      "epoch": 0.7434631226816558,
      "grad_norm": 0.0033721923828125,
      "learning_rate": 5.2359279212335135e-06,
      "loss": 0.8769,
      "step": 32670
    },
    {
      "epoch": 0.7436906902123205,
      "grad_norm": 290.0,
      "learning_rate": 5.231283670815531e-06,
      "loss": 0.5621,
      "step": 32680
    },
    {
      "epoch": 0.7439182577429853,
      "grad_norm": 10.9375,
      "learning_rate": 5.226639420397548e-06,
      "loss": 0.2461,
      "step": 32690
    },
    {
      "epoch": 0.7441458252736499,
      "grad_norm": 191.0,
      "learning_rate": 5.2219951699795655e-06,
      "loss": 1.0081,
      "step": 32700
    },
    {
      "epoch": 0.7443733928043147,
      "grad_norm": 0.0003108978271484375,
      "learning_rate": 5.2173509195615834e-06,
      "loss": 0.4176,
      "step": 32710
    },
    {
      "epoch": 0.7446009603349794,
      "grad_norm": 163.0,
      "learning_rate": 5.2127066691436005e-06,
      "loss": 0.5811,
      "step": 32720
    },
    {
      "epoch": 0.7448285278656441,
      "grad_norm": 0.061767578125,
      "learning_rate": 5.2080624187256175e-06,
      "loss": 1.1029,
      "step": 32730
    },
    {
      "epoch": 0.7450560953963089,
      "grad_norm": 0.0002498626708984375,
      "learning_rate": 5.2034181683076354e-06,
      "loss": 0.5179,
      "step": 32740
    },
    {
      "epoch": 0.7452836629269736,
      "grad_norm": 102.0,
      "learning_rate": 5.198773917889653e-06,
      "loss": 0.9308,
      "step": 32750
    },
    {
      "epoch": 0.7455112304576383,
      "grad_norm": 332.0,
      "learning_rate": 5.19412966747167e-06,
      "loss": 0.3359,
      "step": 32760
    },
    {
      "epoch": 0.745738797988303,
      "grad_norm": 0.11279296875,
      "learning_rate": 5.189485417053688e-06,
      "loss": 0.9387,
      "step": 32770
    },
    {
      "epoch": 0.7459663655189678,
      "grad_norm": 65.0,
      "learning_rate": 5.184841166635705e-06,
      "loss": 0.4667,
      "step": 32780
    },
    {
      "epoch": 0.7461939330496324,
      "grad_norm": 0.004302978515625,
      "learning_rate": 5.180196916217722e-06,
      "loss": 0.3268,
      "step": 32790
    },
    {
      "epoch": 0.7464215005802972,
      "grad_norm": 212.0,
      "learning_rate": 5.17555266579974e-06,
      "loss": 0.9751,
      "step": 32800
    },
    {
      "epoch": 0.7466490681109619,
      "grad_norm": 5.375,
      "learning_rate": 5.170908415381758e-06,
      "loss": 1.1386,
      "step": 32810
    },
    {
      "epoch": 0.7468766356416267,
      "grad_norm": 0.01007080078125,
      "learning_rate": 5.166264164963775e-06,
      "loss": 0.4075,
      "step": 32820
    },
    {
      "epoch": 0.7471042031722914,
      "grad_norm": 0.1650390625,
      "learning_rate": 5.161619914545792e-06,
      "loss": 0.9777,
      "step": 32830
    },
    {
      "epoch": 0.7473317707029561,
      "grad_norm": 160.0,
      "learning_rate": 5.15697566412781e-06,
      "loss": 0.752,
      "step": 32840
    },
    {
      "epoch": 0.7475593382336209,
      "grad_norm": 193.0,
      "learning_rate": 5.152331413709828e-06,
      "loss": 1.0562,
      "step": 32850
    },
    {
      "epoch": 0.7477869057642855,
      "grad_norm": 58.5,
      "learning_rate": 5.147687163291844e-06,
      "loss": 0.3038,
      "step": 32860
    },
    {
      "epoch": 0.7480144732949503,
      "grad_norm": 33.0,
      "learning_rate": 5.143042912873862e-06,
      "loss": 0.5354,
      "step": 32870
    },
    {
      "epoch": 0.748242040825615,
      "grad_norm": 0.055908203125,
      "learning_rate": 5.13839866245588e-06,
      "loss": 1.0672,
      "step": 32880
    },
    {
      "epoch": 0.7484696083562797,
      "grad_norm": 123.5,
      "learning_rate": 5.133754412037898e-06,
      "loss": 0.4356,
      "step": 32890
    },
    {
      "epoch": 0.7486971758869444,
      "grad_norm": 78.5,
      "learning_rate": 5.129110161619914e-06,
      "loss": 0.7123,
      "step": 32900
    },
    {
      "epoch": 0.7489247434176092,
      "grad_norm": 0.003021240234375,
      "learning_rate": 5.124465911201932e-06,
      "loss": 1.1179,
      "step": 32910
    },
    {
      "epoch": 0.749152310948274,
      "grad_norm": 132.0,
      "learning_rate": 5.11982166078395e-06,
      "loss": 0.5054,
      "step": 32920
    },
    {
      "epoch": 0.7493798784789386,
      "grad_norm": 0.01507568359375,
      "learning_rate": 5.115177410365967e-06,
      "loss": 0.3702,
      "step": 32930
    },
    {
      "epoch": 0.7496074460096034,
      "grad_norm": 0.0037994384765625,
      "learning_rate": 5.110533159947985e-06,
      "loss": 0.8239,
      "step": 32940
    },
    {
      "epoch": 0.749835013540268,
      "grad_norm": 0.0002880096435546875,
      "learning_rate": 5.105888909530002e-06,
      "loss": 0.5888,
      "step": 32950
    },
    {
      "epoch": 0.7500625810709328,
      "grad_norm": 0.0059814453125,
      "learning_rate": 5.10124465911202e-06,
      "loss": 0.587,
      "step": 32960
    },
    {
      "epoch": 0.7502901486015975,
      "grad_norm": 130.0,
      "learning_rate": 5.096600408694037e-06,
      "loss": 0.9237,
      "step": 32970
    },
    {
      "epoch": 0.7505177161322623,
      "grad_norm": 3.171875,
      "learning_rate": 5.091956158276055e-06,
      "loss": 1.1101,
      "step": 32980
    },
    {
      "epoch": 0.7507452836629269,
      "grad_norm": 340.0,
      "learning_rate": 5.087311907858072e-06,
      "loss": 1.1635,
      "step": 32990
    },
    {
      "epoch": 0.7509728511935917,
      "grad_norm": 0.1865234375,
      "learning_rate": 5.082667657440089e-06,
      "loss": 0.6334,
      "step": 33000
    },
    {
      "epoch": 0.7512004187242565,
      "grad_norm": 260.0,
      "learning_rate": 5.078023407022107e-06,
      "loss": 0.2618,
      "step": 33010
    },
    {
      "epoch": 0.7514279862549211,
      "grad_norm": 282.0,
      "learning_rate": 5.073379156604125e-06,
      "loss": 0.5989,
      "step": 33020
    },
    {
      "epoch": 0.7516555537855859,
      "grad_norm": 0.0027618408203125,
      "learning_rate": 5.068734906186143e-06,
      "loss": 0.4591,
      "step": 33030
    },
    {
      "epoch": 0.7518831213162506,
      "grad_norm": 0.0013275146484375,
      "learning_rate": 5.064090655768159e-06,
      "loss": 1.3463,
      "step": 33040
    },
    {
      "epoch": 0.7521106888469153,
      "grad_norm": 0.0240478515625,
      "learning_rate": 5.059446405350177e-06,
      "loss": 0.3583,
      "step": 33050
    },
    {
      "epoch": 0.75233825637758,
      "grad_norm": 218.0,
      "learning_rate": 5.054802154932195e-06,
      "loss": 0.8252,
      "step": 33060
    },
    {
      "epoch": 0.7525658239082448,
      "grad_norm": 3.09375,
      "learning_rate": 5.050157904514211e-06,
      "loss": 0.8593,
      "step": 33070
    },
    {
      "epoch": 0.7527933914389094,
      "grad_norm": 0.00592041015625,
      "learning_rate": 5.045513654096229e-06,
      "loss": 0.9507,
      "step": 33080
    },
    {
      "epoch": 0.7530209589695742,
      "grad_norm": 89.0,
      "learning_rate": 5.040869403678247e-06,
      "loss": 1.0212,
      "step": 33090
    },
    {
      "epoch": 0.753248526500239,
      "grad_norm": 0.86328125,
      "learning_rate": 5.036225153260265e-06,
      "loss": 0.5734,
      "step": 33100
    },
    {
      "epoch": 0.7534760940309037,
      "grad_norm": 384.0,
      "learning_rate": 5.031580902842281e-06,
      "loss": 1.2742,
      "step": 33110
    },
    {
      "epoch": 0.7537036615615684,
      "grad_norm": 0.00347900390625,
      "learning_rate": 5.026936652424299e-06,
      "loss": 0.754,
      "step": 33120
    },
    {
      "epoch": 0.7539312290922331,
      "grad_norm": 71.5,
      "learning_rate": 5.022292402006317e-06,
      "loss": 1.6129,
      "step": 33130
    },
    {
      "epoch": 0.7541587966228979,
      "grad_norm": 0.00396728515625,
      "learning_rate": 5.017648151588334e-06,
      "loss": 0.4926,
      "step": 33140
    },
    {
      "epoch": 0.7543863641535625,
      "grad_norm": 53.0,
      "learning_rate": 5.013003901170352e-06,
      "loss": 0.5782,
      "step": 33150
    },
    {
      "epoch": 0.7546139316842273,
      "grad_norm": 59.0,
      "learning_rate": 5.008359650752369e-06,
      "loss": 0.3223,
      "step": 33160
    },
    {
      "epoch": 0.754841499214892,
      "grad_norm": 8.6875,
      "learning_rate": 5.003715400334387e-06,
      "loss": 0.774,
      "step": 33170
    },
    {
      "epoch": 0.7550690667455567,
      "grad_norm": 108.5,
      "learning_rate": 4.9990711499164045e-06,
      "loss": 0.6756,
      "step": 33180
    },
    {
      "epoch": 0.7552966342762215,
      "grad_norm": 260.0,
      "learning_rate": 4.9944268994984216e-06,
      "loss": 1.5512,
      "step": 33190
    },
    {
      "epoch": 0.7555242018068862,
      "grad_norm": 112.0,
      "learning_rate": 4.9897826490804395e-06,
      "loss": 1.3506,
      "step": 33200
    },
    {
      "epoch": 0.755751769337551,
      "grad_norm": 115.0,
      "learning_rate": 4.9851383986624565e-06,
      "loss": 0.6016,
      "step": 33210
    },
    {
      "epoch": 0.7559793368682156,
      "grad_norm": 82.0,
      "learning_rate": 4.9804941482444736e-06,
      "loss": 1.1684,
      "step": 33220
    },
    {
      "epoch": 0.7562069043988804,
      "grad_norm": 5.0625,
      "learning_rate": 4.9758498978264915e-06,
      "loss": 0.6942,
      "step": 33230
    },
    {
      "epoch": 0.756434471929545,
      "grad_norm": 96.0,
      "learning_rate": 4.9712056474085085e-06,
      "loss": 0.3083,
      "step": 33240
    },
    {
      "epoch": 0.7566620394602098,
      "grad_norm": 282.0,
      "learning_rate": 4.966561396990526e-06,
      "loss": 1.321,
      "step": 33250
    },
    {
      "epoch": 0.7568896069908746,
      "grad_norm": 46.75,
      "learning_rate": 4.9619171465725435e-06,
      "loss": 0.6418,
      "step": 33260
    },
    {
      "epoch": 0.7571171745215393,
      "grad_norm": 0.0032196044921875,
      "learning_rate": 4.957272896154561e-06,
      "loss": 0.6699,
      "step": 33270
    },
    {
      "epoch": 0.757344742052204,
      "grad_norm": 0.0030059814453125,
      "learning_rate": 4.9526286457365784e-06,
      "loss": 0.2823,
      "step": 33280
    },
    {
      "epoch": 0.7575723095828687,
      "grad_norm": 0.037353515625,
      "learning_rate": 4.9479843953185955e-06,
      "loss": 0.7372,
      "step": 33290
    },
    {
      "epoch": 0.7577998771135335,
      "grad_norm": 2.671875,
      "learning_rate": 4.943340144900613e-06,
      "loss": 0.7365,
      "step": 33300
    },
    {
      "epoch": 0.7580274446441981,
      "grad_norm": 128.0,
      "learning_rate": 4.9386958944826304e-06,
      "loss": 1.7636,
      "step": 33310
    },
    {
      "epoch": 0.7582550121748629,
      "grad_norm": 330.0,
      "learning_rate": 4.934051644064648e-06,
      "loss": 1.3305,
      "step": 33320
    },
    {
      "epoch": 0.7584825797055276,
      "grad_norm": 23.5,
      "learning_rate": 4.929407393646665e-06,
      "loss": 0.9756,
      "step": 33330
    },
    {
      "epoch": 0.7587101472361923,
      "grad_norm": 0.109375,
      "learning_rate": 4.924763143228683e-06,
      "loss": 0.7742,
      "step": 33340
    },
    {
      "epoch": 0.7589377147668571,
      "grad_norm": 0.0159912109375,
      "learning_rate": 4.9201188928107e-06,
      "loss": 0.7168,
      "step": 33350
    },
    {
      "epoch": 0.7591652822975218,
      "grad_norm": 209.0,
      "learning_rate": 4.915474642392718e-06,
      "loss": 0.4259,
      "step": 33360
    },
    {
      "epoch": 0.7593928498281866,
      "grad_norm": 116.5,
      "learning_rate": 4.910830391974736e-06,
      "loss": 0.9307,
      "step": 33370
    },
    {
      "epoch": 0.7596204173588512,
      "grad_norm": 330.0,
      "learning_rate": 4.906186141556753e-06,
      "loss": 0.6475,
      "step": 33380
    },
    {
      "epoch": 0.759847984889516,
      "grad_norm": 4.34375,
      "learning_rate": 4.901541891138771e-06,
      "loss": 0.9563,
      "step": 33390
    },
    {
      "epoch": 0.7600755524201807,
      "grad_norm": 0.001068115234375,
      "learning_rate": 4.896897640720788e-06,
      "loss": 0.2594,
      "step": 33400
    },
    {
      "epoch": 0.7603031199508454,
      "grad_norm": 97.5,
      "learning_rate": 4.892253390302806e-06,
      "loss": 0.7482,
      "step": 33410
    },
    {
      "epoch": 0.7605306874815101,
      "grad_norm": 0.0018463134765625,
      "learning_rate": 4.887609139884823e-06,
      "loss": 0.6553,
      "step": 33420
    },
    {
      "epoch": 0.7607582550121749,
      "grad_norm": 169.0,
      "learning_rate": 4.88296488946684e-06,
      "loss": 0.9909,
      "step": 33430
    },
    {
      "epoch": 0.7609858225428396,
      "grad_norm": 270.0,
      "learning_rate": 4.878320639048858e-06,
      "loss": 0.5479,
      "step": 33440
    },
    {
      "epoch": 0.7612133900735043,
      "grad_norm": 17.25,
      "learning_rate": 4.873676388630875e-06,
      "loss": 1.0176,
      "step": 33450
    },
    {
      "epoch": 0.7614409576041691,
      "grad_norm": 50.75,
      "learning_rate": 4.869032138212893e-06,
      "loss": 1.0031,
      "step": 33460
    },
    {
      "epoch": 0.7616685251348337,
      "grad_norm": 932.0,
      "learning_rate": 4.86438788779491e-06,
      "loss": 0.4872,
      "step": 33470
    },
    {
      "epoch": 0.7618960926654985,
      "grad_norm": 0.060302734375,
      "learning_rate": 4.859743637376928e-06,
      "loss": 0.3155,
      "step": 33480
    },
    {
      "epoch": 0.7621236601961632,
      "grad_norm": 18.25,
      "learning_rate": 4.855099386958945e-06,
      "loss": 0.2922,
      "step": 33490
    },
    {
      "epoch": 0.762351227726828,
      "grad_norm": 5.78125,
      "learning_rate": 4.850455136540962e-06,
      "loss": 0.1818,
      "step": 33500
    },
    {
      "epoch": 0.7625787952574926,
      "grad_norm": 0.0693359375,
      "learning_rate": 4.84581088612298e-06,
      "loss": 0.7284,
      "step": 33510
    },
    {
      "epoch": 0.7628063627881574,
      "grad_norm": 174.0,
      "learning_rate": 4.841166635704997e-06,
      "loss": 1.8108,
      "step": 33520
    },
    {
      "epoch": 0.7630339303188222,
      "grad_norm": 0.53515625,
      "learning_rate": 4.836522385287015e-06,
      "loss": 0.7205,
      "step": 33530
    },
    {
      "epoch": 0.7632614978494868,
      "grad_norm": 42.5,
      "learning_rate": 4.831878134869033e-06,
      "loss": 0.3432,
      "step": 33540
    },
    {
      "epoch": 0.7634890653801516,
      "grad_norm": 1.03125,
      "learning_rate": 4.82723388445105e-06,
      "loss": 0.5571,
      "step": 33550
    },
    {
      "epoch": 0.7637166329108163,
      "grad_norm": 5.125,
      "learning_rate": 4.822589634033068e-06,
      "loss": 0.4755,
      "step": 33560
    },
    {
      "epoch": 0.763944200441481,
      "grad_norm": 0.005584716796875,
      "learning_rate": 4.817945383615085e-06,
      "loss": 0.7014,
      "step": 33570
    },
    {
      "epoch": 0.7641717679721457,
      "grad_norm": 212.0,
      "learning_rate": 4.813301133197103e-06,
      "loss": 0.8947,
      "step": 33580
    },
    {
      "epoch": 0.7643993355028105,
      "grad_norm": 109.0,
      "learning_rate": 4.80865688277912e-06,
      "loss": 0.7993,
      "step": 33590
    },
    {
      "epoch": 0.7646269030334751,
      "grad_norm": 356.0,
      "learning_rate": 4.804012632361138e-06,
      "loss": 0.4398,
      "step": 33600
    },
    {
      "epoch": 0.7648544705641399,
      "grad_norm": 58.25,
      "learning_rate": 4.799368381943155e-06,
      "loss": 0.9515,
      "step": 33610
    },
    {
      "epoch": 0.7650820380948047,
      "grad_norm": 0.455078125,
      "learning_rate": 4.794724131525173e-06,
      "loss": 1.6853,
      "step": 33620
    },
    {
      "epoch": 0.7653096056254693,
      "grad_norm": 147.0,
      "learning_rate": 4.79007988110719e-06,
      "loss": 1.0366,
      "step": 33630
    },
    {
      "epoch": 0.7655371731561341,
      "grad_norm": 1.5703125,
      "learning_rate": 4.785435630689207e-06,
      "loss": 0.8633,
      "step": 33640
    },
    {
      "epoch": 0.7657647406867988,
      "grad_norm": 210.0,
      "learning_rate": 4.780791380271225e-06,
      "loss": 1.2256,
      "step": 33650
    },
    {
      "epoch": 0.7659923082174636,
      "grad_norm": 374.0,
      "learning_rate": 4.776147129853242e-06,
      "loss": 1.2271,
      "step": 33660
    },
    {
      "epoch": 0.7662198757481282,
      "grad_norm": 49.5,
      "learning_rate": 4.77150287943526e-06,
      "loss": 1.1173,
      "step": 33670
    },
    {
      "epoch": 0.766447443278793,
      "grad_norm": 37.25,
      "learning_rate": 4.766858629017277e-06,
      "loss": 0.5934,
      "step": 33680
    },
    {
      "epoch": 0.7666750108094577,
      "grad_norm": 294.0,
      "learning_rate": 4.762214378599295e-06,
      "loss": 0.5344,
      "step": 33690
    },
    {
      "epoch": 0.7669025783401224,
      "grad_norm": 30.625,
      "learning_rate": 4.757570128181312e-06,
      "loss": 1.0645,
      "step": 33700
    },
    {
      "epoch": 0.7671301458707872,
      "grad_norm": 360.0,
      "learning_rate": 4.752925877763329e-06,
      "loss": 1.3593,
      "step": 33710
    },
    {
      "epoch": 0.7673577134014519,
      "grad_norm": 34.5,
      "learning_rate": 4.748281627345347e-06,
      "loss": 0.5545,
      "step": 33720
    },
    {
      "epoch": 0.7675852809321166,
      "grad_norm": 0.01495361328125,
      "learning_rate": 4.7436373769273645e-06,
      "loss": 0.9161,
      "step": 33730
    },
    {
      "epoch": 0.7678128484627813,
      "grad_norm": 0.1953125,
      "learning_rate": 4.738993126509382e-06,
      "loss": 0.3554,
      "step": 33740
    },
    {
      "epoch": 0.7680404159934461,
      "grad_norm": 0.004638671875,
      "learning_rate": 4.7343488760913995e-06,
      "loss": 1.0434,
      "step": 33750
    },
    {
      "epoch": 0.7682679835241107,
      "grad_norm": 0.0625,
      "learning_rate": 4.7297046256734166e-06,
      "loss": 0.8742,
      "step": 33760
    },
    {
      "epoch": 0.7684955510547755,
      "grad_norm": 312.0,
      "learning_rate": 4.7250603752554345e-06,
      "loss": 0.7768,
      "step": 33770
    },
    {
      "epoch": 0.7687231185854402,
      "grad_norm": 194.0,
      "learning_rate": 4.7204161248374515e-06,
      "loss": 0.5586,
      "step": 33780
    },
    {
      "epoch": 0.768950686116105,
      "grad_norm": 170.0,
      "learning_rate": 4.715771874419469e-06,
      "loss": 0.8125,
      "step": 33790
    },
    {
      "epoch": 0.7691782536467697,
      "grad_norm": 123.5,
      "learning_rate": 4.7111276240014865e-06,
      "loss": 1.2531,
      "step": 33800
    },
    {
      "epoch": 0.7694058211774344,
      "grad_norm": 0.00360107421875,
      "learning_rate": 4.706483373583504e-06,
      "loss": 0.5784,
      "step": 33810
    },
    {
      "epoch": 0.7696333887080992,
      "grad_norm": 0.00799560546875,
      "learning_rate": 4.7018391231655214e-06,
      "loss": 0.5269,
      "step": 33820
    },
    {
      "epoch": 0.7698609562387638,
      "grad_norm": 280.0,
      "learning_rate": 4.697194872747539e-06,
      "loss": 0.8091,
      "step": 33830
    },
    {
      "epoch": 0.7700885237694286,
      "grad_norm": 27.75,
      "learning_rate": 4.692550622329556e-06,
      "loss": 0.6047,
      "step": 33840
    },
    {
      "epoch": 0.7703160913000933,
      "grad_norm": 212.0,
      "learning_rate": 4.6879063719115734e-06,
      "loss": 0.401,
      "step": 33850
    },
    {
      "epoch": 0.770543658830758,
      "grad_norm": 0.0016632080078125,
      "learning_rate": 4.683262121493591e-06,
      "loss": 0.5788,
      "step": 33860
    },
    {
      "epoch": 0.7707712263614227,
      "grad_norm": 100.0,
      "learning_rate": 4.678617871075608e-06,
      "loss": 1.1948,
      "step": 33870
    },
    {
      "epoch": 0.7709987938920875,
      "grad_norm": 0.0157470703125,
      "learning_rate": 4.673973620657626e-06,
      "loss": 0.1832,
      "step": 33880
    },
    {
      "epoch": 0.7712263614227522,
      "grad_norm": 92.5,
      "learning_rate": 4.669329370239643e-06,
      "loss": 0.7473,
      "step": 33890
    },
    {
      "epoch": 0.7714539289534169,
      "grad_norm": 50.5,
      "learning_rate": 4.664685119821661e-06,
      "loss": 0.5739,
      "step": 33900
    },
    {
      "epoch": 0.7716814964840817,
      "grad_norm": 184.0,
      "learning_rate": 4.660040869403678e-06,
      "loss": 0.8654,
      "step": 33910
    },
    {
      "epoch": 0.7719090640147463,
      "grad_norm": 0.259765625,
      "learning_rate": 4.655396618985696e-06,
      "loss": 0.4546,
      "step": 33920
    },
    {
      "epoch": 0.7721366315454111,
      "grad_norm": 101.5,
      "learning_rate": 4.650752368567713e-06,
      "loss": 0.7966,
      "step": 33930
    },
    {
      "epoch": 0.7723641990760758,
      "grad_norm": 92.0,
      "learning_rate": 4.646108118149731e-06,
      "loss": 1.1002,
      "step": 33940
    },
    {
      "epoch": 0.7725917666067406,
      "grad_norm": 117.0,
      "learning_rate": 4.641463867731748e-06,
      "loss": 0.4622,
      "step": 33950
    },
    {
      "epoch": 0.7728193341374052,
      "grad_norm": 217.0,
      "learning_rate": 4.636819617313766e-06,
      "loss": 0.5537,
      "step": 33960
    },
    {
      "epoch": 0.77304690166807,
      "grad_norm": 0.01263427734375,
      "learning_rate": 4.632175366895784e-06,
      "loss": 1.1942,
      "step": 33970
    },
    {
      "epoch": 0.7732744691987348,
      "grad_norm": 0.0048828125,
      "learning_rate": 4.627531116477801e-06,
      "loss": 0.6684,
      "step": 33980
    },
    {
      "epoch": 0.7735020367293994,
      "grad_norm": 0.73828125,
      "learning_rate": 4.622886866059818e-06,
      "loss": 0.5888,
      "step": 33990
    },
    {
      "epoch": 0.7737296042600642,
      "grad_norm": 0.0020599365234375,
      "learning_rate": 4.618242615641836e-06,
      "loss": 0.6542,
      "step": 34000
    },
    {
      "epoch": 0.7739571717907289,
      "grad_norm": 13.75,
      "learning_rate": 4.613598365223853e-06,
      "loss": 0.1006,
      "step": 34010
    },
    {
      "epoch": 0.7741847393213936,
      "grad_norm": 93.0,
      "learning_rate": 4.608954114805871e-06,
      "loss": 0.3673,
      "step": 34020
    },
    {
      "epoch": 0.7744123068520583,
      "grad_norm": 0.75390625,
      "learning_rate": 4.604309864387888e-06,
      "loss": 0.3281,
      "step": 34030
    },
    {
      "epoch": 0.7746398743827231,
      "grad_norm": 121.5,
      "learning_rate": 4.599665613969906e-06,
      "loss": 0.6437,
      "step": 34040
    },
    {
      "epoch": 0.7748674419133877,
      "grad_norm": 151.0,
      "learning_rate": 4.595021363551923e-06,
      "loss": 0.7988,
      "step": 34050
    },
    {
      "epoch": 0.7750950094440525,
      "grad_norm": 162.0,
      "learning_rate": 4.59037711313394e-06,
      "loss": 0.9661,
      "step": 34060
    },
    {
      "epoch": 0.7753225769747173,
      "grad_norm": 680.0,
      "learning_rate": 4.585732862715958e-06,
      "loss": 0.9771,
      "step": 34070
    },
    {
      "epoch": 0.775550144505382,
      "grad_norm": 0.01611328125,
      "learning_rate": 4.581088612297975e-06,
      "loss": 0.2354,
      "step": 34080
    },
    {
      "epoch": 0.7757777120360467,
      "grad_norm": 57.0,
      "learning_rate": 4.576444361879993e-06,
      "loss": 0.7108,
      "step": 34090
    },
    {
      "epoch": 0.7760052795667114,
      "grad_norm": 156.0,
      "learning_rate": 4.57180011146201e-06,
      "loss": 0.8346,
      "step": 34100
    },
    {
      "epoch": 0.7762328470973762,
      "grad_norm": 74.0,
      "learning_rate": 4.567155861044028e-06,
      "loss": 1.8852,
      "step": 34110
    },
    {
      "epoch": 0.7764604146280408,
      "grad_norm": 67.5,
      "learning_rate": 4.562511610626045e-06,
      "loss": 0.6649,
      "step": 34120
    },
    {
      "epoch": 0.7766879821587056,
      "grad_norm": 5.15625,
      "learning_rate": 4.557867360208063e-06,
      "loss": 0.8679,
      "step": 34130
    },
    {
      "epoch": 0.7769155496893703,
      "grad_norm": 42.5,
      "learning_rate": 4.55322310979008e-06,
      "loss": 0.5629,
      "step": 34140
    },
    {
      "epoch": 0.777143117220035,
      "grad_norm": 0.006622314453125,
      "learning_rate": 4.548578859372098e-06,
      "loss": 0.6572,
      "step": 34150
    },
    {
      "epoch": 0.7773706847506998,
      "grad_norm": 95.0,
      "learning_rate": 4.543934608954116e-06,
      "loss": 0.6961,
      "step": 34160
    },
    {
      "epoch": 0.7775982522813645,
      "grad_norm": 35.5,
      "learning_rate": 4.539290358536133e-06,
      "loss": 0.8068,
      "step": 34170
    },
    {
      "epoch": 0.7778258198120293,
      "grad_norm": 0.00083160400390625,
      "learning_rate": 4.534646108118151e-06,
      "loss": 0.5425,
      "step": 34180
    },
    {
      "epoch": 0.7780533873426939,
      "grad_norm": 0.0069580078125,
      "learning_rate": 4.530001857700168e-06,
      "loss": 0.6577,
      "step": 34190
    },
    {
      "epoch": 0.7782809548733587,
      "grad_norm": 338.0,
      "learning_rate": 4.525357607282185e-06,
      "loss": 1.3804,
      "step": 34200
    },
    {
      "epoch": 0.7785085224040234,
      "grad_norm": 0.000732421875,
      "learning_rate": 4.520713356864203e-06,
      "loss": 0.8354,
      "step": 34210
    },
    {
      "epoch": 0.7787360899346881,
      "grad_norm": 75.0,
      "learning_rate": 4.51606910644622e-06,
      "loss": 1.2119,
      "step": 34220
    },
    {
      "epoch": 0.7789636574653529,
      "grad_norm": 211.0,
      "learning_rate": 4.511424856028238e-06,
      "loss": 1.4703,
      "step": 34230
    },
    {
      "epoch": 0.7791912249960176,
      "grad_norm": 209.0,
      "learning_rate": 4.506780605610255e-06,
      "loss": 1.4949,
      "step": 34240
    },
    {
      "epoch": 0.7794187925266823,
      "grad_norm": 286.0,
      "learning_rate": 4.5021363551922726e-06,
      "loss": 1.8654,
      "step": 34250
    },
    {
      "epoch": 0.779646360057347,
      "grad_norm": 81.0,
      "learning_rate": 4.49749210477429e-06,
      "loss": 0.2158,
      "step": 34260
    },
    {
      "epoch": 0.7798739275880118,
      "grad_norm": 221.0,
      "learning_rate": 4.492847854356307e-06,
      "loss": 1.5384,
      "step": 34270
    },
    {
      "epoch": 0.7801014951186764,
      "grad_norm": 68.5,
      "learning_rate": 4.488203603938325e-06,
      "loss": 0.6592,
      "step": 34280
    },
    {
      "epoch": 0.7803290626493412,
      "grad_norm": 0.01611328125,
      "learning_rate": 4.483559353520342e-06,
      "loss": 1.2643,
      "step": 34290
    },
    {
      "epoch": 0.7805566301800059,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.4789151031023595e-06,
      "loss": 0.5136,
      "step": 34300
    },
    {
      "epoch": 0.7807841977106706,
      "grad_norm": 0.00115966796875,
      "learning_rate": 4.474270852684377e-06,
      "loss": 0.2839,
      "step": 34310
    },
    {
      "epoch": 0.7810117652413354,
      "grad_norm": 0.000667572021484375,
      "learning_rate": 4.4696266022663945e-06,
      "loss": 0.6133,
      "step": 34320
    },
    {
      "epoch": 0.7812393327720001,
      "grad_norm": 314.0,
      "learning_rate": 4.464982351848412e-06,
      "loss": 1.2368,
      "step": 34330
    },
    {
      "epoch": 0.7814669003026649,
      "grad_norm": 258.0,
      "learning_rate": 4.4603381014304295e-06,
      "loss": 0.3896,
      "step": 34340
    },
    {
      "epoch": 0.7816944678333295,
      "grad_norm": 0.0751953125,
      "learning_rate": 4.455693851012447e-06,
      "loss": 0.5974,
      "step": 34350
    },
    {
      "epoch": 0.7819220353639943,
      "grad_norm": 132.0,
      "learning_rate": 4.451049600594464e-06,
      "loss": 1.3146,
      "step": 34360
    },
    {
      "epoch": 0.782149602894659,
      "grad_norm": 0.00022983551025390625,
      "learning_rate": 4.446405350176482e-06,
      "loss": 0.1737,
      "step": 34370
    },
    {
      "epoch": 0.7823771704253237,
      "grad_norm": 181.0,
      "learning_rate": 4.441761099758499e-06,
      "loss": 1.1424,
      "step": 34380
    },
    {
      "epoch": 0.7826047379559884,
      "grad_norm": 161.0,
      "learning_rate": 4.437116849340517e-06,
      "loss": 1.2486,
      "step": 34390
    },
    {
      "epoch": 0.7828323054866532,
      "grad_norm": 97.5,
      "learning_rate": 4.432472598922534e-06,
      "loss": 0.9539,
      "step": 34400
    },
    {
      "epoch": 0.7830598730173179,
      "grad_norm": 0.0087890625,
      "learning_rate": 4.427828348504551e-06,
      "loss": 0.5835,
      "step": 34410
    },
    {
      "epoch": 0.7832874405479826,
      "grad_norm": 66.0,
      "learning_rate": 4.423184098086569e-06,
      "loss": 1.2339,
      "step": 34420
    },
    {
      "epoch": 0.7835150080786474,
      "grad_norm": 171.0,
      "learning_rate": 4.418539847668586e-06,
      "loss": 0.7504,
      "step": 34430
    },
    {
      "epoch": 0.783742575609312,
      "grad_norm": 183.0,
      "learning_rate": 4.413895597250604e-06,
      "loss": 1.2637,
      "step": 34440
    },
    {
      "epoch": 0.7839701431399768,
      "grad_norm": 0.003021240234375,
      "learning_rate": 4.409251346832621e-06,
      "loss": 0.4016,
      "step": 34450
    },
    {
      "epoch": 0.7841977106706415,
      "grad_norm": 0.00183868408203125,
      "learning_rate": 4.404607096414639e-06,
      "loss": 1.1838,
      "step": 34460
    },
    {
      "epoch": 0.7844252782013063,
      "grad_norm": 82.5,
      "learning_rate": 4.399962845996656e-06,
      "loss": 0.3881,
      "step": 34470
    },
    {
      "epoch": 0.7846528457319709,
      "grad_norm": 1.6328125,
      "learning_rate": 4.395318595578673e-06,
      "loss": 0.891,
      "step": 34480
    },
    {
      "epoch": 0.7848804132626357,
      "grad_norm": 62.5,
      "learning_rate": 4.390674345160691e-06,
      "loss": 0.4636,
      "step": 34490
    },
    {
      "epoch": 0.7851079807933005,
      "grad_norm": 290.0,
      "learning_rate": 4.386030094742709e-06,
      "loss": 1.4101,
      "step": 34500
    },
    {
      "epoch": 0.7853355483239651,
      "grad_norm": 0.0026702880859375,
      "learning_rate": 4.381385844324726e-06,
      "loss": 0.7543,
      "step": 34510
    },
    {
      "epoch": 0.7855631158546299,
      "grad_norm": 1000.0,
      "learning_rate": 4.376741593906744e-06,
      "loss": 1.5729,
      "step": 34520
    },
    {
      "epoch": 0.7857906833852946,
      "grad_norm": 35.5,
      "learning_rate": 4.372097343488761e-06,
      "loss": 0.8616,
      "step": 34530
    },
    {
      "epoch": 0.7860182509159593,
      "grad_norm": 174.0,
      "learning_rate": 4.367453093070779e-06,
      "loss": 1.3624,
      "step": 34540
    },
    {
      "epoch": 0.786245818446624,
      "grad_norm": 3.34375,
      "learning_rate": 4.362808842652796e-06,
      "loss": 1.049,
      "step": 34550
    },
    {
      "epoch": 0.7864733859772888,
      "grad_norm": 7.40625,
      "learning_rate": 4.358164592234814e-06,
      "loss": 1.1966,
      "step": 34560
    },
    {
      "epoch": 0.7867009535079534,
      "grad_norm": 0.06787109375,
      "learning_rate": 4.353520341816831e-06,
      "loss": 0.5237,
      "step": 34570
    },
    {
      "epoch": 0.7869285210386182,
      "grad_norm": 51.75,
      "learning_rate": 4.348876091398849e-06,
      "loss": 1.0273,
      "step": 34580
    },
    {
      "epoch": 0.787156088569283,
      "grad_norm": 77.0,
      "learning_rate": 4.344231840980866e-06,
      "loss": 0.071,
      "step": 34590
    },
    {
      "epoch": 0.7873836560999476,
      "grad_norm": 34.5,
      "learning_rate": 4.339587590562884e-06,
      "loss": 1.0871,
      "step": 34600
    },
    {
      "epoch": 0.7876112236306124,
      "grad_norm": 0.000919342041015625,
      "learning_rate": 4.334943340144901e-06,
      "loss": 1.2659,
      "step": 34610
    },
    {
      "epoch": 0.7878387911612771,
      "grad_norm": 102.0,
      "learning_rate": 4.330299089726919e-06,
      "loss": 1.93,
      "step": 34620
    },
    {
      "epoch": 0.7880663586919419,
      "grad_norm": 151.0,
      "learning_rate": 4.325654839308936e-06,
      "loss": 1.1869,
      "step": 34630
    },
    {
      "epoch": 0.7882939262226065,
      "grad_norm": 2.640625,
      "learning_rate": 4.321010588890953e-06,
      "loss": 0.6584,
      "step": 34640
    },
    {
      "epoch": 0.7885214937532713,
      "grad_norm": 17.125,
      "learning_rate": 4.316366338472971e-06,
      "loss": 0.9343,
      "step": 34650
    },
    {
      "epoch": 0.788749061283936,
      "grad_norm": 25.25,
      "learning_rate": 4.311722088054988e-06,
      "loss": 0.6961,
      "step": 34660
    },
    {
      "epoch": 0.7889766288146007,
      "grad_norm": 252.0,
      "learning_rate": 4.307077837637006e-06,
      "loss": 0.4789,
      "step": 34670
    },
    {
      "epoch": 0.7892041963452655,
      "grad_norm": 39.5,
      "learning_rate": 4.302433587219023e-06,
      "loss": 0.2024,
      "step": 34680
    },
    {
      "epoch": 0.7894317638759302,
      "grad_norm": 1.640625,
      "learning_rate": 4.297789336801041e-06,
      "loss": 0.7032,
      "step": 34690
    },
    {
      "epoch": 0.7896593314065949,
      "grad_norm": 0.0021514892578125,
      "learning_rate": 4.293145086383058e-06,
      "loss": 0.0536,
      "step": 34700
    },
    {
      "epoch": 0.7898868989372596,
      "grad_norm": 134.0,
      "learning_rate": 4.288500835965076e-06,
      "loss": 0.4407,
      "step": 34710
    },
    {
      "epoch": 0.7901144664679244,
      "grad_norm": 113.0,
      "learning_rate": 4.283856585547093e-06,
      "loss": 0.6476,
      "step": 34720
    },
    {
      "epoch": 0.790342033998589,
      "grad_norm": 61.0,
      "learning_rate": 4.279212335129111e-06,
      "loss": 0.5796,
      "step": 34730
    },
    {
      "epoch": 0.7905696015292538,
      "grad_norm": 141.0,
      "learning_rate": 4.274568084711128e-06,
      "loss": 0.5367,
      "step": 34740
    },
    {
      "epoch": 0.7907971690599185,
      "grad_norm": 2.453125,
      "learning_rate": 4.269923834293146e-06,
      "loss": 0.8599,
      "step": 34750
    },
    {
      "epoch": 0.7910247365905833,
      "grad_norm": 140.0,
      "learning_rate": 4.2652795838751636e-06,
      "loss": 1.1373,
      "step": 34760
    },
    {
      "epoch": 0.791252304121248,
      "grad_norm": 9.9375,
      "learning_rate": 4.260635333457181e-06,
      "loss": 0.7951,
      "step": 34770
    },
    {
      "epoch": 0.7914798716519127,
      "grad_norm": 0.007080078125,
      "learning_rate": 4.255991083039198e-06,
      "loss": 1.0265,
      "step": 34780
    },
    {
      "epoch": 0.7917074391825775,
      "grad_norm": 620.0,
      "learning_rate": 4.2513468326212156e-06,
      "loss": 1.0167,
      "step": 34790
    },
    {
      "epoch": 0.7919350067132421,
      "grad_norm": 56.75,
      "learning_rate": 4.246702582203233e-06,
      "loss": 0.7284,
      "step": 34800
    },
    {
      "epoch": 0.7921625742439069,
      "grad_norm": 524.0,
      "learning_rate": 4.2420583317852505e-06,
      "loss": 0.8765,
      "step": 34810
    },
    {
      "epoch": 0.7923901417745716,
      "grad_norm": 0.002532958984375,
      "learning_rate": 4.2374140813672676e-06,
      "loss": 0.4113,
      "step": 34820
    },
    {
      "epoch": 0.7926177093052363,
      "grad_norm": 79.0,
      "learning_rate": 4.2327698309492855e-06,
      "loss": 0.7494,
      "step": 34830
    },
    {
      "epoch": 0.792845276835901,
      "grad_norm": 128.0,
      "learning_rate": 4.2281255805313025e-06,
      "loss": 0.6965,
      "step": 34840
    },
    {
      "epoch": 0.7930728443665658,
      "grad_norm": 318.0,
      "learning_rate": 4.22348133011332e-06,
      "loss": 1.3472,
      "step": 34850
    },
    {
      "epoch": 0.7933004118972306,
      "grad_norm": 55.5,
      "learning_rate": 4.2188370796953375e-06,
      "loss": 0.6159,
      "step": 34860
    },
    {
      "epoch": 0.7935279794278952,
      "grad_norm": 141.0,
      "learning_rate": 4.2141928292773545e-06,
      "loss": 1.3901,
      "step": 34870
    },
    {
      "epoch": 0.79375554695856,
      "grad_norm": 0.162109375,
      "learning_rate": 4.2095485788593724e-06,
      "loss": 0.7453,
      "step": 34880
    },
    {
      "epoch": 0.7939831144892247,
      "grad_norm": 10.6875,
      "learning_rate": 4.2049043284413895e-06,
      "loss": 0.2888,
      "step": 34890
    },
    {
      "epoch": 0.7942106820198894,
      "grad_norm": 318.0,
      "learning_rate": 4.200260078023407e-06,
      "loss": 0.6738,
      "step": 34900
    },
    {
      "epoch": 0.7944382495505541,
      "grad_norm": 0.04736328125,
      "learning_rate": 4.1956158276054245e-06,
      "loss": 1.5442,
      "step": 34910
    },
    {
      "epoch": 0.7946658170812189,
      "grad_norm": 31.5,
      "learning_rate": 4.190971577187442e-06,
      "loss": 0.5108,
      "step": 34920
    },
    {
      "epoch": 0.7948933846118835,
      "grad_norm": 112.0,
      "learning_rate": 4.18632732676946e-06,
      "loss": 0.9403,
      "step": 34930
    },
    {
      "epoch": 0.7951209521425483,
      "grad_norm": 0.0341796875,
      "learning_rate": 4.181683076351477e-06,
      "loss": 0.8197,
      "step": 34940
    },
    {
      "epoch": 0.7953485196732131,
      "grad_norm": 89.0,
      "learning_rate": 4.177038825933495e-06,
      "loss": 0.5499,
      "step": 34950
    },
    {
      "epoch": 0.7955760872038777,
      "grad_norm": 55.75,
      "learning_rate": 4.172394575515512e-06,
      "loss": 1.1121,
      "step": 34960
    },
    {
      "epoch": 0.7958036547345425,
      "grad_norm": 0.0771484375,
      "learning_rate": 4.16775032509753e-06,
      "loss": 0.064,
      "step": 34970
    },
    {
      "epoch": 0.7960312222652072,
      "grad_norm": 3.671875,
      "learning_rate": 4.163106074679547e-06,
      "loss": 0.4133,
      "step": 34980
    },
    {
      "epoch": 0.796258789795872,
      "grad_norm": 0.01483154296875,
      "learning_rate": 4.158461824261564e-06,
      "loss": 0.8712,
      "step": 34990
    },
    {
      "epoch": 0.7964863573265366,
      "grad_norm": 0.02587890625,
      "learning_rate": 4.153817573843582e-06,
      "loss": 0.4902,
      "step": 35000
    },
    {
      "epoch": 0.7967139248572014,
      "grad_norm": 133.0,
      "learning_rate": 4.149173323425599e-06,
      "loss": 0.965,
      "step": 35010
    },
    {
      "epoch": 0.796941492387866,
      "grad_norm": 122.0,
      "learning_rate": 4.144529073007617e-06,
      "loss": 0.8051,
      "step": 35020
    },
    {
      "epoch": 0.7971690599185308,
      "grad_norm": 0.10888671875,
      "learning_rate": 4.139884822589634e-06,
      "loss": 0.3471,
      "step": 35030
    },
    {
      "epoch": 0.7973966274491956,
      "grad_norm": 174.0,
      "learning_rate": 4.135240572171652e-06,
      "loss": 0.7128,
      "step": 35040
    },
    {
      "epoch": 0.7976241949798603,
      "grad_norm": 95.5,
      "learning_rate": 4.130596321753669e-06,
      "loss": 0.6152,
      "step": 35050
    },
    {
      "epoch": 0.797851762510525,
      "grad_norm": 11.25,
      "learning_rate": 4.125952071335686e-06,
      "loss": 0.2278,
      "step": 35060
    },
    {
      "epoch": 0.7980793300411897,
      "grad_norm": 60.25,
      "learning_rate": 4.121307820917704e-06,
      "loss": 0.9441,
      "step": 35070
    },
    {
      "epoch": 0.7983068975718545,
      "grad_norm": 148.0,
      "learning_rate": 4.116663570499721e-06,
      "loss": 0.3205,
      "step": 35080
    },
    {
      "epoch": 0.7985344651025191,
      "grad_norm": 41.25,
      "learning_rate": 4.112019320081739e-06,
      "loss": 0.4517,
      "step": 35090
    },
    {
      "epoch": 0.7987620326331839,
      "grad_norm": 145.0,
      "learning_rate": 4.107375069663756e-06,
      "loss": 0.5093,
      "step": 35100
    },
    {
      "epoch": 0.7989896001638486,
      "grad_norm": 56.75,
      "learning_rate": 4.102730819245774e-06,
      "loss": 1.5531,
      "step": 35110
    },
    {
      "epoch": 0.7992171676945133,
      "grad_norm": 107.5,
      "learning_rate": 4.098086568827792e-06,
      "loss": 0.331,
      "step": 35120
    },
    {
      "epoch": 0.7994447352251781,
      "grad_norm": 223.0,
      "learning_rate": 4.093442318409809e-06,
      "loss": 1.4207,
      "step": 35130
    },
    {
      "epoch": 0.7996723027558428,
      "grad_norm": 0.64453125,
      "learning_rate": 4.088798067991827e-06,
      "loss": 0.5879,
      "step": 35140
    },
    {
      "epoch": 0.7998998702865076,
      "grad_norm": 0.005340576171875,
      "learning_rate": 4.084153817573844e-06,
      "loss": 0.6367,
      "step": 35150
    },
    {
      "epoch": 0.8001274378171722,
      "grad_norm": 0.7578125,
      "learning_rate": 4.079509567155862e-06,
      "loss": 0.9204,
      "step": 35160
    },
    {
      "epoch": 0.800355005347837,
      "grad_norm": 0.0031280517578125,
      "learning_rate": 4.074865316737879e-06,
      "loss": 1.5267,
      "step": 35170
    },
    {
      "epoch": 0.8005825728785017,
      "grad_norm": 109.5,
      "learning_rate": 4.070221066319897e-06,
      "loss": 1.774,
      "step": 35180
    },
    {
      "epoch": 0.8008101404091664,
      "grad_norm": 82.0,
      "learning_rate": 4.065576815901914e-06,
      "loss": 0.5863,
      "step": 35190
    },
    {
      "epoch": 0.8010377079398311,
      "grad_norm": 104.5,
      "learning_rate": 4.060932565483931e-06,
      "loss": 0.4524,
      "step": 35200
    },
    {
      "epoch": 0.8012652754704959,
      "grad_norm": 167.0,
      "learning_rate": 4.056288315065949e-06,
      "loss": 1.6478,
      "step": 35210
    },
    {
      "epoch": 0.8014928430011606,
      "grad_norm": 0.006805419921875,
      "learning_rate": 4.051644064647966e-06,
      "loss": 0.7998,
      "step": 35220
    },
    {
      "epoch": 0.8017204105318253,
      "grad_norm": 19.875,
      "learning_rate": 4.046999814229984e-06,
      "loss": 0.4662,
      "step": 35230
    },
    {
      "epoch": 0.8019479780624901,
      "grad_norm": 104.5,
      "learning_rate": 4.042355563812001e-06,
      "loss": 0.1962,
      "step": 35240
    },
    {
      "epoch": 0.8021755455931547,
      "grad_norm": 123.0,
      "learning_rate": 4.037711313394019e-06,
      "loss": 0.9709,
      "step": 35250
    },
    {
      "epoch": 0.8024031131238195,
      "grad_norm": 113.0,
      "learning_rate": 4.033067062976036e-06,
      "loss": 0.5296,
      "step": 35260
    },
    {
      "epoch": 0.8026306806544842,
      "grad_norm": 0.0277099609375,
      "learning_rate": 4.028422812558053e-06,
      "loss": 0.8251,
      "step": 35270
    },
    {
      "epoch": 0.802858248185149,
      "grad_norm": 26.25,
      "learning_rate": 4.023778562140071e-06,
      "loss": 0.2832,
      "step": 35280
    },
    {
      "epoch": 0.8030858157158137,
      "grad_norm": 0.005767822265625,
      "learning_rate": 4.019134311722089e-06,
      "loss": 0.2567,
      "step": 35290
    },
    {
      "epoch": 0.8033133832464784,
      "grad_norm": 88.5,
      "learning_rate": 4.014490061304106e-06,
      "loss": 1.0019,
      "step": 35300
    },
    {
      "epoch": 0.8035409507771432,
      "grad_norm": 130.0,
      "learning_rate": 4.009845810886124e-06,
      "loss": 0.6091,
      "step": 35310
    },
    {
      "epoch": 0.8037685183078078,
      "grad_norm": 96.0,
      "learning_rate": 4.005201560468141e-06,
      "loss": 0.8902,
      "step": 35320
    },
    {
      "epoch": 0.8039960858384726,
      "grad_norm": 0.024169921875,
      "learning_rate": 4.0005573100501586e-06,
      "loss": 0.6822,
      "step": 35330
    },
    {
      "epoch": 0.8042236533691373,
      "grad_norm": 0.002044677734375,
      "learning_rate": 3.995913059632176e-06,
      "loss": 0.1501,
      "step": 35340
    },
    {
      "epoch": 0.804451220899802,
      "grad_norm": 378.0,
      "learning_rate": 3.9912688092141935e-06,
      "loss": 1.0958,
      "step": 35350
    },
    {
      "epoch": 0.8046787884304667,
      "grad_norm": 0.005859375,
      "learning_rate": 3.9866245587962106e-06,
      "loss": 0.781,
      "step": 35360
    },
    {
      "epoch": 0.8049063559611315,
      "grad_norm": 131.0,
      "learning_rate": 3.9819803083782285e-06,
      "loss": 0.8034,
      "step": 35370
    },
    {
      "epoch": 0.8051339234917962,
      "grad_norm": 43.25,
      "learning_rate": 3.9773360579602455e-06,
      "loss": 0.5361,
      "step": 35380
    },
    {
      "epoch": 0.8053614910224609,
      "grad_norm": 121.0,
      "learning_rate": 3.972691807542263e-06,
      "loss": 2.5527,
      "step": 35390
    },
    {
      "epoch": 0.8055890585531257,
      "grad_norm": 276.0,
      "learning_rate": 3.9680475571242805e-06,
      "loss": 0.3911,
      "step": 35400
    },
    {
      "epoch": 0.8058166260837903,
      "grad_norm": 0.0012664794921875,
      "learning_rate": 3.9634033067062975e-06,
      "loss": 0.5361,
      "step": 35410
    },
    {
      "epoch": 0.8060441936144551,
      "grad_norm": 0.34375,
      "learning_rate": 3.9587590562883154e-06,
      "loss": 0.7028,
      "step": 35420
    },
    {
      "epoch": 0.8062717611451198,
      "grad_norm": 132.0,
      "learning_rate": 3.9541148058703325e-06,
      "loss": 0.6728,
      "step": 35430
    },
    {
      "epoch": 0.8064993286757846,
      "grad_norm": 1.703125,
      "learning_rate": 3.94947055545235e-06,
      "loss": 0.6725,
      "step": 35440
    },
    {
      "epoch": 0.8067268962064492,
      "grad_norm": 0.00372314453125,
      "learning_rate": 3.9448263050343674e-06,
      "loss": 0.7857,
      "step": 35450
    },
    {
      "epoch": 0.806954463737114,
      "grad_norm": 55.0,
      "learning_rate": 3.940182054616385e-06,
      "loss": 0.3697,
      "step": 35460
    },
    {
      "epoch": 0.8071820312677788,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.935537804198402e-06,
      "loss": 0.6536,
      "step": 35470
    },
    {
      "epoch": 0.8074095987984434,
      "grad_norm": 0.0537109375,
      "learning_rate": 3.93089355378042e-06,
      "loss": 1.2075,
      "step": 35480
    },
    {
      "epoch": 0.8076371663291082,
      "grad_norm": 100.5,
      "learning_rate": 3.926249303362437e-06,
      "loss": 0.8859,
      "step": 35490
    },
    {
      "epoch": 0.8078647338597729,
      "grad_norm": 9.4375,
      "learning_rate": 3.921605052944455e-06,
      "loss": 0.2887,
      "step": 35500
    },
    {
      "epoch": 0.8080923013904376,
      "grad_norm": 632.0,
      "learning_rate": 3.916960802526472e-06,
      "loss": 1.3845,
      "step": 35510
    },
    {
      "epoch": 0.8083198689211023,
      "grad_norm": 61.5,
      "learning_rate": 3.91231655210849e-06,
      "loss": 0.7365,
      "step": 35520
    },
    {
      "epoch": 0.8085474364517671,
      "grad_norm": 0.007110595703125,
      "learning_rate": 3.907672301690508e-06,
      "loss": 0.4621,
      "step": 35530
    },
    {
      "epoch": 0.8087750039824317,
      "grad_norm": 207.0,
      "learning_rate": 3.903028051272525e-06,
      "loss": 0.7739,
      "step": 35540
    },
    {
      "epoch": 0.8090025715130965,
      "grad_norm": 104.0,
      "learning_rate": 3.898383800854542e-06,
      "loss": 0.356,
      "step": 35550
    },
    {
      "epoch": 0.8092301390437613,
      "grad_norm": 176.0,
      "learning_rate": 3.89373955043656e-06,
      "loss": 0.8192,
      "step": 35560
    },
    {
      "epoch": 0.809457706574426,
      "grad_norm": 211.0,
      "learning_rate": 3.889095300018577e-06,
      "loss": 0.3281,
      "step": 35570
    },
    {
      "epoch": 0.8096852741050907,
      "grad_norm": 118.5,
      "learning_rate": 3.884451049600595e-06,
      "loss": 1.1694,
      "step": 35580
    },
    {
      "epoch": 0.8099128416357554,
      "grad_norm": 0.00408935546875,
      "learning_rate": 3.879806799182612e-06,
      "loss": 1.1135,
      "step": 35590
    },
    {
      "epoch": 0.8101404091664202,
      "grad_norm": 0.006072998046875,
      "learning_rate": 3.87516254876463e-06,
      "loss": 0.8995,
      "step": 35600
    },
    {
      "epoch": 0.8103679766970848,
      "grad_norm": 0.000545501708984375,
      "learning_rate": 3.870518298346647e-06,
      "loss": 0.9028,
      "step": 35610
    },
    {
      "epoch": 0.8105955442277496,
      "grad_norm": 0.0023651123046875,
      "learning_rate": 3.865874047928664e-06,
      "loss": 0.2737,
      "step": 35620
    },
    {
      "epoch": 0.8108231117584143,
      "grad_norm": 101.5,
      "learning_rate": 3.861229797510682e-06,
      "loss": 0.788,
      "step": 35630
    },
    {
      "epoch": 0.811050679289079,
      "grad_norm": 320.0,
      "learning_rate": 3.856585547092699e-06,
      "loss": 1.619,
      "step": 35640
    },
    {
      "epoch": 0.8112782468197438,
      "grad_norm": 165.0,
      "learning_rate": 3.851941296674717e-06,
      "loss": 1.0128,
      "step": 35650
    },
    {
      "epoch": 0.8115058143504085,
      "grad_norm": 133.0,
      "learning_rate": 3.847297046256734e-06,
      "loss": 0.8878,
      "step": 35660
    },
    {
      "epoch": 0.8117333818810732,
      "grad_norm": 154.0,
      "learning_rate": 3.842652795838752e-06,
      "loss": 0.7354,
      "step": 35670
    },
    {
      "epoch": 0.8119609494117379,
      "grad_norm": 127.0,
      "learning_rate": 3.838008545420769e-06,
      "loss": 1.0767,
      "step": 35680
    },
    {
      "epoch": 0.8121885169424027,
      "grad_norm": 187.0,
      "learning_rate": 3.833364295002787e-06,
      "loss": 0.9246,
      "step": 35690
    },
    {
      "epoch": 0.8124160844730673,
      "grad_norm": 0.2109375,
      "learning_rate": 3.828720044584804e-06,
      "loss": 0.9109,
      "step": 35700
    },
    {
      "epoch": 0.8126436520037321,
      "grad_norm": 99.5,
      "learning_rate": 3.824075794166822e-06,
      "loss": 0.9211,
      "step": 35710
    },
    {
      "epoch": 0.8128712195343968,
      "grad_norm": 102.5,
      "learning_rate": 3.81943154374884e-06,
      "loss": 0.6501,
      "step": 35720
    },
    {
      "epoch": 0.8130987870650616,
      "grad_norm": 0.5859375,
      "learning_rate": 3.814787293330857e-06,
      "loss": 0.9523,
      "step": 35730
    },
    {
      "epoch": 0.8133263545957263,
      "grad_norm": 0.0038299560546875,
      "learning_rate": 3.8101430429128743e-06,
      "loss": 0.3959,
      "step": 35740
    },
    {
      "epoch": 0.813553922126391,
      "grad_norm": 91.5,
      "learning_rate": 3.805498792494892e-06,
      "loss": 0.4729,
      "step": 35750
    },
    {
      "epoch": 0.8137814896570558,
      "grad_norm": 12.9375,
      "learning_rate": 3.800854542076909e-06,
      "loss": 0.4477,
      "step": 35760
    },
    {
      "epoch": 0.8140090571877204,
      "grad_norm": 87.5,
      "learning_rate": 3.7962102916589268e-06,
      "loss": 1.039,
      "step": 35770
    },
    {
      "epoch": 0.8142366247183852,
      "grad_norm": 106.5,
      "learning_rate": 3.791566041240944e-06,
      "loss": 0.542,
      "step": 35780
    },
    {
      "epoch": 0.8144641922490499,
      "grad_norm": 0.000377655029296875,
      "learning_rate": 3.7869217908229617e-06,
      "loss": 1.3955,
      "step": 35790
    },
    {
      "epoch": 0.8146917597797146,
      "grad_norm": 6.90625,
      "learning_rate": 3.7822775404049788e-06,
      "loss": 0.7089,
      "step": 35800
    },
    {
      "epoch": 0.8149193273103793,
      "grad_norm": 472.0,
      "learning_rate": 3.7776332899869967e-06,
      "loss": 1.2375,
      "step": 35810
    },
    {
      "epoch": 0.8151468948410441,
      "grad_norm": 74.5,
      "learning_rate": 3.7729890395690137e-06,
      "loss": 0.2331,
      "step": 35820
    },
    {
      "epoch": 0.8153744623717089,
      "grad_norm": 83.0,
      "learning_rate": 3.768344789151031e-06,
      "loss": 1.3552,
      "step": 35830
    },
    {
      "epoch": 0.8156020299023735,
      "grad_norm": 0.00457763671875,
      "learning_rate": 3.7637005387330487e-06,
      "loss": 0.8349,
      "step": 35840
    },
    {
      "epoch": 0.8158295974330383,
      "grad_norm": 44.75,
      "learning_rate": 3.759056288315066e-06,
      "loss": 0.71,
      "step": 35850
    },
    {
      "epoch": 0.816057164963703,
      "grad_norm": 0.109375,
      "learning_rate": 3.754412037897084e-06,
      "loss": 1.0302,
      "step": 35860
    },
    {
      "epoch": 0.8162847324943677,
      "grad_norm": 29.625,
      "learning_rate": 3.749767787479101e-06,
      "loss": 0.4066,
      "step": 35870
    },
    {
      "epoch": 0.8165123000250324,
      "grad_norm": 0.5703125,
      "learning_rate": 3.745123537061119e-06,
      "loss": 0.4334,
      "step": 35880
    },
    {
      "epoch": 0.8167398675556972,
      "grad_norm": 398.0,
      "learning_rate": 3.740479286643136e-06,
      "loss": 1.0484,
      "step": 35890
    },
    {
      "epoch": 0.8169674350863618,
      "grad_norm": 86.5,
      "learning_rate": 3.735835036225153e-06,
      "loss": 0.5296,
      "step": 35900
    },
    {
      "epoch": 0.8171950026170266,
      "grad_norm": 88.5,
      "learning_rate": 3.731190785807171e-06,
      "loss": 0.4069,
      "step": 35910
    },
    {
      "epoch": 0.8174225701476914,
      "grad_norm": 1032.0,
      "learning_rate": 3.7265465353891885e-06,
      "loss": 1.0553,
      "step": 35920
    },
    {
      "epoch": 0.817650137678356,
      "grad_norm": 83.5,
      "learning_rate": 3.721902284971206e-06,
      "loss": 0.7672,
      "step": 35930
    },
    {
      "epoch": 0.8178777052090208,
      "grad_norm": 86.5,
      "learning_rate": 3.7172580345532235e-06,
      "loss": 1.2129,
      "step": 35940
    },
    {
      "epoch": 0.8181052727396855,
      "grad_norm": 0.0888671875,
      "learning_rate": 3.712613784135241e-06,
      "loss": 1.0739,
      "step": 35950
    },
    {
      "epoch": 0.8183328402703502,
      "grad_norm": 0.451171875,
      "learning_rate": 3.7079695337172584e-06,
      "loss": 0.5745,
      "step": 35960
    },
    {
      "epoch": 0.8185604078010149,
      "grad_norm": 0.007476806640625,
      "learning_rate": 3.7033252832992755e-06,
      "loss": 0.6136,
      "step": 35970
    },
    {
      "epoch": 0.8187879753316797,
      "grad_norm": 196.0,
      "learning_rate": 3.6986810328812934e-06,
      "loss": 0.4984,
      "step": 35980
    },
    {
      "epoch": 0.8190155428623443,
      "grad_norm": 6.59375,
      "learning_rate": 3.6940367824633104e-06,
      "loss": 1.1809,
      "step": 35990
    },
    {
      "epoch": 0.8192431103930091,
      "grad_norm": 142.0,
      "learning_rate": 3.6893925320453283e-06,
      "loss": 0.4723,
      "step": 36000
    },
    {
      "epoch": 0.8194706779236739,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.6847482816273454e-06,
      "loss": 0.2952,
      "step": 36010
    },
    {
      "epoch": 0.8196982454543386,
      "grad_norm": 0.03759765625,
      "learning_rate": 3.6801040312093633e-06,
      "loss": 0.4858,
      "step": 36020
    },
    {
      "epoch": 0.8199258129850033,
      "grad_norm": 200.0,
      "learning_rate": 3.6754597807913803e-06,
      "loss": 0.8237,
      "step": 36030
    },
    {
      "epoch": 0.820153380515668,
      "grad_norm": 203.0,
      "learning_rate": 3.670815530373398e-06,
      "loss": 0.9794,
      "step": 36040
    },
    {
      "epoch": 0.8203809480463328,
      "grad_norm": 0.004730224609375,
      "learning_rate": 3.6661712799554157e-06,
      "loss": 1.0119,
      "step": 36050
    },
    {
      "epoch": 0.8206085155769974,
      "grad_norm": 1432.0,
      "learning_rate": 3.6615270295374328e-06,
      "loss": 1.1913,
      "step": 36060
    },
    {
      "epoch": 0.8208360831076622,
      "grad_norm": 0.000385284423828125,
      "learning_rate": 3.6568827791194507e-06,
      "loss": 1.1917,
      "step": 36070
    },
    {
      "epoch": 0.8210636506383269,
      "grad_norm": 197.0,
      "learning_rate": 3.6522385287014677e-06,
      "loss": 0.7417,
      "step": 36080
    },
    {
      "epoch": 0.8212912181689916,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.6475942782834856e-06,
      "loss": 0.8677,
      "step": 36090
    },
    {
      "epoch": 0.8215187856996564,
      "grad_norm": 18.25,
      "learning_rate": 3.6429500278655027e-06,
      "loss": 0.4982,
      "step": 36100
    },
    {
      "epoch": 0.8217463532303211,
      "grad_norm": 204.0,
      "learning_rate": 3.63830577744752e-06,
      "loss": 1.2077,
      "step": 36110
    },
    {
      "epoch": 0.8219739207609859,
      "grad_norm": 62.25,
      "learning_rate": 3.6336615270295377e-06,
      "loss": 0.8629,
      "step": 36120
    },
    {
      "epoch": 0.8222014882916505,
      "grad_norm": 254.0,
      "learning_rate": 3.629017276611555e-06,
      "loss": 1.0013,
      "step": 36130
    },
    {
      "epoch": 0.8224290558223153,
      "grad_norm": 211.0,
      "learning_rate": 3.6243730261935726e-06,
      "loss": 1.5575,
      "step": 36140
    },
    {
      "epoch": 0.82265662335298,
      "grad_norm": 84.0,
      "learning_rate": 3.61972877577559e-06,
      "loss": 0.8066,
      "step": 36150
    },
    {
      "epoch": 0.8228841908836447,
      "grad_norm": 191.0,
      "learning_rate": 3.615084525357608e-06,
      "loss": 0.7149,
      "step": 36160
    },
    {
      "epoch": 0.8231117584143094,
      "grad_norm": 10.5,
      "learning_rate": 3.610440274939625e-06,
      "loss": 0.5763,
      "step": 36170
    },
    {
      "epoch": 0.8233393259449742,
      "grad_norm": 0.003936767578125,
      "learning_rate": 3.605796024521642e-06,
      "loss": 0.6642,
      "step": 36180
    },
    {
      "epoch": 0.8235668934756389,
      "grad_norm": 3.65625,
      "learning_rate": 3.60115177410366e-06,
      "loss": 0.9072,
      "step": 36190
    },
    {
      "epoch": 0.8237944610063036,
      "grad_norm": 0.00433349609375,
      "learning_rate": 3.596507523685677e-06,
      "loss": 0.792,
      "step": 36200
    },
    {
      "epoch": 0.8240220285369684,
      "grad_norm": 0.578125,
      "learning_rate": 3.591863273267695e-06,
      "loss": 0.4959,
      "step": 36210
    },
    {
      "epoch": 0.824249596067633,
      "grad_norm": 174.0,
      "learning_rate": 3.5872190228497124e-06,
      "loss": 1.0424,
      "step": 36220
    },
    {
      "epoch": 0.8244771635982978,
      "grad_norm": 0.026123046875,
      "learning_rate": 3.58257477243173e-06,
      "loss": 0.7105,
      "step": 36230
    },
    {
      "epoch": 0.8247047311289625,
      "grad_norm": 144.0,
      "learning_rate": 3.5779305220137474e-06,
      "loss": 0.7311,
      "step": 36240
    },
    {
      "epoch": 0.8249322986596273,
      "grad_norm": 84.0,
      "learning_rate": 3.5732862715957645e-06,
      "loss": 0.3587,
      "step": 36250
    },
    {
      "epoch": 0.825159866190292,
      "grad_norm": 107.0,
      "learning_rate": 3.5686420211777824e-06,
      "loss": 0.3666,
      "step": 36260
    },
    {
      "epoch": 0.8253874337209567,
      "grad_norm": 15.6875,
      "learning_rate": 3.5639977707597994e-06,
      "loss": 1.1279,
      "step": 36270
    },
    {
      "epoch": 0.8256150012516215,
      "grad_norm": 3.25,
      "learning_rate": 3.5593535203418173e-06,
      "loss": 0.4852,
      "step": 36280
    },
    {
      "epoch": 0.8258425687822861,
      "grad_norm": 203.0,
      "learning_rate": 3.5547092699238344e-06,
      "loss": 0.2943,
      "step": 36290
    },
    {
      "epoch": 0.8260701363129509,
      "grad_norm": 0.2412109375,
      "learning_rate": 3.5500650195058523e-06,
      "loss": 0.4492,
      "step": 36300
    },
    {
      "epoch": 0.8262977038436156,
      "grad_norm": 752.0,
      "learning_rate": 3.5454207690878693e-06,
      "loss": 1.364,
      "step": 36310
    },
    {
      "epoch": 0.8265252713742803,
      "grad_norm": 194.0,
      "learning_rate": 3.540776518669887e-06,
      "loss": 0.3604,
      "step": 36320
    },
    {
      "epoch": 0.826752838904945,
      "grad_norm": 115.5,
      "learning_rate": 3.5361322682519043e-06,
      "loss": 1.0495,
      "step": 36330
    },
    {
      "epoch": 0.8269804064356098,
      "grad_norm": 0.01373291015625,
      "learning_rate": 3.5314880178339218e-06,
      "loss": 0.9358,
      "step": 36340
    },
    {
      "epoch": 0.8272079739662745,
      "grad_norm": 0.003692626953125,
      "learning_rate": 3.5268437674159397e-06,
      "loss": 0.7612,
      "step": 36350
    },
    {
      "epoch": 0.8274355414969392,
      "grad_norm": 94.5,
      "learning_rate": 3.5221995169979567e-06,
      "loss": 0.8492,
      "step": 36360
    },
    {
      "epoch": 0.827663109027604,
      "grad_norm": 169.0,
      "learning_rate": 3.5175552665799746e-06,
      "loss": 0.6884,
      "step": 36370
    },
    {
      "epoch": 0.8278906765582686,
      "grad_norm": 274.0,
      "learning_rate": 3.5129110161619917e-06,
      "loss": 0.8339,
      "step": 36380
    },
    {
      "epoch": 0.8281182440889334,
      "grad_norm": 42.5,
      "learning_rate": 3.5082667657440087e-06,
      "loss": 0.7294,
      "step": 36390
    },
    {
      "epoch": 0.8283458116195981,
      "grad_norm": 0.01953125,
      "learning_rate": 3.5036225153260266e-06,
      "loss": 0.5342,
      "step": 36400
    },
    {
      "epoch": 0.8285733791502629,
      "grad_norm": 0.01519775390625,
      "learning_rate": 3.498978264908044e-06,
      "loss": 0.5502,
      "step": 36410
    },
    {
      "epoch": 0.8288009466809275,
      "grad_norm": 35.5,
      "learning_rate": 3.4943340144900616e-06,
      "loss": 1.4139,
      "step": 36420
    },
    {
      "epoch": 0.8290285142115923,
      "grad_norm": 146.0,
      "learning_rate": 3.489689764072079e-06,
      "loss": 1.5905,
      "step": 36430
    },
    {
      "epoch": 0.8292560817422571,
      "grad_norm": 428.0,
      "learning_rate": 3.4850455136540965e-06,
      "loss": 1.185,
      "step": 36440
    },
    {
      "epoch": 0.8294836492729217,
      "grad_norm": 0.00946044921875,
      "learning_rate": 3.480401263236114e-06,
      "loss": 0.7056,
      "step": 36450
    },
    {
      "epoch": 0.8297112168035865,
      "grad_norm": 70.0,
      "learning_rate": 3.475757012818132e-06,
      "loss": 0.5355,
      "step": 36460
    },
    {
      "epoch": 0.8299387843342512,
      "grad_norm": 94.5,
      "learning_rate": 3.471112762400149e-06,
      "loss": 0.8507,
      "step": 36470
    },
    {
      "epoch": 0.8301663518649159,
      "grad_norm": 152.0,
      "learning_rate": 3.466468511982166e-06,
      "loss": 1.4564,
      "step": 36480
    },
    {
      "epoch": 0.8303939193955806,
      "grad_norm": 0.294921875,
      "learning_rate": 3.461824261564184e-06,
      "loss": 0.8107,
      "step": 36490
    },
    {
      "epoch": 0.8306214869262454,
      "grad_norm": 122.5,
      "learning_rate": 3.457180011146201e-06,
      "loss": 0.249,
      "step": 36500
    },
    {
      "epoch": 0.83084905445691,
      "grad_norm": 0.341796875,
      "learning_rate": 3.452535760728219e-06,
      "loss": 0.7191,
      "step": 36510
    },
    {
      "epoch": 0.8310766219875748,
      "grad_norm": 0.0004825592041015625,
      "learning_rate": 3.4478915103102364e-06,
      "loss": 0.6621,
      "step": 36520
    },
    {
      "epoch": 0.8313041895182396,
      "grad_norm": 156.0,
      "learning_rate": 3.443247259892254e-06,
      "loss": 1.1363,
      "step": 36530
    },
    {
      "epoch": 0.8315317570489043,
      "grad_norm": 162.0,
      "learning_rate": 3.4386030094742713e-06,
      "loss": 0.3841,
      "step": 36540
    },
    {
      "epoch": 0.831759324579569,
      "grad_norm": 48.5,
      "learning_rate": 3.4339587590562884e-06,
      "loss": 0.9027,
      "step": 36550
    },
    {
      "epoch": 0.8319868921102337,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.4293145086383063e-06,
      "loss": 0.2517,
      "step": 36560
    },
    {
      "epoch": 0.8322144596408985,
      "grad_norm": 107.0,
      "learning_rate": 3.4246702582203233e-06,
      "loss": 0.5416,
      "step": 36570
    },
    {
      "epoch": 0.8324420271715631,
      "grad_norm": 197.0,
      "learning_rate": 3.4200260078023412e-06,
      "loss": 0.8416,
      "step": 36580
    },
    {
      "epoch": 0.8326695947022279,
      "grad_norm": 0.0012969970703125,
      "learning_rate": 3.4153817573843583e-06,
      "loss": 0.8415,
      "step": 36590
    },
    {
      "epoch": 0.8328971622328926,
      "grad_norm": 2.359375,
      "learning_rate": 3.410737506966376e-06,
      "loss": 0.5321,
      "step": 36600
    },
    {
      "epoch": 0.8331247297635573,
      "grad_norm": 74.5,
      "learning_rate": 3.4060932565483933e-06,
      "loss": 0.9315,
      "step": 36610
    },
    {
      "epoch": 0.8333522972942221,
      "grad_norm": 0.0140380859375,
      "learning_rate": 3.4014490061304107e-06,
      "loss": 0.379,
      "step": 36620
    },
    {
      "epoch": 0.8335798648248868,
      "grad_norm": 0.01123046875,
      "learning_rate": 3.396804755712428e-06,
      "loss": 0.8327,
      "step": 36630
    },
    {
      "epoch": 0.8338074323555515,
      "grad_norm": 90.5,
      "learning_rate": 3.3921605052944457e-06,
      "loss": 0.9565,
      "step": 36640
    },
    {
      "epoch": 0.8340349998862162,
      "grad_norm": 78.5,
      "learning_rate": 3.3875162548764636e-06,
      "loss": 0.6883,
      "step": 36650
    },
    {
      "epoch": 0.834262567416881,
      "grad_norm": 183.0,
      "learning_rate": 3.3828720044584806e-06,
      "loss": 0.9348,
      "step": 36660
    },
    {
      "epoch": 0.8344901349475456,
      "grad_norm": 0.03125,
      "learning_rate": 3.3782277540404985e-06,
      "loss": 1.6461,
      "step": 36670
    },
    {
      "epoch": 0.8347177024782104,
      "grad_norm": 0.0089111328125,
      "learning_rate": 3.3735835036225156e-06,
      "loss": 1.2373,
      "step": 36680
    },
    {
      "epoch": 0.8349452700088751,
      "grad_norm": 59.25,
      "learning_rate": 3.3689392532045327e-06,
      "loss": 1.2515,
      "step": 36690
    },
    {
      "epoch": 0.8351728375395399,
      "grad_norm": 0.00138092041015625,
      "learning_rate": 3.3642950027865506e-06,
      "loss": 0.3459,
      "step": 36700
    },
    {
      "epoch": 0.8354004050702046,
      "grad_norm": 51.0,
      "learning_rate": 3.359650752368568e-06,
      "loss": 0.2199,
      "step": 36710
    },
    {
      "epoch": 0.8356279726008693,
      "grad_norm": 158.0,
      "learning_rate": 3.3550065019505855e-06,
      "loss": 0.7559,
      "step": 36720
    },
    {
      "epoch": 0.8358555401315341,
      "grad_norm": 324.0,
      "learning_rate": 3.350362251532603e-06,
      "loss": 0.5995,
      "step": 36730
    },
    {
      "epoch": 0.8360831076621987,
      "grad_norm": 0.003753662109375,
      "learning_rate": 3.3457180011146205e-06,
      "loss": 0.2693,
      "step": 36740
    },
    {
      "epoch": 0.8363106751928635,
      "grad_norm": 1.9296875,
      "learning_rate": 3.341073750696638e-06,
      "loss": 1.778,
      "step": 36750
    },
    {
      "epoch": 0.8365382427235282,
      "grad_norm": 143.0,
      "learning_rate": 3.336429500278655e-06,
      "loss": 0.7647,
      "step": 36760
    },
    {
      "epoch": 0.8367658102541929,
      "grad_norm": 0.84765625,
      "learning_rate": 3.331785249860673e-06,
      "loss": 0.5585,
      "step": 36770
    },
    {
      "epoch": 0.8369933777848576,
      "grad_norm": 0.0087890625,
      "learning_rate": 3.32714099944269e-06,
      "loss": 1.1354,
      "step": 36780
    },
    {
      "epoch": 0.8372209453155224,
      "grad_norm": 158.0,
      "learning_rate": 3.322496749024708e-06,
      "loss": 0.3952,
      "step": 36790
    },
    {
      "epoch": 0.8374485128461872,
      "grad_norm": 126.0,
      "learning_rate": 3.317852498606725e-06,
      "loss": 0.7406,
      "step": 36800
    },
    {
      "epoch": 0.8376760803768518,
      "grad_norm": 0.267578125,
      "learning_rate": 3.313208248188743e-06,
      "loss": 0.2815,
      "step": 36810
    },
    {
      "epoch": 0.8379036479075166,
      "grad_norm": 96.0,
      "learning_rate": 3.3085639977707603e-06,
      "loss": 0.8999,
      "step": 36820
    },
    {
      "epoch": 0.8381312154381813,
      "grad_norm": 151.0,
      "learning_rate": 3.3039197473527774e-06,
      "loss": 0.4653,
      "step": 36830
    },
    {
      "epoch": 0.838358782968846,
      "grad_norm": 552.0,
      "learning_rate": 3.2992754969347953e-06,
      "loss": 0.233,
      "step": 36840
    },
    {
      "epoch": 0.8385863504995107,
      "grad_norm": 0.00103759765625,
      "learning_rate": 3.2946312465168123e-06,
      "loss": 0.6373,
      "step": 36850
    },
    {
      "epoch": 0.8388139180301755,
      "grad_norm": 193.0,
      "learning_rate": 3.2899869960988302e-06,
      "loss": 1.2057,
      "step": 36860
    },
    {
      "epoch": 0.8390414855608401,
      "grad_norm": 7.53125,
      "learning_rate": 3.2853427456808473e-06,
      "loss": 0.3808,
      "step": 36870
    },
    {
      "epoch": 0.8392690530915049,
      "grad_norm": 0.0027313232421875,
      "learning_rate": 3.280698495262865e-06,
      "loss": 0.6334,
      "step": 36880
    },
    {
      "epoch": 0.8394966206221697,
      "grad_norm": 0.0732421875,
      "learning_rate": 3.2760542448448822e-06,
      "loss": 0.4672,
      "step": 36890
    },
    {
      "epoch": 0.8397241881528343,
      "grad_norm": 368.0,
      "learning_rate": 3.2714099944268997e-06,
      "loss": 1.2501,
      "step": 36900
    },
    {
      "epoch": 0.8399517556834991,
      "grad_norm": 86.5,
      "learning_rate": 3.266765744008917e-06,
      "loss": 0.6497,
      "step": 36910
    },
    {
      "epoch": 0.8401793232141638,
      "grad_norm": 21.75,
      "learning_rate": 3.2621214935909347e-06,
      "loss": 0.3415,
      "step": 36920
    },
    {
      "epoch": 0.8404068907448285,
      "grad_norm": 234.0,
      "learning_rate": 3.257477243172952e-06,
      "loss": 1.5189,
      "step": 36930
    },
    {
      "epoch": 0.8406344582754932,
      "grad_norm": 92.0,
      "learning_rate": 3.2528329927549696e-06,
      "loss": 0.5246,
      "step": 36940
    },
    {
      "epoch": 0.840862025806158,
      "grad_norm": 101.0,
      "learning_rate": 3.2481887423369875e-06,
      "loss": 1.515,
      "step": 36950
    },
    {
      "epoch": 0.8410895933368226,
      "grad_norm": 152.0,
      "learning_rate": 3.2435444919190046e-06,
      "loss": 0.4679,
      "step": 36960
    },
    {
      "epoch": 0.8413171608674874,
      "grad_norm": 77.5,
      "learning_rate": 3.2389002415010216e-06,
      "loss": 0.8459,
      "step": 36970
    },
    {
      "epoch": 0.8415447283981522,
      "grad_norm": 420.0,
      "learning_rate": 3.2342559910830395e-06,
      "loss": 0.7959,
      "step": 36980
    },
    {
      "epoch": 0.8417722959288169,
      "grad_norm": 390.0,
      "learning_rate": 3.2296117406650566e-06,
      "loss": 0.7391,
      "step": 36990
    },
    {
      "epoch": 0.8419998634594816,
      "grad_norm": 47.75,
      "learning_rate": 3.2249674902470745e-06,
      "loss": 1.3744,
      "step": 37000
    },
    {
      "epoch": 0.8422274309901463,
      "grad_norm": 0.043212890625,
      "learning_rate": 3.220323239829092e-06,
      "loss": 0.2153,
      "step": 37010
    },
    {
      "epoch": 0.8424549985208111,
      "grad_norm": 486.0,
      "learning_rate": 3.2156789894111094e-06,
      "loss": 1.0625,
      "step": 37020
    },
    {
      "epoch": 0.8426825660514757,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.211034738993127e-06,
      "loss": 0.4297,
      "step": 37030
    },
    {
      "epoch": 0.8429101335821405,
      "grad_norm": 0.001312255859375,
      "learning_rate": 3.206390488575144e-06,
      "loss": 1.2573,
      "step": 37040
    },
    {
      "epoch": 0.8431377011128052,
      "grad_norm": 0.052978515625,
      "learning_rate": 3.201746238157162e-06,
      "loss": 1.4472,
      "step": 37050
    },
    {
      "epoch": 0.84336526864347,
      "grad_norm": 2.953125,
      "learning_rate": 3.197101987739179e-06,
      "loss": 0.861,
      "step": 37060
    },
    {
      "epoch": 0.8435928361741347,
      "grad_norm": 164.0,
      "learning_rate": 3.192457737321197e-06,
      "loss": 1.1493,
      "step": 37070
    },
    {
      "epoch": 0.8438204037047994,
      "grad_norm": 5.34375,
      "learning_rate": 3.187813486903214e-06,
      "loss": 0.6872,
      "step": 37080
    },
    {
      "epoch": 0.8440479712354642,
      "grad_norm": 204.0,
      "learning_rate": 3.183169236485232e-06,
      "loss": 0.7484,
      "step": 37090
    },
    {
      "epoch": 0.8442755387661288,
      "grad_norm": 3.9375,
      "learning_rate": 3.178524986067249e-06,
      "loss": 1.6068,
      "step": 37100
    },
    {
      "epoch": 0.8445031062967936,
      "grad_norm": 1.015625,
      "learning_rate": 3.1738807356492663e-06,
      "loss": 1.2351,
      "step": 37110
    },
    {
      "epoch": 0.8447306738274583,
      "grad_norm": 57.25,
      "learning_rate": 3.1692364852312842e-06,
      "loss": 0.5311,
      "step": 37120
    },
    {
      "epoch": 0.844958241358123,
      "grad_norm": 0.373046875,
      "learning_rate": 3.1645922348133013e-06,
      "loss": 0.3901,
      "step": 37130
    },
    {
      "epoch": 0.8451858088887877,
      "grad_norm": 0.028564453125,
      "learning_rate": 3.159947984395319e-06,
      "loss": 0.2723,
      "step": 37140
    },
    {
      "epoch": 0.8454133764194525,
      "grad_norm": 6.0,
      "learning_rate": 3.1553037339773362e-06,
      "loss": 0.8138,
      "step": 37150
    },
    {
      "epoch": 0.8456409439501172,
      "grad_norm": 47.25,
      "learning_rate": 3.150659483559354e-06,
      "loss": 0.4094,
      "step": 37160
    },
    {
      "epoch": 0.8458685114807819,
      "grad_norm": 2.53125,
      "learning_rate": 3.146015233141371e-06,
      "loss": 0.6714,
      "step": 37170
    },
    {
      "epoch": 0.8460960790114467,
      "grad_norm": 0.007049560546875,
      "learning_rate": 3.1413709827233887e-06,
      "loss": 0.4559,
      "step": 37180
    },
    {
      "epoch": 0.8463236465421113,
      "grad_norm": 128.0,
      "learning_rate": 3.136726732305406e-06,
      "loss": 0.3019,
      "step": 37190
    },
    {
      "epoch": 0.8465512140727761,
      "grad_norm": 1624.0,
      "learning_rate": 3.1320824818874236e-06,
      "loss": 0.9804,
      "step": 37200
    },
    {
      "epoch": 0.8467787816034408,
      "grad_norm": 22.75,
      "learning_rate": 3.127438231469441e-06,
      "loss": 0.6202,
      "step": 37210
    },
    {
      "epoch": 0.8470063491341056,
      "grad_norm": 0.0016021728515625,
      "learning_rate": 3.1227939810514586e-06,
      "loss": 0.9937,
      "step": 37220
    },
    {
      "epoch": 0.8472339166647703,
      "grad_norm": 4.5625,
      "learning_rate": 3.118149730633476e-06,
      "loss": 0.2964,
      "step": 37230
    },
    {
      "epoch": 0.847461484195435,
      "grad_norm": 0.00101470947265625,
      "learning_rate": 3.1135054802154935e-06,
      "loss": 0.6394,
      "step": 37240
    },
    {
      "epoch": 0.8476890517260998,
      "grad_norm": 260.0,
      "learning_rate": 3.1088612297975106e-06,
      "loss": 0.958,
      "step": 37250
    },
    {
      "epoch": 0.8479166192567644,
      "grad_norm": 107.5,
      "learning_rate": 3.1042169793795285e-06,
      "loss": 0.7758,
      "step": 37260
    },
    {
      "epoch": 0.8481441867874292,
      "grad_norm": 79.0,
      "learning_rate": 3.0995727289615456e-06,
      "loss": 1.1645,
      "step": 37270
    },
    {
      "epoch": 0.8483717543180939,
      "grad_norm": 0.515625,
      "learning_rate": 3.0949284785435635e-06,
      "loss": 0.9962,
      "step": 37280
    },
    {
      "epoch": 0.8485993218487586,
      "grad_norm": 580.0,
      "learning_rate": 3.0902842281255805e-06,
      "loss": 0.9814,
      "step": 37290
    },
    {
      "epoch": 0.8488268893794233,
      "grad_norm": 0.28125,
      "learning_rate": 3.0856399777075984e-06,
      "loss": 0.5737,
      "step": 37300
    },
    {
      "epoch": 0.8490544569100881,
      "grad_norm": 3.609375,
      "learning_rate": 3.080995727289616e-06,
      "loss": 0.9064,
      "step": 37310
    },
    {
      "epoch": 0.8492820244407528,
      "grad_norm": 68.0,
      "learning_rate": 3.076351476871633e-06,
      "loss": 0.6521,
      "step": 37320
    },
    {
      "epoch": 0.8495095919714175,
      "grad_norm": 79.5,
      "learning_rate": 3.071707226453651e-06,
      "loss": 1.1957,
      "step": 37330
    },
    {
      "epoch": 0.8497371595020823,
      "grad_norm": 0.322265625,
      "learning_rate": 3.067062976035668e-06,
      "loss": 0.964,
      "step": 37340
    },
    {
      "epoch": 0.849964727032747,
      "grad_norm": 0.037353515625,
      "learning_rate": 3.062418725617686e-06,
      "loss": 1.2012,
      "step": 37350
    },
    {
      "epoch": 0.8501922945634117,
      "grad_norm": 46.75,
      "learning_rate": 3.057774475199703e-06,
      "loss": 0.6458,
      "step": 37360
    },
    {
      "epoch": 0.8504198620940764,
      "grad_norm": 0.0341796875,
      "learning_rate": 3.0531302247817208e-06,
      "loss": 0.4323,
      "step": 37370
    },
    {
      "epoch": 0.8506474296247412,
      "grad_norm": 0.0013275146484375,
      "learning_rate": 3.048485974363738e-06,
      "loss": 0.5469,
      "step": 37380
    },
    {
      "epoch": 0.8508749971554058,
      "grad_norm": 0.017578125,
      "learning_rate": 3.0438417239457553e-06,
      "loss": 0.709,
      "step": 37390
    },
    {
      "epoch": 0.8511025646860706,
      "grad_norm": 0.000152587890625,
      "learning_rate": 3.0391974735277728e-06,
      "loss": 0.5888,
      "step": 37400
    },
    {
      "epoch": 0.8513301322167354,
      "grad_norm": 17.0,
      "learning_rate": 3.0345532231097903e-06,
      "loss": 0.9634,
      "step": 37410
    },
    {
      "epoch": 0.8515576997474,
      "grad_norm": 207.0,
      "learning_rate": 3.029908972691808e-06,
      "loss": 0.9425,
      "step": 37420
    },
    {
      "epoch": 0.8517852672780648,
      "grad_norm": 181.0,
      "learning_rate": 3.0252647222738252e-06,
      "loss": 0.6937,
      "step": 37430
    },
    {
      "epoch": 0.8520128348087295,
      "grad_norm": 720.0,
      "learning_rate": 3.020620471855843e-06,
      "loss": 0.6997,
      "step": 37440
    },
    {
      "epoch": 0.8522404023393942,
      "grad_norm": 148.0,
      "learning_rate": 3.01597622143786e-06,
      "loss": 0.5342,
      "step": 37450
    },
    {
      "epoch": 0.8524679698700589,
      "grad_norm": 128.0,
      "learning_rate": 3.0113319710198772e-06,
      "loss": 0.8169,
      "step": 37460
    },
    {
      "epoch": 0.8526955374007237,
      "grad_norm": 386.0,
      "learning_rate": 3.006687720601895e-06,
      "loss": 1.1508,
      "step": 37470
    },
    {
      "epoch": 0.8529231049313883,
      "grad_norm": 199.0,
      "learning_rate": 3.0020434701839126e-06,
      "loss": 0.6555,
      "step": 37480
    },
    {
      "epoch": 0.8531506724620531,
      "grad_norm": 0.45703125,
      "learning_rate": 2.99739921976593e-06,
      "loss": 0.3029,
      "step": 37490
    },
    {
      "epoch": 0.8533782399927179,
      "grad_norm": 490.0,
      "learning_rate": 2.9927549693479476e-06,
      "loss": 0.6338,
      "step": 37500
    },
    {
      "epoch": 0.8536058075233826,
      "grad_norm": 0.0017547607421875,
      "learning_rate": 2.988110718929965e-06,
      "loss": 0.6076,
      "step": 37510
    },
    {
      "epoch": 0.8538333750540473,
      "grad_norm": 125.0,
      "learning_rate": 2.9834664685119825e-06,
      "loss": 0.7814,
      "step": 37520
    },
    {
      "epoch": 0.854060942584712,
      "grad_norm": 156.0,
      "learning_rate": 2.9788222180939996e-06,
      "loss": 0.2773,
      "step": 37530
    },
    {
      "epoch": 0.8542885101153768,
      "grad_norm": 236.0,
      "learning_rate": 2.9741779676760175e-06,
      "loss": 0.9726,
      "step": 37540
    },
    {
      "epoch": 0.8545160776460414,
      "grad_norm": 31.5,
      "learning_rate": 2.9695337172580345e-06,
      "loss": 0.3637,
      "step": 37550
    },
    {
      "epoch": 0.8547436451767062,
      "grad_norm": 0.71875,
      "learning_rate": 2.9648894668400524e-06,
      "loss": 1.4841,
      "step": 37560
    },
    {
      "epoch": 0.8549712127073709,
      "grad_norm": 221.0,
      "learning_rate": 2.9602452164220695e-06,
      "loss": 0.5392,
      "step": 37570
    },
    {
      "epoch": 0.8551987802380356,
      "grad_norm": 23.875,
      "learning_rate": 2.9556009660040874e-06,
      "loss": 0.412,
      "step": 37580
    },
    {
      "epoch": 0.8554263477687004,
      "grad_norm": 142.0,
      "learning_rate": 2.9509567155861044e-06,
      "loss": 0.7894,
      "step": 37590
    },
    {
      "epoch": 0.8556539152993651,
      "grad_norm": 9.918212890625e-05,
      "learning_rate": 2.946312465168122e-06,
      "loss": 0.3653,
      "step": 37600
    },
    {
      "epoch": 0.8558814828300298,
      "grad_norm": 94.0,
      "learning_rate": 2.94166821475014e-06,
      "loss": 0.879,
      "step": 37610
    },
    {
      "epoch": 0.8561090503606945,
      "grad_norm": 0.0269775390625,
      "learning_rate": 2.937023964332157e-06,
      "loss": 0.4148,
      "step": 37620
    },
    {
      "epoch": 0.8563366178913593,
      "grad_norm": 130.0,
      "learning_rate": 2.9323797139141748e-06,
      "loss": 0.397,
      "step": 37630
    },
    {
      "epoch": 0.856564185422024,
      "grad_norm": 0.00164031982421875,
      "learning_rate": 2.927735463496192e-06,
      "loss": 0.5978,
      "step": 37640
    },
    {
      "epoch": 0.8567917529526887,
      "grad_norm": 0.004150390625,
      "learning_rate": 2.9230912130782097e-06,
      "loss": 0.2544,
      "step": 37650
    },
    {
      "epoch": 0.8570193204833534,
      "grad_norm": 223.0,
      "learning_rate": 2.918446962660227e-06,
      "loss": 0.6235,
      "step": 37660
    },
    {
      "epoch": 0.8572468880140182,
      "grad_norm": 28.375,
      "learning_rate": 2.9138027122422443e-06,
      "loss": 1.0042,
      "step": 37670
    },
    {
      "epoch": 0.8574744555446829,
      "grad_norm": 83.5,
      "learning_rate": 2.9091584618242618e-06,
      "loss": 0.7284,
      "step": 37680
    },
    {
      "epoch": 0.8577020230753476,
      "grad_norm": 296.0,
      "learning_rate": 2.9045142114062792e-06,
      "loss": 0.6937,
      "step": 37690
    },
    {
      "epoch": 0.8579295906060124,
      "grad_norm": 0.48828125,
      "learning_rate": 2.8998699609882967e-06,
      "loss": 0.5004,
      "step": 37700
    },
    {
      "epoch": 0.858157158136677,
      "grad_norm": 250.0,
      "learning_rate": 2.895225710570314e-06,
      "loss": 0.6805,
      "step": 37710
    },
    {
      "epoch": 0.8583847256673418,
      "grad_norm": 262.0,
      "learning_rate": 2.8905814601523317e-06,
      "loss": 1.0976,
      "step": 37720
    },
    {
      "epoch": 0.8586122931980065,
      "grad_norm": 0.431640625,
      "learning_rate": 2.885937209734349e-06,
      "loss": 0.7135,
      "step": 37730
    },
    {
      "epoch": 0.8588398607286712,
      "grad_norm": 328.0,
      "learning_rate": 2.881292959316366e-06,
      "loss": 0.9131,
      "step": 37740
    },
    {
      "epoch": 0.8590674282593359,
      "grad_norm": 98.0,
      "learning_rate": 2.876648708898384e-06,
      "loss": 1.331,
      "step": 37750
    },
    {
      "epoch": 0.8592949957900007,
      "grad_norm": 0.046630859375,
      "learning_rate": 2.872004458480401e-06,
      "loss": 0.9098,
      "step": 37760
    },
    {
      "epoch": 0.8595225633206655,
      "grad_norm": 144.0,
      "learning_rate": 2.867360208062419e-06,
      "loss": 0.9397,
      "step": 37770
    },
    {
      "epoch": 0.8597501308513301,
      "grad_norm": 43.75,
      "learning_rate": 2.8627159576444365e-06,
      "loss": 1.167,
      "step": 37780
    },
    {
      "epoch": 0.8599776983819949,
      "grad_norm": 0.0074462890625,
      "learning_rate": 2.858071707226454e-06,
      "loss": 0.5232,
      "step": 37790
    },
    {
      "epoch": 0.8602052659126596,
      "grad_norm": 0.03466796875,
      "learning_rate": 2.8534274568084715e-06,
      "loss": 0.6691,
      "step": 37800
    },
    {
      "epoch": 0.8604328334433243,
      "grad_norm": 168.0,
      "learning_rate": 2.8487832063904885e-06,
      "loss": 0.5781,
      "step": 37810
    },
    {
      "epoch": 0.860660400973989,
      "grad_norm": 149.0,
      "learning_rate": 2.8441389559725064e-06,
      "loss": 0.8373,
      "step": 37820
    },
    {
      "epoch": 0.8608879685046538,
      "grad_norm": 65.0,
      "learning_rate": 2.8394947055545235e-06,
      "loss": 0.3185,
      "step": 37830
    },
    {
      "epoch": 0.8611155360353184,
      "grad_norm": 0.031494140625,
      "learning_rate": 2.8348504551365414e-06,
      "loss": 1.0476,
      "step": 37840
    },
    {
      "epoch": 0.8613431035659832,
      "grad_norm": 284.0,
      "learning_rate": 2.8302062047185585e-06,
      "loss": 0.3016,
      "step": 37850
    },
    {
      "epoch": 0.861570671096648,
      "grad_norm": 115.5,
      "learning_rate": 2.8255619543005764e-06,
      "loss": 1.1202,
      "step": 37860
    },
    {
      "epoch": 0.8617982386273126,
      "grad_norm": 450.0,
      "learning_rate": 2.8209177038825934e-06,
      "loss": 0.9652,
      "step": 37870
    },
    {
      "epoch": 0.8620258061579774,
      "grad_norm": 109.5,
      "learning_rate": 2.816273453464611e-06,
      "loss": 0.2732,
      "step": 37880
    },
    {
      "epoch": 0.8622533736886421,
      "grad_norm": 0.008056640625,
      "learning_rate": 2.8116292030466284e-06,
      "loss": 1.2132,
      "step": 37890
    },
    {
      "epoch": 0.8624809412193069,
      "grad_norm": 141.0,
      "learning_rate": 2.806984952628646e-06,
      "loss": 0.7935,
      "step": 37900
    },
    {
      "epoch": 0.8627085087499715,
      "grad_norm": 117.0,
      "learning_rate": 2.8023407022106638e-06,
      "loss": 0.6853,
      "step": 37910
    },
    {
      "epoch": 0.8629360762806363,
      "grad_norm": 134.0,
      "learning_rate": 2.797696451792681e-06,
      "loss": 0.6329,
      "step": 37920
    },
    {
      "epoch": 0.863163643811301,
      "grad_norm": 0.045654296875,
      "learning_rate": 2.7930522013746987e-06,
      "loss": 0.3806,
      "step": 37930
    },
    {
      "epoch": 0.8633912113419657,
      "grad_norm": 181.0,
      "learning_rate": 2.7884079509567158e-06,
      "loss": 0.3467,
      "step": 37940
    },
    {
      "epoch": 0.8636187788726305,
      "grad_norm": 78.0,
      "learning_rate": 2.783763700538733e-06,
      "loss": 0.794,
      "step": 37950
    },
    {
      "epoch": 0.8638463464032952,
      "grad_norm": 14.9375,
      "learning_rate": 2.7791194501207507e-06,
      "loss": 0.4774,
      "step": 37960
    },
    {
      "epoch": 0.8640739139339599,
      "grad_norm": 288.0,
      "learning_rate": 2.774475199702768e-06,
      "loss": 1.1021,
      "step": 37970
    },
    {
      "epoch": 0.8643014814646246,
      "grad_norm": 41.75,
      "learning_rate": 2.7698309492847857e-06,
      "loss": 0.8135,
      "step": 37980
    },
    {
      "epoch": 0.8645290489952894,
      "grad_norm": 15.375,
      "learning_rate": 2.765186698866803e-06,
      "loss": 1.4092,
      "step": 37990
    },
    {
      "epoch": 0.864756616525954,
      "grad_norm": 1.9921875,
      "learning_rate": 2.7605424484488206e-06,
      "loss": 0.7986,
      "step": 38000
    },
    {
      "epoch": 0.8649841840566188,
      "grad_norm": 9.5367431640625e-05,
      "learning_rate": 2.755898198030838e-06,
      "loss": 0.5902,
      "step": 38010
    },
    {
      "epoch": 0.8652117515872835,
      "grad_norm": 0.0015716552734375,
      "learning_rate": 2.751253947612855e-06,
      "loss": 0.6114,
      "step": 38020
    },
    {
      "epoch": 0.8654393191179482,
      "grad_norm": 104.5,
      "learning_rate": 2.746609697194873e-06,
      "loss": 1.7117,
      "step": 38030
    },
    {
      "epoch": 0.865666886648613,
      "grad_norm": 131.0,
      "learning_rate": 2.74196544677689e-06,
      "loss": 1.4261,
      "step": 38040
    },
    {
      "epoch": 0.8658944541792777,
      "grad_norm": 108.5,
      "learning_rate": 2.737321196358908e-06,
      "loss": 1.1102,
      "step": 38050
    },
    {
      "epoch": 0.8661220217099425,
      "grad_norm": 165.0,
      "learning_rate": 2.732676945940925e-06,
      "loss": 1.4162,
      "step": 38060
    },
    {
      "epoch": 0.8663495892406071,
      "grad_norm": 0.0478515625,
      "learning_rate": 2.728032695522943e-06,
      "loss": 0.8285,
      "step": 38070
    },
    {
      "epoch": 0.8665771567712719,
      "grad_norm": 528.0,
      "learning_rate": 2.7233884451049605e-06,
      "loss": 0.6101,
      "step": 38080
    },
    {
      "epoch": 0.8668047243019366,
      "grad_norm": 366.0,
      "learning_rate": 2.7187441946869775e-06,
      "loss": 0.7081,
      "step": 38090
    },
    {
      "epoch": 0.8670322918326013,
      "grad_norm": 92.0,
      "learning_rate": 2.7140999442689954e-06,
      "loss": 0.6953,
      "step": 38100
    },
    {
      "epoch": 0.867259859363266,
      "grad_norm": 52.25,
      "learning_rate": 2.7094556938510125e-06,
      "loss": 1.0633,
      "step": 38110
    },
    {
      "epoch": 0.8674874268939308,
      "grad_norm": 39.0,
      "learning_rate": 2.7048114434330304e-06,
      "loss": 0.5285,
      "step": 38120
    },
    {
      "epoch": 0.8677149944245955,
      "grad_norm": 0.0301513671875,
      "learning_rate": 2.7001671930150474e-06,
      "loss": 0.0967,
      "step": 38130
    },
    {
      "epoch": 0.8679425619552602,
      "grad_norm": 648.0,
      "learning_rate": 2.6955229425970653e-06,
      "loss": 1.3772,
      "step": 38140
    },
    {
      "epoch": 0.868170129485925,
      "grad_norm": 0.0030517578125,
      "learning_rate": 2.6908786921790824e-06,
      "loss": 0.6069,
      "step": 38150
    },
    {
      "epoch": 0.8683976970165896,
      "grad_norm": 0.009033203125,
      "learning_rate": 2.6862344417611e-06,
      "loss": 0.874,
      "step": 38160
    },
    {
      "epoch": 0.8686252645472544,
      "grad_norm": 0.0014190673828125,
      "learning_rate": 2.6815901913431173e-06,
      "loss": 0.5441,
      "step": 38170
    },
    {
      "epoch": 0.8688528320779191,
      "grad_norm": 213.0,
      "learning_rate": 2.676945940925135e-06,
      "loss": 0.3126,
      "step": 38180
    },
    {
      "epoch": 0.8690803996085839,
      "grad_norm": 13.375,
      "learning_rate": 2.6723016905071523e-06,
      "loss": 0.3812,
      "step": 38190
    },
    {
      "epoch": 0.8693079671392486,
      "grad_norm": 172.0,
      "learning_rate": 2.6676574400891698e-06,
      "loss": 1.0021,
      "step": 38200
    },
    {
      "epoch": 0.8695355346699133,
      "grad_norm": 33.5,
      "learning_rate": 2.6630131896711877e-06,
      "loss": 0.0799,
      "step": 38210
    },
    {
      "epoch": 0.8697631022005781,
      "grad_norm": 117.0,
      "learning_rate": 2.6583689392532047e-06,
      "loss": 0.4817,
      "step": 38220
    },
    {
      "epoch": 0.8699906697312427,
      "grad_norm": 153.0,
      "learning_rate": 2.653724688835222e-06,
      "loss": 0.8601,
      "step": 38230
    },
    {
      "epoch": 0.8702182372619075,
      "grad_norm": 0.00067138671875,
      "learning_rate": 2.6490804384172397e-06,
      "loss": 0.6808,
      "step": 38240
    },
    {
      "epoch": 0.8704458047925722,
      "grad_norm": 174.0,
      "learning_rate": 2.6444361879992568e-06,
      "loss": 0.563,
      "step": 38250
    },
    {
      "epoch": 0.8706733723232369,
      "grad_norm": 268.0,
      "learning_rate": 2.6397919375812747e-06,
      "loss": 0.3454,
      "step": 38260
    },
    {
      "epoch": 0.8709009398539016,
      "grad_norm": 93.0,
      "learning_rate": 2.635147687163292e-06,
      "loss": 0.4794,
      "step": 38270
    },
    {
      "epoch": 0.8711285073845664,
      "grad_norm": 252.0,
      "learning_rate": 2.6305034367453096e-06,
      "loss": 0.7577,
      "step": 38280
    },
    {
      "epoch": 0.8713560749152311,
      "grad_norm": 0.001922607421875,
      "learning_rate": 2.625859186327327e-06,
      "loss": 1.0244,
      "step": 38290
    },
    {
      "epoch": 0.8715836424458958,
      "grad_norm": 37.25,
      "learning_rate": 2.621214935909344e-06,
      "loss": 0.8107,
      "step": 38300
    },
    {
      "epoch": 0.8718112099765606,
      "grad_norm": 88.0,
      "learning_rate": 2.616570685491362e-06,
      "loss": 1.0488,
      "step": 38310
    },
    {
      "epoch": 0.8720387775072252,
      "grad_norm": 624.0,
      "learning_rate": 2.611926435073379e-06,
      "loss": 1.5232,
      "step": 38320
    },
    {
      "epoch": 0.87226634503789,
      "grad_norm": 198.0,
      "learning_rate": 2.607282184655397e-06,
      "loss": 0.827,
      "step": 38330
    },
    {
      "epoch": 0.8724939125685547,
      "grad_norm": 58.0,
      "learning_rate": 2.602637934237414e-06,
      "loss": 1.3417,
      "step": 38340
    },
    {
      "epoch": 0.8727214800992195,
      "grad_norm": 0.043701171875,
      "learning_rate": 2.597993683819432e-06,
      "loss": 0.5065,
      "step": 38350
    },
    {
      "epoch": 0.8729490476298841,
      "grad_norm": 12.0625,
      "learning_rate": 2.593349433401449e-06,
      "loss": 0.4259,
      "step": 38360
    },
    {
      "epoch": 0.8731766151605489,
      "grad_norm": 152.0,
      "learning_rate": 2.588705182983467e-06,
      "loss": 0.7806,
      "step": 38370
    },
    {
      "epoch": 0.8734041826912137,
      "grad_norm": 38.75,
      "learning_rate": 2.584060932565484e-06,
      "loss": 0.8262,
      "step": 38380
    },
    {
      "epoch": 0.8736317502218783,
      "grad_norm": 197.0,
      "learning_rate": 2.5794166821475015e-06,
      "loss": 1.4511,
      "step": 38390
    },
    {
      "epoch": 0.8738593177525431,
      "grad_norm": 8.5,
      "learning_rate": 2.5747724317295194e-06,
      "loss": 0.9592,
      "step": 38400
    },
    {
      "epoch": 0.8740868852832078,
      "grad_norm": 0.00439453125,
      "learning_rate": 2.5701281813115364e-06,
      "loss": 1.0694,
      "step": 38410
    },
    {
      "epoch": 0.8743144528138725,
      "grad_norm": 0.0045166015625,
      "learning_rate": 2.5654839308935543e-06,
      "loss": 0.711,
      "step": 38420
    },
    {
      "epoch": 0.8745420203445372,
      "grad_norm": 176.0,
      "learning_rate": 2.5608396804755714e-06,
      "loss": 1.7328,
      "step": 38430
    },
    {
      "epoch": 0.874769587875202,
      "grad_norm": 398.0,
      "learning_rate": 2.5561954300575893e-06,
      "loss": 1.7606,
      "step": 38440
    },
    {
      "epoch": 0.8749971554058666,
      "grad_norm": 0.00323486328125,
      "learning_rate": 2.5515511796396063e-06,
      "loss": 0.4511,
      "step": 38450
    },
    {
      "epoch": 0.8752247229365314,
      "grad_norm": 107.5,
      "learning_rate": 2.546906929221624e-06,
      "loss": 0.7491,
      "step": 38460
    },
    {
      "epoch": 0.8754522904671962,
      "grad_norm": 0.006683349609375,
      "learning_rate": 2.5422626788036413e-06,
      "loss": 0.1302,
      "step": 38470
    },
    {
      "epoch": 0.8756798579978609,
      "grad_norm": 0.0019683837890625,
      "learning_rate": 2.5376184283856588e-06,
      "loss": 0.123,
      "step": 38480
    },
    {
      "epoch": 0.8759074255285256,
      "grad_norm": 1.3515625,
      "learning_rate": 2.5329741779676762e-06,
      "loss": 0.4095,
      "step": 38490
    },
    {
      "epoch": 0.8761349930591903,
      "grad_norm": 0.001007080078125,
      "learning_rate": 2.5283299275496937e-06,
      "loss": 0.6416,
      "step": 38500
    },
    {
      "epoch": 0.8763625605898551,
      "grad_norm": 158.0,
      "learning_rate": 2.5236856771317116e-06,
      "loss": 0.6458,
      "step": 38510
    },
    {
      "epoch": 0.8765901281205197,
      "grad_norm": 74.0,
      "learning_rate": 2.5190414267137287e-06,
      "loss": 0.5827,
      "step": 38520
    },
    {
      "epoch": 0.8768176956511845,
      "grad_norm": 0.000705718994140625,
      "learning_rate": 2.5143971762957457e-06,
      "loss": 0.3351,
      "step": 38530
    },
    {
      "epoch": 0.8770452631818492,
      "grad_norm": 692.0,
      "learning_rate": 2.5097529258777636e-06,
      "loss": 0.6072,
      "step": 38540
    },
    {
      "epoch": 0.8772728307125139,
      "grad_norm": 140.0,
      "learning_rate": 2.5051086754597807e-06,
      "loss": 1.0702,
      "step": 38550
    },
    {
      "epoch": 0.8775003982431787,
      "grad_norm": 178.0,
      "learning_rate": 2.5004644250417986e-06,
      "loss": 0.5,
      "step": 38560
    },
    {
      "epoch": 0.8777279657738434,
      "grad_norm": 0.006683349609375,
      "learning_rate": 2.495820174623816e-06,
      "loss": 0.4341,
      "step": 38570
    },
    {
      "epoch": 0.8779555333045082,
      "grad_norm": 0.06982421875,
      "learning_rate": 2.4911759242058335e-06,
      "loss": 0.414,
      "step": 38580
    },
    {
      "epoch": 0.8781831008351728,
      "grad_norm": 0.0625,
      "learning_rate": 2.486531673787851e-06,
      "loss": 1.1987,
      "step": 38590
    },
    {
      "epoch": 0.8784106683658376,
      "grad_norm": 0.1171875,
      "learning_rate": 2.4818874233698685e-06,
      "loss": 0.0302,
      "step": 38600
    },
    {
      "epoch": 0.8786382358965023,
      "grad_norm": 30.5,
      "learning_rate": 2.477243172951886e-06,
      "loss": 0.5378,
      "step": 38610
    },
    {
      "epoch": 0.878865803427167,
      "grad_norm": 9.4375,
      "learning_rate": 2.472598922533903e-06,
      "loss": 0.4307,
      "step": 38620
    },
    {
      "epoch": 0.8790933709578317,
      "grad_norm": 185.0,
      "learning_rate": 2.4679546721159205e-06,
      "loss": 0.5764,
      "step": 38630
    },
    {
      "epoch": 0.8793209384884965,
      "grad_norm": 462.0,
      "learning_rate": 2.463310421697938e-06,
      "loss": 0.8343,
      "step": 38640
    },
    {
      "epoch": 0.8795485060191612,
      "grad_norm": 199.0,
      "learning_rate": 2.4586661712799555e-06,
      "loss": 1.9169,
      "step": 38650
    },
    {
      "epoch": 0.8797760735498259,
      "grad_norm": 0.08740234375,
      "learning_rate": 2.454021920861973e-06,
      "loss": 0.2283,
      "step": 38660
    },
    {
      "epoch": 0.8800036410804907,
      "grad_norm": 0.72265625,
      "learning_rate": 2.4493776704439904e-06,
      "loss": 0.57,
      "step": 38670
    },
    {
      "epoch": 0.8802312086111553,
      "grad_norm": 106.0,
      "learning_rate": 2.444733420026008e-06,
      "loss": 1.2208,
      "step": 38680
    },
    {
      "epoch": 0.8804587761418201,
      "grad_norm": 121.5,
      "learning_rate": 2.4400891696080254e-06,
      "loss": 0.7081,
      "step": 38690
    },
    {
      "epoch": 0.8806863436724848,
      "grad_norm": 106.0,
      "learning_rate": 2.435444919190043e-06,
      "loss": 0.6047,
      "step": 38700
    },
    {
      "epoch": 0.8809139112031495,
      "grad_norm": 60.75,
      "learning_rate": 2.4308006687720603e-06,
      "loss": 0.5824,
      "step": 38710
    },
    {
      "epoch": 0.8811414787338142,
      "grad_norm": 139.0,
      "learning_rate": 2.426156418354078e-06,
      "loss": 0.7305,
      "step": 38720
    },
    {
      "epoch": 0.881369046264479,
      "grad_norm": 108.5,
      "learning_rate": 2.4215121679360953e-06,
      "loss": 1.3065,
      "step": 38730
    },
    {
      "epoch": 0.8815966137951438,
      "grad_norm": 27.5,
      "learning_rate": 2.4168679175181128e-06,
      "loss": 0.6733,
      "step": 38740
    },
    {
      "epoch": 0.8818241813258084,
      "grad_norm": 266.0,
      "learning_rate": 2.4122236671001303e-06,
      "loss": 0.7179,
      "step": 38750
    },
    {
      "epoch": 0.8820517488564732,
      "grad_norm": 122.5,
      "learning_rate": 2.4075794166821477e-06,
      "loss": 1.0378,
      "step": 38760
    },
    {
      "epoch": 0.8822793163871379,
      "grad_norm": 26.75,
      "learning_rate": 2.402935166264165e-06,
      "loss": 1.1369,
      "step": 38770
    },
    {
      "epoch": 0.8825068839178026,
      "grad_norm": 153.0,
      "learning_rate": 2.3982909158461827e-06,
      "loss": 1.0131,
      "step": 38780
    },
    {
      "epoch": 0.8827344514484673,
      "grad_norm": 328.0,
      "learning_rate": 2.3936466654282e-06,
      "loss": 0.7188,
      "step": 38790
    },
    {
      "epoch": 0.8829620189791321,
      "grad_norm": 87.5,
      "learning_rate": 2.3890024150102176e-06,
      "loss": 0.9656,
      "step": 38800
    },
    {
      "epoch": 0.8831895865097967,
      "grad_norm": 0.047607421875,
      "learning_rate": 2.384358164592235e-06,
      "loss": 0.2546,
      "step": 38810
    },
    {
      "epoch": 0.8834171540404615,
      "grad_norm": 232.0,
      "learning_rate": 2.3797139141742526e-06,
      "loss": 1.4063,
      "step": 38820
    },
    {
      "epoch": 0.8836447215711263,
      "grad_norm": 143.0,
      "learning_rate": 2.3750696637562697e-06,
      "loss": 1.1688,
      "step": 38830
    },
    {
      "epoch": 0.8838722891017909,
      "grad_norm": 19.625,
      "learning_rate": 2.370425413338287e-06,
      "loss": 1.2766,
      "step": 38840
    },
    {
      "epoch": 0.8840998566324557,
      "grad_norm": 876.0,
      "learning_rate": 2.3657811629203046e-06,
      "loss": 1.361,
      "step": 38850
    },
    {
      "epoch": 0.8843274241631204,
      "grad_norm": 202.0,
      "learning_rate": 2.361136912502322e-06,
      "loss": 1.0456,
      "step": 38860
    },
    {
      "epoch": 0.8845549916937852,
      "grad_norm": 53.0,
      "learning_rate": 2.35649266208434e-06,
      "loss": 0.3813,
      "step": 38870
    },
    {
      "epoch": 0.8847825592244498,
      "grad_norm": 45.0,
      "learning_rate": 2.3518484116663575e-06,
      "loss": 0.7371,
      "step": 38880
    },
    {
      "epoch": 0.8850101267551146,
      "grad_norm": 100.5,
      "learning_rate": 2.347204161248375e-06,
      "loss": 0.2555,
      "step": 38890
    },
    {
      "epoch": 0.8852376942857793,
      "grad_norm": 98.0,
      "learning_rate": 2.3425599108303924e-06,
      "loss": 0.7791,
      "step": 38900
    },
    {
      "epoch": 0.885465261816444,
      "grad_norm": 308.0,
      "learning_rate": 2.3379156604124095e-06,
      "loss": 0.5103,
      "step": 38910
    },
    {
      "epoch": 0.8856928293471088,
      "grad_norm": 153.0,
      "learning_rate": 2.333271409994427e-06,
      "loss": 0.8535,
      "step": 38920
    },
    {
      "epoch": 0.8859203968777735,
      "grad_norm": 102.5,
      "learning_rate": 2.3286271595764444e-06,
      "loss": 0.6356,
      "step": 38930
    },
    {
      "epoch": 0.8861479644084382,
      "grad_norm": 0.00148773193359375,
      "learning_rate": 2.323982909158462e-06,
      "loss": 1.1517,
      "step": 38940
    },
    {
      "epoch": 0.8863755319391029,
      "grad_norm": 193.0,
      "learning_rate": 2.3193386587404794e-06,
      "loss": 0.4882,
      "step": 38950
    },
    {
      "epoch": 0.8866030994697677,
      "grad_norm": 106.0,
      "learning_rate": 2.314694408322497e-06,
      "loss": 0.8464,
      "step": 38960
    },
    {
      "epoch": 0.8868306670004323,
      "grad_norm": 0.00191497802734375,
      "learning_rate": 2.3100501579045144e-06,
      "loss": 1.453,
      "step": 38970
    },
    {
      "epoch": 0.8870582345310971,
      "grad_norm": 1.515625,
      "learning_rate": 2.305405907486532e-06,
      "loss": 0.4114,
      "step": 38980
    },
    {
      "epoch": 0.8872858020617618,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3007616570685493e-06,
      "loss": 0.6132,
      "step": 38990
    },
    {
      "epoch": 0.8875133695924265,
      "grad_norm": 0.875,
      "learning_rate": 2.296117406650567e-06,
      "loss": 0.7628,
      "step": 39000
    },
    {
      "epoch": 0.8877409371230913,
      "grad_norm": 146.0,
      "learning_rate": 2.2914731562325843e-06,
      "loss": 0.8817,
      "step": 39010
    },
    {
      "epoch": 0.887968504653756,
      "grad_norm": 87.5,
      "learning_rate": 2.2868289058146017e-06,
      "loss": 0.4443,
      "step": 39020
    },
    {
      "epoch": 0.8881960721844208,
      "grad_norm": 0.04736328125,
      "learning_rate": 2.2821846553966192e-06,
      "loss": 0.2328,
      "step": 39030
    },
    {
      "epoch": 0.8884236397150854,
      "grad_norm": 124.5,
      "learning_rate": 2.2775404049786367e-06,
      "loss": 0.4294,
      "step": 39040
    },
    {
      "epoch": 0.8886512072457502,
      "grad_norm": 25.875,
      "learning_rate": 2.272896154560654e-06,
      "loss": 0.7789,
      "step": 39050
    },
    {
      "epoch": 0.8888787747764149,
      "grad_norm": 0.275390625,
      "learning_rate": 2.2682519041426717e-06,
      "loss": 1.4648,
      "step": 39060
    },
    {
      "epoch": 0.8891063423070796,
      "grad_norm": 0.0206298828125,
      "learning_rate": 2.263607653724689e-06,
      "loss": 0.3648,
      "step": 39070
    },
    {
      "epoch": 0.8893339098377443,
      "grad_norm": 0.1220703125,
      "learning_rate": 2.2589634033067066e-06,
      "loss": 0.4381,
      "step": 39080
    },
    {
      "epoch": 0.8895614773684091,
      "grad_norm": 0.026611328125,
      "learning_rate": 2.254319152888724e-06,
      "loss": 1.4355,
      "step": 39090
    },
    {
      "epoch": 0.8897890448990738,
      "grad_norm": 0.04345703125,
      "learning_rate": 2.2496749024707416e-06,
      "loss": 1.5589,
      "step": 39100
    },
    {
      "epoch": 0.8900166124297385,
      "grad_norm": 0.00830078125,
      "learning_rate": 2.245030652052759e-06,
      "loss": 0.4328,
      "step": 39110
    },
    {
      "epoch": 0.8902441799604033,
      "grad_norm": 141.0,
      "learning_rate": 2.240386401634776e-06,
      "loss": 1.0442,
      "step": 39120
    },
    {
      "epoch": 0.8904717474910679,
      "grad_norm": 129.0,
      "learning_rate": 2.2357421512167936e-06,
      "loss": 0.6684,
      "step": 39130
    },
    {
      "epoch": 0.8906993150217327,
      "grad_norm": 0.482421875,
      "learning_rate": 2.231097900798811e-06,
      "loss": 0.9156,
      "step": 39140
    },
    {
      "epoch": 0.8909268825523974,
      "grad_norm": 50.5,
      "learning_rate": 2.2264536503808285e-06,
      "loss": 0.6132,
      "step": 39150
    },
    {
      "epoch": 0.8911544500830622,
      "grad_norm": 1.59375,
      "learning_rate": 2.221809399962846e-06,
      "loss": 0.6362,
      "step": 39160
    },
    {
      "epoch": 0.8913820176137268,
      "grad_norm": 175.0,
      "learning_rate": 2.217165149544864e-06,
      "loss": 0.834,
      "step": 39170
    },
    {
      "epoch": 0.8916095851443916,
      "grad_norm": 115.5,
      "learning_rate": 2.2125208991268814e-06,
      "loss": 0.411,
      "step": 39180
    },
    {
      "epoch": 0.8918371526750564,
      "grad_norm": 197.0,
      "learning_rate": 2.2078766487088985e-06,
      "loss": 0.7732,
      "step": 39190
    },
    {
      "epoch": 0.892064720205721,
      "grad_norm": 243.0,
      "learning_rate": 2.203232398290916e-06,
      "loss": 0.994,
      "step": 39200
    },
    {
      "epoch": 0.8922922877363858,
      "grad_norm": 27.875,
      "learning_rate": 2.1985881478729334e-06,
      "loss": 0.3322,
      "step": 39210
    },
    {
      "epoch": 0.8925198552670505,
      "grad_norm": 26.0,
      "learning_rate": 2.193943897454951e-06,
      "loss": 1.1581,
      "step": 39220
    },
    {
      "epoch": 0.8927474227977152,
      "grad_norm": 172.0,
      "learning_rate": 2.1892996470369684e-06,
      "loss": 0.8136,
      "step": 39230
    },
    {
      "epoch": 0.8929749903283799,
      "grad_norm": 1.1484375,
      "learning_rate": 2.184655396618986e-06,
      "loss": 0.8088,
      "step": 39240
    },
    {
      "epoch": 0.8932025578590447,
      "grad_norm": 94.0,
      "learning_rate": 2.1800111462010033e-06,
      "loss": 0.5515,
      "step": 39250
    },
    {
      "epoch": 0.8934301253897095,
      "grad_norm": 260.0,
      "learning_rate": 2.175366895783021e-06,
      "loss": 1.0181,
      "step": 39260
    },
    {
      "epoch": 0.8936576929203741,
      "grad_norm": 45.25,
      "learning_rate": 2.1707226453650383e-06,
      "loss": 1.3966,
      "step": 39270
    },
    {
      "epoch": 0.8938852604510389,
      "grad_norm": 146.0,
      "learning_rate": 2.1660783949470558e-06,
      "loss": 1.2331,
      "step": 39280
    },
    {
      "epoch": 0.8941128279817036,
      "grad_norm": 215.0,
      "learning_rate": 2.1614341445290732e-06,
      "loss": 0.3853,
      "step": 39290
    },
    {
      "epoch": 0.8943403955123683,
      "grad_norm": 0.0498046875,
      "learning_rate": 2.1567898941110907e-06,
      "loss": 0.5673,
      "step": 39300
    },
    {
      "epoch": 0.894567963043033,
      "grad_norm": 0.00118255615234375,
      "learning_rate": 2.152145643693108e-06,
      "loss": 0.9072,
      "step": 39310
    },
    {
      "epoch": 0.8947955305736978,
      "grad_norm": 93.5,
      "learning_rate": 2.1475013932751257e-06,
      "loss": 0.3777,
      "step": 39320
    },
    {
      "epoch": 0.8950230981043624,
      "grad_norm": 100.0,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.5091,
      "step": 39330
    },
    {
      "epoch": 0.8952506656350272,
      "grad_norm": 6.818771362304688e-05,
      "learning_rate": 2.13821289243916e-06,
      "loss": 0.3332,
      "step": 39340
    },
    {
      "epoch": 0.895478233165692,
      "grad_norm": 0.70703125,
      "learning_rate": 2.133568642021178e-06,
      "loss": 0.2085,
      "step": 39350
    },
    {
      "epoch": 0.8957058006963566,
      "grad_norm": 0.00506591796875,
      "learning_rate": 2.1289243916031956e-06,
      "loss": 0.3323,
      "step": 39360
    },
    {
      "epoch": 0.8959333682270214,
      "grad_norm": 81.0,
      "learning_rate": 2.124280141185213e-06,
      "loss": 0.3852,
      "step": 39370
    },
    {
      "epoch": 0.8961609357576861,
      "grad_norm": 0.0311279296875,
      "learning_rate": 2.1196358907672305e-06,
      "loss": 0.9096,
      "step": 39380
    },
    {
      "epoch": 0.8963885032883508,
      "grad_norm": 4.9375,
      "learning_rate": 2.114991640349248e-06,
      "loss": 1.2164,
      "step": 39390
    },
    {
      "epoch": 0.8966160708190155,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.110347389931265e-06,
      "loss": 0.7443,
      "step": 39400
    },
    {
      "epoch": 0.8968436383496803,
      "grad_norm": 106.5,
      "learning_rate": 2.1057031395132826e-06,
      "loss": 0.422,
      "step": 39410
    },
    {
      "epoch": 0.897071205880345,
      "grad_norm": 0.0115966796875,
      "learning_rate": 2.1010588890953e-06,
      "loss": 0.9382,
      "step": 39420
    },
    {
      "epoch": 0.8972987734110097,
      "grad_norm": 64.0,
      "learning_rate": 2.0964146386773175e-06,
      "loss": 0.5984,
      "step": 39430
    },
    {
      "epoch": 0.8975263409416745,
      "grad_norm": 151.0,
      "learning_rate": 2.091770388259335e-06,
      "loss": 0.8815,
      "step": 39440
    },
    {
      "epoch": 0.8977539084723392,
      "grad_norm": 348.0,
      "learning_rate": 2.0871261378413525e-06,
      "loss": 1.0328,
      "step": 39450
    },
    {
      "epoch": 0.8979814760030039,
      "grad_norm": 0.0027923583984375,
      "learning_rate": 2.08248188742337e-06,
      "loss": 0.6886,
      "step": 39460
    },
    {
      "epoch": 0.8982090435336686,
      "grad_norm": 116.0,
      "learning_rate": 2.0778376370053874e-06,
      "loss": 0.7691,
      "step": 39470
    },
    {
      "epoch": 0.8984366110643334,
      "grad_norm": 48.0,
      "learning_rate": 2.073193386587405e-06,
      "loss": 0.3808,
      "step": 39480
    },
    {
      "epoch": 0.898664178594998,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.0685491361694224e-06,
      "loss": 0.3899,
      "step": 39490
    },
    {
      "epoch": 0.8988917461256628,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.06390488575144e-06,
      "loss": 0.2678,
      "step": 39500
    },
    {
      "epoch": 0.8991193136563275,
      "grad_norm": 31.125,
      "learning_rate": 2.0592606353334573e-06,
      "loss": 0.2722,
      "step": 39510
    },
    {
      "epoch": 0.8993468811869922,
      "grad_norm": 5.25,
      "learning_rate": 2.054616384915475e-06,
      "loss": 0.6616,
      "step": 39520
    },
    {
      "epoch": 0.899574448717657,
      "grad_norm": 179.0,
      "learning_rate": 2.0499721344974923e-06,
      "loss": 1.2084,
      "step": 39530
    },
    {
      "epoch": 0.8998020162483217,
      "grad_norm": 102.5,
      "learning_rate": 2.0453278840795098e-06,
      "loss": 0.8453,
      "step": 39540
    },
    {
      "epoch": 0.9000295837789865,
      "grad_norm": 52.0,
      "learning_rate": 2.0406836336615273e-06,
      "loss": 0.5723,
      "step": 39550
    },
    {
      "epoch": 0.9002571513096511,
      "grad_norm": 251.0,
      "learning_rate": 2.0360393832435447e-06,
      "loss": 1.3839,
      "step": 39560
    },
    {
      "epoch": 0.9004847188403159,
      "grad_norm": 31.875,
      "learning_rate": 2.0313951328255622e-06,
      "loss": 0.6992,
      "step": 39570
    },
    {
      "epoch": 0.9007122863709806,
      "grad_norm": 112.5,
      "learning_rate": 2.0267508824075797e-06,
      "loss": 1.8413,
      "step": 39580
    },
    {
      "epoch": 0.9009398539016453,
      "grad_norm": 105.0,
      "learning_rate": 2.022106631989597e-06,
      "loss": 0.617,
      "step": 39590
    },
    {
      "epoch": 0.90116742143231,
      "grad_norm": 38.75,
      "learning_rate": 2.0174623815716146e-06,
      "loss": 1.247,
      "step": 39600
    },
    {
      "epoch": 0.9013949889629748,
      "grad_norm": 86.0,
      "learning_rate": 2.0128181311536317e-06,
      "loss": 0.4337,
      "step": 39610
    },
    {
      "epoch": 0.9016225564936395,
      "grad_norm": 148.0,
      "learning_rate": 2.008173880735649e-06,
      "loss": 0.685,
      "step": 39620
    },
    {
      "epoch": 0.9018501240243042,
      "grad_norm": 2.171875,
      "learning_rate": 2.0035296303176667e-06,
      "loss": 0.4842,
      "step": 39630
    },
    {
      "epoch": 0.902077691554969,
      "grad_norm": 260.0,
      "learning_rate": 1.998885379899684e-06,
      "loss": 1.1652,
      "step": 39640
    },
    {
      "epoch": 0.9023052590856336,
      "grad_norm": 37.5,
      "learning_rate": 1.994241129481702e-06,
      "loss": 0.6106,
      "step": 39650
    },
    {
      "epoch": 0.9025328266162984,
      "grad_norm": 102.0,
      "learning_rate": 1.9895968790637195e-06,
      "loss": 0.7818,
      "step": 39660
    },
    {
      "epoch": 0.9027603941469631,
      "grad_norm": 85.0,
      "learning_rate": 1.984952628645737e-06,
      "loss": 1.0692,
      "step": 39670
    },
    {
      "epoch": 0.9029879616776278,
      "grad_norm": 42.0,
      "learning_rate": 1.980308378227754e-06,
      "loss": 0.8222,
      "step": 39680
    },
    {
      "epoch": 0.9032155292082925,
      "grad_norm": 74.0,
      "learning_rate": 1.9756641278097715e-06,
      "loss": 0.3199,
      "step": 39690
    },
    {
      "epoch": 0.9034430967389573,
      "grad_norm": 0.000537872314453125,
      "learning_rate": 1.971019877391789e-06,
      "loss": 0.184,
      "step": 39700
    },
    {
      "epoch": 0.9036706642696221,
      "grad_norm": 165.0,
      "learning_rate": 1.9663756269738065e-06,
      "loss": 1.0572,
      "step": 39710
    },
    {
      "epoch": 0.9038982318002867,
      "grad_norm": 59.5,
      "learning_rate": 1.961731376555824e-06,
      "loss": 0.4825,
      "step": 39720
    },
    {
      "epoch": 0.9041257993309515,
      "grad_norm": 6.6875,
      "learning_rate": 1.9570871261378414e-06,
      "loss": 1.1153,
      "step": 39730
    },
    {
      "epoch": 0.9043533668616162,
      "grad_norm": 45.75,
      "learning_rate": 1.952442875719859e-06,
      "loss": 1.3418,
      "step": 39740
    },
    {
      "epoch": 0.9045809343922809,
      "grad_norm": 0.0064697265625,
      "learning_rate": 1.9477986253018764e-06,
      "loss": 0.6987,
      "step": 39750
    },
    {
      "epoch": 0.9048085019229456,
      "grad_norm": 370.0,
      "learning_rate": 1.943154374883894e-06,
      "loss": 0.4526,
      "step": 39760
    },
    {
      "epoch": 0.9050360694536104,
      "grad_norm": 0.0201416015625,
      "learning_rate": 1.9385101244659114e-06,
      "loss": 0.286,
      "step": 39770
    },
    {
      "epoch": 0.905263636984275,
      "grad_norm": 123.0,
      "learning_rate": 1.933865874047929e-06,
      "loss": 0.5775,
      "step": 39780
    },
    {
      "epoch": 0.9054912045149398,
      "grad_norm": 0.62109375,
      "learning_rate": 1.9292216236299463e-06,
      "loss": 1.9849,
      "step": 39790
    },
    {
      "epoch": 0.9057187720456046,
      "grad_norm": 41.75,
      "learning_rate": 1.924577373211964e-06,
      "loss": 0.4552,
      "step": 39800
    },
    {
      "epoch": 0.9059463395762692,
      "grad_norm": 101.0,
      "learning_rate": 1.9199331227939813e-06,
      "loss": 1.2271,
      "step": 39810
    },
    {
      "epoch": 0.906173907106934,
      "grad_norm": 5.1875,
      "learning_rate": 1.9152888723759988e-06,
      "loss": 0.6729,
      "step": 39820
    },
    {
      "epoch": 0.9064014746375987,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9106446219580162e-06,
      "loss": 1.6097,
      "step": 39830
    },
    {
      "epoch": 0.9066290421682635,
      "grad_norm": 0.0220947265625,
      "learning_rate": 1.9060003715400335e-06,
      "loss": 0.8023,
      "step": 39840
    },
    {
      "epoch": 0.9068566096989281,
      "grad_norm": 2496.0,
      "learning_rate": 1.901356121122051e-06,
      "loss": 0.7743,
      "step": 39850
    },
    {
      "epoch": 0.9070841772295929,
      "grad_norm": 8.5,
      "learning_rate": 1.8967118707040685e-06,
      "loss": 0.9532,
      "step": 39860
    },
    {
      "epoch": 0.9073117447602576,
      "grad_norm": 0.2890625,
      "learning_rate": 1.8920676202860861e-06,
      "loss": 1.4144,
      "step": 39870
    },
    {
      "epoch": 0.9075393122909223,
      "grad_norm": 87.0,
      "learning_rate": 1.8874233698681036e-06,
      "loss": 1.1352,
      "step": 39880
    },
    {
      "epoch": 0.9077668798215871,
      "grad_norm": 0.00994873046875,
      "learning_rate": 1.882779119450121e-06,
      "loss": 0.3861,
      "step": 39890
    },
    {
      "epoch": 0.9079944473522518,
      "grad_norm": 0.005340576171875,
      "learning_rate": 1.8781348690321384e-06,
      "loss": 0.3776,
      "step": 39900
    },
    {
      "epoch": 0.9082220148829165,
      "grad_norm": 354.0,
      "learning_rate": 1.8734906186141558e-06,
      "loss": 1.0079,
      "step": 39910
    },
    {
      "epoch": 0.9084495824135812,
      "grad_norm": 249.0,
      "learning_rate": 1.8688463681961733e-06,
      "loss": 0.2589,
      "step": 39920
    },
    {
      "epoch": 0.908677149944246,
      "grad_norm": 203.0,
      "learning_rate": 1.8642021177781908e-06,
      "loss": 0.7516,
      "step": 39930
    },
    {
      "epoch": 0.9089047174749106,
      "grad_norm": 138.0,
      "learning_rate": 1.8595578673602083e-06,
      "loss": 0.5554,
      "step": 39940
    },
    {
      "epoch": 0.9091322850055754,
      "grad_norm": 82.0,
      "learning_rate": 1.8549136169422258e-06,
      "loss": 1.1278,
      "step": 39950
    },
    {
      "epoch": 0.9093598525362401,
      "grad_norm": 1.0390625,
      "learning_rate": 1.8502693665242432e-06,
      "loss": 0.6187,
      "step": 39960
    },
    {
      "epoch": 0.9095874200669049,
      "grad_norm": 0.01068115234375,
      "learning_rate": 1.8456251161062605e-06,
      "loss": 0.7118,
      "step": 39970
    },
    {
      "epoch": 0.9098149875975696,
      "grad_norm": 79.0,
      "learning_rate": 1.840980865688278e-06,
      "loss": 1.0158,
      "step": 39980
    },
    {
      "epoch": 0.9100425551282343,
      "grad_norm": 0.001312255859375,
      "learning_rate": 1.8363366152702955e-06,
      "loss": 0.4229,
      "step": 39990
    },
    {
      "epoch": 0.9102701226588991,
      "grad_norm": 0.0027618408203125,
      "learning_rate": 1.831692364852313e-06,
      "loss": 0.2937,
      "step": 40000
    },
    {
      "epoch": 0.9104976901895637,
      "grad_norm": 25.875,
      "learning_rate": 1.8270481144343304e-06,
      "loss": 0.1502,
      "step": 40010
    },
    {
      "epoch": 0.9107252577202285,
      "grad_norm": 220.0,
      "learning_rate": 1.822403864016348e-06,
      "loss": 0.9015,
      "step": 40020
    },
    {
      "epoch": 0.9109528252508932,
      "grad_norm": 4.6875,
      "learning_rate": 1.8177596135983656e-06,
      "loss": 0.3266,
      "step": 40030
    },
    {
      "epoch": 0.9111803927815579,
      "grad_norm": 192.0,
      "learning_rate": 1.8131153631803826e-06,
      "loss": 1.4997,
      "step": 40040
    },
    {
      "epoch": 0.9114079603122226,
      "grad_norm": 51.5,
      "learning_rate": 1.8084711127624003e-06,
      "loss": 0.5175,
      "step": 40050
    },
    {
      "epoch": 0.9116355278428874,
      "grad_norm": 0.04052734375,
      "learning_rate": 1.8038268623444178e-06,
      "loss": 0.2849,
      "step": 40060
    },
    {
      "epoch": 0.9118630953735521,
      "grad_norm": 6.4375,
      "learning_rate": 1.7991826119264353e-06,
      "loss": 2.3122,
      "step": 40070
    },
    {
      "epoch": 0.9120906629042168,
      "grad_norm": 178.0,
      "learning_rate": 1.7945383615084528e-06,
      "loss": 0.773,
      "step": 40080
    },
    {
      "epoch": 0.9123182304348816,
      "grad_norm": 0.443359375,
      "learning_rate": 1.7898941110904702e-06,
      "loss": 0.5007,
      "step": 40090
    },
    {
      "epoch": 0.9125457979655462,
      "grad_norm": 0.92578125,
      "learning_rate": 1.7852498606724877e-06,
      "loss": 0.5697,
      "step": 40100
    },
    {
      "epoch": 0.912773365496211,
      "grad_norm": 54.0,
      "learning_rate": 1.780605610254505e-06,
      "loss": 0.6472,
      "step": 40110
    },
    {
      "epoch": 0.9130009330268757,
      "grad_norm": 139.0,
      "learning_rate": 1.7759613598365225e-06,
      "loss": 1.1982,
      "step": 40120
    },
    {
      "epoch": 0.9132285005575405,
      "grad_norm": 0.004730224609375,
      "learning_rate": 1.77131710941854e-06,
      "loss": 0.2094,
      "step": 40130
    },
    {
      "epoch": 0.9134560680882051,
      "grad_norm": 75.5,
      "learning_rate": 1.7666728590005574e-06,
      "loss": 0.3811,
      "step": 40140
    },
    {
      "epoch": 0.9136836356188699,
      "grad_norm": 186.0,
      "learning_rate": 1.762028608582575e-06,
      "loss": 1.2799,
      "step": 40150
    },
    {
      "epoch": 0.9139112031495347,
      "grad_norm": 0.6484375,
      "learning_rate": 1.7573843581645924e-06,
      "loss": 0.7681,
      "step": 40160
    },
    {
      "epoch": 0.9141387706801993,
      "grad_norm": 0.0038604736328125,
      "learning_rate": 1.7527401077466099e-06,
      "loss": 2.2484,
      "step": 40170
    },
    {
      "epoch": 0.9143663382108641,
      "grad_norm": 14.3125,
      "learning_rate": 1.7480958573286271e-06,
      "loss": 0.6297,
      "step": 40180
    },
    {
      "epoch": 0.9145939057415288,
      "grad_norm": 3.9375,
      "learning_rate": 1.7434516069106446e-06,
      "loss": 0.3909,
      "step": 40190
    },
    {
      "epoch": 0.9148214732721935,
      "grad_norm": 0.5859375,
      "learning_rate": 1.7388073564926623e-06,
      "loss": 1.346,
      "step": 40200
    },
    {
      "epoch": 0.9150490408028582,
      "grad_norm": 186.0,
      "learning_rate": 1.7341631060746798e-06,
      "loss": 0.6037,
      "step": 40210
    },
    {
      "epoch": 0.915276608333523,
      "grad_norm": 1.4375,
      "learning_rate": 1.7295188556566973e-06,
      "loss": 0.4518,
      "step": 40220
    },
    {
      "epoch": 0.9155041758641878,
      "grad_norm": 0.0703125,
      "learning_rate": 1.7248746052387147e-06,
      "loss": 0.7153,
      "step": 40230
    },
    {
      "epoch": 0.9157317433948524,
      "grad_norm": 0.026611328125,
      "learning_rate": 1.7202303548207322e-06,
      "loss": 0.6838,
      "step": 40240
    },
    {
      "epoch": 0.9159593109255172,
      "grad_norm": 195.0,
      "learning_rate": 1.7155861044027495e-06,
      "loss": 0.5748,
      "step": 40250
    },
    {
      "epoch": 0.9161868784561819,
      "grad_norm": 125.5,
      "learning_rate": 1.710941853984767e-06,
      "loss": 1.8679,
      "step": 40260
    },
    {
      "epoch": 0.9164144459868466,
      "grad_norm": 126.5,
      "learning_rate": 1.7062976035667844e-06,
      "loss": 1.228,
      "step": 40270
    },
    {
      "epoch": 0.9166420135175113,
      "grad_norm": 0.0145263671875,
      "learning_rate": 1.701653353148802e-06,
      "loss": 0.1398,
      "step": 40280
    },
    {
      "epoch": 0.9168695810481761,
      "grad_norm": 154.0,
      "learning_rate": 1.6970091027308194e-06,
      "loss": 1.3622,
      "step": 40290
    },
    {
      "epoch": 0.9170971485788407,
      "grad_norm": 0.01031494140625,
      "learning_rate": 1.6923648523128369e-06,
      "loss": 0.6799,
      "step": 40300
    },
    {
      "epoch": 0.9173247161095055,
      "grad_norm": 202.0,
      "learning_rate": 1.6877206018948543e-06,
      "loss": 1.6579,
      "step": 40310
    },
    {
      "epoch": 0.9175522836401703,
      "grad_norm": 528.0,
      "learning_rate": 1.6830763514768716e-06,
      "loss": 0.9635,
      "step": 40320
    },
    {
      "epoch": 0.9177798511708349,
      "grad_norm": 77.0,
      "learning_rate": 1.678432101058889e-06,
      "loss": 0.7486,
      "step": 40330
    },
    {
      "epoch": 0.9180074187014997,
      "grad_norm": 490.0,
      "learning_rate": 1.6737878506409066e-06,
      "loss": 0.9566,
      "step": 40340
    },
    {
      "epoch": 0.9182349862321644,
      "grad_norm": 83.5,
      "learning_rate": 1.669143600222924e-06,
      "loss": 0.8304,
      "step": 40350
    },
    {
      "epoch": 0.9184625537628291,
      "grad_norm": 360.0,
      "learning_rate": 1.6644993498049417e-06,
      "loss": 0.6268,
      "step": 40360
    },
    {
      "epoch": 0.9186901212934938,
      "grad_norm": 0.02587890625,
      "learning_rate": 1.6598550993869592e-06,
      "loss": 0.6046,
      "step": 40370
    },
    {
      "epoch": 0.9189176888241586,
      "grad_norm": 239.0,
      "learning_rate": 1.6552108489689767e-06,
      "loss": 0.5684,
      "step": 40380
    },
    {
      "epoch": 0.9191452563548232,
      "grad_norm": 97.5,
      "learning_rate": 1.650566598550994e-06,
      "loss": 0.9473,
      "step": 40390
    },
    {
      "epoch": 0.919372823885488,
      "grad_norm": 9.0,
      "learning_rate": 1.6459223481330114e-06,
      "loss": 0.4115,
      "step": 40400
    },
    {
      "epoch": 0.9196003914161528,
      "grad_norm": 52.25,
      "learning_rate": 1.641278097715029e-06,
      "loss": 0.6657,
      "step": 40410
    },
    {
      "epoch": 0.9198279589468175,
      "grad_norm": 46.0,
      "learning_rate": 1.6366338472970464e-06,
      "loss": 1.3878,
      "step": 40420
    },
    {
      "epoch": 0.9200555264774822,
      "grad_norm": 0.04248046875,
      "learning_rate": 1.6319895968790639e-06,
      "loss": 1.0309,
      "step": 40430
    },
    {
      "epoch": 0.9202830940081469,
      "grad_norm": 0.06201171875,
      "learning_rate": 1.6273453464610814e-06,
      "loss": 1.096,
      "step": 40440
    },
    {
      "epoch": 0.9205106615388117,
      "grad_norm": 1.0859375,
      "learning_rate": 1.6227010960430988e-06,
      "loss": 0.6528,
      "step": 40450
    },
    {
      "epoch": 0.9207382290694763,
      "grad_norm": 0.0023345947265625,
      "learning_rate": 1.618056845625116e-06,
      "loss": 1.2704,
      "step": 40460
    },
    {
      "epoch": 0.9209657966001411,
      "grad_norm": 0.46875,
      "learning_rate": 1.6134125952071336e-06,
      "loss": 0.6184,
      "step": 40470
    },
    {
      "epoch": 0.9211933641308058,
      "grad_norm": 36.25,
      "learning_rate": 1.608768344789151e-06,
      "loss": 0.9206,
      "step": 40480
    },
    {
      "epoch": 0.9214209316614705,
      "grad_norm": 0.056640625,
      "learning_rate": 1.6041240943711685e-06,
      "loss": 0.8128,
      "step": 40490
    },
    {
      "epoch": 0.9216484991921353,
      "grad_norm": 239.0,
      "learning_rate": 1.599479843953186e-06,
      "loss": 0.9394,
      "step": 40500
    },
    {
      "epoch": 0.9218760667228,
      "grad_norm": 193.0,
      "learning_rate": 1.5948355935352037e-06,
      "loss": 0.5975,
      "step": 40510
    },
    {
      "epoch": 0.9221036342534648,
      "grad_norm": 125.5,
      "learning_rate": 1.5901913431172212e-06,
      "loss": 0.9721,
      "step": 40520
    },
    {
      "epoch": 0.9223312017841294,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.5855470926992384e-06,
      "loss": 0.3291,
      "step": 40530
    },
    {
      "epoch": 0.9225587693147942,
      "grad_norm": 2.0625,
      "learning_rate": 1.580902842281256e-06,
      "loss": 0.4145,
      "step": 40540
    },
    {
      "epoch": 0.9227863368454589,
      "grad_norm": 0.0419921875,
      "learning_rate": 1.5762585918632734e-06,
      "loss": 0.5751,
      "step": 40550
    },
    {
      "epoch": 0.9230139043761236,
      "grad_norm": 140.0,
      "learning_rate": 1.5716143414452909e-06,
      "loss": 0.9489,
      "step": 40560
    },
    {
      "epoch": 0.9232414719067883,
      "grad_norm": 496.0,
      "learning_rate": 1.5669700910273084e-06,
      "loss": 1.4961,
      "step": 40570
    },
    {
      "epoch": 0.9234690394374531,
      "grad_norm": 113.0,
      "learning_rate": 1.5623258406093258e-06,
      "loss": 1.2155,
      "step": 40580
    },
    {
      "epoch": 0.9236966069681178,
      "grad_norm": 0.00360107421875,
      "learning_rate": 1.5576815901913433e-06,
      "loss": 0.8227,
      "step": 40590
    },
    {
      "epoch": 0.9239241744987825,
      "grad_norm": 2.15625,
      "learning_rate": 1.5530373397733606e-06,
      "loss": 1.0849,
      "step": 40600
    },
    {
      "epoch": 0.9241517420294473,
      "grad_norm": 0.06298828125,
      "learning_rate": 1.548393089355378e-06,
      "loss": 1.0473,
      "step": 40610
    },
    {
      "epoch": 0.9243793095601119,
      "grad_norm": 62.5,
      "learning_rate": 1.5437488389373955e-06,
      "loss": 1.0451,
      "step": 40620
    },
    {
      "epoch": 0.9246068770907767,
      "grad_norm": 146.0,
      "learning_rate": 1.539104588519413e-06,
      "loss": 0.4225,
      "step": 40630
    },
    {
      "epoch": 0.9248344446214414,
      "grad_norm": 0.03857421875,
      "learning_rate": 1.5344603381014305e-06,
      "loss": 0.3821,
      "step": 40640
    },
    {
      "epoch": 0.9250620121521062,
      "grad_norm": 0.51171875,
      "learning_rate": 1.529816087683448e-06,
      "loss": 1.0289,
      "step": 40650
    },
    {
      "epoch": 0.9252895796827708,
      "grad_norm": 35.75,
      "learning_rate": 1.5251718372654657e-06,
      "loss": 0.4319,
      "step": 40660
    },
    {
      "epoch": 0.9255171472134356,
      "grad_norm": 1.65625,
      "learning_rate": 1.5205275868474827e-06,
      "loss": 1.8188,
      "step": 40670
    },
    {
      "epoch": 0.9257447147441004,
      "grad_norm": 246.0,
      "learning_rate": 1.5158833364295002e-06,
      "loss": 0.4857,
      "step": 40680
    },
    {
      "epoch": 0.925972282274765,
      "grad_norm": 111.5,
      "learning_rate": 1.5112390860115179e-06,
      "loss": 0.4388,
      "step": 40690
    },
    {
      "epoch": 0.9261998498054298,
      "grad_norm": 149.0,
      "learning_rate": 1.5065948355935354e-06,
      "loss": 1.21,
      "step": 40700
    },
    {
      "epoch": 0.9264274173360945,
      "grad_norm": 25.875,
      "learning_rate": 1.5019505851755528e-06,
      "loss": 1.0039,
      "step": 40710
    },
    {
      "epoch": 0.9266549848667592,
      "grad_norm": 0.0181884765625,
      "learning_rate": 1.4973063347575703e-06,
      "loss": 0.6732,
      "step": 40720
    },
    {
      "epoch": 0.9268825523974239,
      "grad_norm": 0.068359375,
      "learning_rate": 1.4926620843395878e-06,
      "loss": 0.6945,
      "step": 40730
    },
    {
      "epoch": 0.9271101199280887,
      "grad_norm": 38.5,
      "learning_rate": 1.488017833921605e-06,
      "loss": 0.6279,
      "step": 40740
    },
    {
      "epoch": 0.9273376874587533,
      "grad_norm": 57.5,
      "learning_rate": 1.4833735835036226e-06,
      "loss": 1.4734,
      "step": 40750
    },
    {
      "epoch": 0.9275652549894181,
      "grad_norm": 2.421875,
      "learning_rate": 1.47872933308564e-06,
      "loss": 0.4427,
      "step": 40760
    },
    {
      "epoch": 0.9277928225200829,
      "grad_norm": 215.0,
      "learning_rate": 1.4740850826676575e-06,
      "loss": 0.6717,
      "step": 40770
    },
    {
      "epoch": 0.9280203900507475,
      "grad_norm": 246.0,
      "learning_rate": 1.469440832249675e-06,
      "loss": 0.6142,
      "step": 40780
    },
    {
      "epoch": 0.9282479575814123,
      "grad_norm": 440.0,
      "learning_rate": 1.4647965818316925e-06,
      "loss": 1.0173,
      "step": 40790
    },
    {
      "epoch": 0.928475525112077,
      "grad_norm": 58.75,
      "learning_rate": 1.46015233141371e-06,
      "loss": 0.802,
      "step": 40800
    },
    {
      "epoch": 0.9287030926427418,
      "grad_norm": 154.0,
      "learning_rate": 1.4555080809957276e-06,
      "loss": 0.5632,
      "step": 40810
    },
    {
      "epoch": 0.9289306601734064,
      "grad_norm": 22.25,
      "learning_rate": 1.4508638305777447e-06,
      "loss": 0.6134,
      "step": 40820
    },
    {
      "epoch": 0.9291582277040712,
      "grad_norm": 0.005218505859375,
      "learning_rate": 1.4462195801597622e-06,
      "loss": 0.8432,
      "step": 40830
    },
    {
      "epoch": 0.9293857952347359,
      "grad_norm": 416.0,
      "learning_rate": 1.4415753297417799e-06,
      "loss": 0.9989,
      "step": 40840
    },
    {
      "epoch": 0.9296133627654006,
      "grad_norm": 0.002471923828125,
      "learning_rate": 1.4369310793237973e-06,
      "loss": 0.9845,
      "step": 40850
    },
    {
      "epoch": 0.9298409302960654,
      "grad_norm": 0.00154876708984375,
      "learning_rate": 1.4322868289058148e-06,
      "loss": 0.8265,
      "step": 40860
    },
    {
      "epoch": 0.9300684978267301,
      "grad_norm": 85.0,
      "learning_rate": 1.4276425784878323e-06,
      "loss": 0.7665,
      "step": 40870
    },
    {
      "epoch": 0.9302960653573948,
      "grad_norm": 151.0,
      "learning_rate": 1.4229983280698498e-06,
      "loss": 1.0652,
      "step": 40880
    },
    {
      "epoch": 0.9305236328880595,
      "grad_norm": 0.0306396484375,
      "learning_rate": 1.418354077651867e-06,
      "loss": 1.1202,
      "step": 40890
    },
    {
      "epoch": 0.9307512004187243,
      "grad_norm": 0.05712890625,
      "learning_rate": 1.4137098272338845e-06,
      "loss": 0.5472,
      "step": 40900
    },
    {
      "epoch": 0.9309787679493889,
      "grad_norm": 96.5,
      "learning_rate": 1.409065576815902e-06,
      "loss": 0.32,
      "step": 40910
    },
    {
      "epoch": 0.9312063354800537,
      "grad_norm": 0.10400390625,
      "learning_rate": 1.4044213263979195e-06,
      "loss": 0.7863,
      "step": 40920
    },
    {
      "epoch": 0.9314339030107184,
      "grad_norm": 167.0,
      "learning_rate": 1.399777075979937e-06,
      "loss": 0.8551,
      "step": 40930
    },
    {
      "epoch": 0.9316614705413832,
      "grad_norm": 3.515625,
      "learning_rate": 1.3951328255619544e-06,
      "loss": 0.2319,
      "step": 40940
    },
    {
      "epoch": 0.9318890380720479,
      "grad_norm": 99.5,
      "learning_rate": 1.390488575143972e-06,
      "loss": 0.7769,
      "step": 40950
    },
    {
      "epoch": 0.9321166056027126,
      "grad_norm": 0.06103515625,
      "learning_rate": 1.3858443247259892e-06,
      "loss": 0.4053,
      "step": 40960
    },
    {
      "epoch": 0.9323441731333774,
      "grad_norm": 392.0,
      "learning_rate": 1.3812000743080067e-06,
      "loss": 0.7167,
      "step": 40970
    },
    {
      "epoch": 0.932571740664042,
      "grad_norm": 624.0,
      "learning_rate": 1.3765558238900241e-06,
      "loss": 0.6587,
      "step": 40980
    },
    {
      "epoch": 0.9327993081947068,
      "grad_norm": 352.0,
      "learning_rate": 1.3719115734720418e-06,
      "loss": 1.5605,
      "step": 40990
    },
    {
      "epoch": 0.9330268757253715,
      "grad_norm": 0.0002651214599609375,
      "learning_rate": 1.3672673230540593e-06,
      "loss": 0.0836,
      "step": 41000
    },
    {
      "epoch": 0.9332544432560362,
      "grad_norm": 156.0,
      "learning_rate": 1.3626230726360768e-06,
      "loss": 1.4145,
      "step": 41010
    },
    {
      "epoch": 0.9334820107867009,
      "grad_norm": 58.0,
      "learning_rate": 1.3579788222180943e-06,
      "loss": 0.9058,
      "step": 41020
    },
    {
      "epoch": 0.9337095783173657,
      "grad_norm": 39.25,
      "learning_rate": 1.3533345718001115e-06,
      "loss": 0.9311,
      "step": 41030
    },
    {
      "epoch": 0.9339371458480304,
      "grad_norm": 215.0,
      "learning_rate": 1.348690321382129e-06,
      "loss": 0.4867,
      "step": 41040
    },
    {
      "epoch": 0.9341647133786951,
      "grad_norm": 89.5,
      "learning_rate": 1.3440460709641465e-06,
      "loss": 0.465,
      "step": 41050
    },
    {
      "epoch": 0.9343922809093599,
      "grad_norm": 109.5,
      "learning_rate": 1.339401820546164e-06,
      "loss": 0.4145,
      "step": 41060
    },
    {
      "epoch": 0.9346198484400245,
      "grad_norm": 137.0,
      "learning_rate": 1.3347575701281814e-06,
      "loss": 0.9795,
      "step": 41070
    },
    {
      "epoch": 0.9348474159706893,
      "grad_norm": 0.00244140625,
      "learning_rate": 1.330113319710199e-06,
      "loss": 0.6129,
      "step": 41080
    },
    {
      "epoch": 0.935074983501354,
      "grad_norm": 0.01092529296875,
      "learning_rate": 1.3254690692922164e-06,
      "loss": 0.6366,
      "step": 41090
    },
    {
      "epoch": 0.9353025510320188,
      "grad_norm": 0.01300048828125,
      "learning_rate": 1.3208248188742337e-06,
      "loss": 0.9284,
      "step": 41100
    },
    {
      "epoch": 0.9355301185626834,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3161805684562511e-06,
      "loss": 0.4144,
      "step": 41110
    },
    {
      "epoch": 0.9357576860933482,
      "grad_norm": 0.005462646484375,
      "learning_rate": 1.3115363180382686e-06,
      "loss": 0.3014,
      "step": 41120
    },
    {
      "epoch": 0.935985253624013,
      "grad_norm": 0.0057373046875,
      "learning_rate": 1.306892067620286e-06,
      "loss": 1.7698,
      "step": 41130
    },
    {
      "epoch": 0.9362128211546776,
      "grad_norm": 205.0,
      "learning_rate": 1.3022478172023038e-06,
      "loss": 0.4364,
      "step": 41140
    },
    {
      "epoch": 0.9364403886853424,
      "grad_norm": 0.003692626953125,
      "learning_rate": 1.2976035667843213e-06,
      "loss": 0.7499,
      "step": 41150
    },
    {
      "epoch": 0.9366679562160071,
      "grad_norm": 0.000385284423828125,
      "learning_rate": 1.2929593163663387e-06,
      "loss": 0.3143,
      "step": 41160
    },
    {
      "epoch": 0.9368955237466718,
      "grad_norm": 162.0,
      "learning_rate": 1.288315065948356e-06,
      "loss": 0.6341,
      "step": 41170
    },
    {
      "epoch": 0.9371230912773365,
      "grad_norm": 16.5,
      "learning_rate": 1.2836708155303735e-06,
      "loss": 0.9434,
      "step": 41180
    },
    {
      "epoch": 0.9373506588080013,
      "grad_norm": 58.5,
      "learning_rate": 1.279026565112391e-06,
      "loss": 1.0873,
      "step": 41190
    },
    {
      "epoch": 0.937578226338666,
      "grad_norm": 163.0,
      "learning_rate": 1.2743823146944084e-06,
      "loss": 1.5616,
      "step": 41200
    },
    {
      "epoch": 0.9378057938693307,
      "grad_norm": 104.5,
      "learning_rate": 1.269738064276426e-06,
      "loss": 0.3836,
      "step": 41210
    },
    {
      "epoch": 0.9380333613999955,
      "grad_norm": 2.53125,
      "learning_rate": 1.2650938138584434e-06,
      "loss": 0.9532,
      "step": 41220
    },
    {
      "epoch": 0.9382609289306602,
      "grad_norm": 167.0,
      "learning_rate": 1.2604495634404609e-06,
      "loss": 0.8201,
      "step": 41230
    },
    {
      "epoch": 0.9384884964613249,
      "grad_norm": 85.5,
      "learning_rate": 1.2558053130224781e-06,
      "loss": 1.0521,
      "step": 41240
    },
    {
      "epoch": 0.9387160639919896,
      "grad_norm": 38.75,
      "learning_rate": 1.2511610626044956e-06,
      "loss": 0.8827,
      "step": 41250
    },
    {
      "epoch": 0.9389436315226544,
      "grad_norm": 4.59375,
      "learning_rate": 1.246516812186513e-06,
      "loss": 0.6915,
      "step": 41260
    },
    {
      "epoch": 0.939171199053319,
      "grad_norm": 101.5,
      "learning_rate": 1.2418725617685306e-06,
      "loss": 0.6162,
      "step": 41270
    },
    {
      "epoch": 0.9393987665839838,
      "grad_norm": 174.0,
      "learning_rate": 1.237228311350548e-06,
      "loss": 0.4506,
      "step": 41280
    },
    {
      "epoch": 0.9396263341146486,
      "grad_norm": 170.0,
      "learning_rate": 1.2325840609325655e-06,
      "loss": 0.4677,
      "step": 41290
    },
    {
      "epoch": 0.9398539016453132,
      "grad_norm": 91.0,
      "learning_rate": 1.227939810514583e-06,
      "loss": 0.8939,
      "step": 41300
    },
    {
      "epoch": 0.940081469175978,
      "grad_norm": 206.0,
      "learning_rate": 1.2232955600966005e-06,
      "loss": 0.9096,
      "step": 41310
    },
    {
      "epoch": 0.9403090367066427,
      "grad_norm": 1.28125,
      "learning_rate": 1.218651309678618e-06,
      "loss": 1.1665,
      "step": 41320
    },
    {
      "epoch": 0.9405366042373074,
      "grad_norm": 584.0,
      "learning_rate": 1.2140070592606355e-06,
      "loss": 0.6406,
      "step": 41330
    },
    {
      "epoch": 0.9407641717679721,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.209362808842653e-06,
      "loss": 0.4356,
      "step": 41340
    },
    {
      "epoch": 0.9409917392986369,
      "grad_norm": 310.0,
      "learning_rate": 1.2047185584246704e-06,
      "loss": 0.7865,
      "step": 41350
    },
    {
      "epoch": 0.9412193068293015,
      "grad_norm": 70.0,
      "learning_rate": 1.2000743080066877e-06,
      "loss": 1.5738,
      "step": 41360
    },
    {
      "epoch": 0.9414468743599663,
      "grad_norm": 144.0,
      "learning_rate": 1.1954300575887052e-06,
      "loss": 1.0219,
      "step": 41370
    },
    {
      "epoch": 0.9416744418906311,
      "grad_norm": 14.1875,
      "learning_rate": 1.1907858071707228e-06,
      "loss": 0.6794,
      "step": 41380
    },
    {
      "epoch": 0.9419020094212958,
      "grad_norm": 92.0,
      "learning_rate": 1.1861415567527403e-06,
      "loss": 0.7143,
      "step": 41390
    },
    {
      "epoch": 0.9421295769519605,
      "grad_norm": 182.0,
      "learning_rate": 1.1814973063347576e-06,
      "loss": 0.7581,
      "step": 41400
    },
    {
      "epoch": 0.9423571444826252,
      "grad_norm": 0.00323486328125,
      "learning_rate": 1.176853055916775e-06,
      "loss": 0.4724,
      "step": 41410
    },
    {
      "epoch": 0.94258471201329,
      "grad_norm": 0.322265625,
      "learning_rate": 1.1722088054987925e-06,
      "loss": 0.2051,
      "step": 41420
    },
    {
      "epoch": 0.9428122795439546,
      "grad_norm": 0.000553131103515625,
      "learning_rate": 1.16756455508081e-06,
      "loss": 1.1938,
      "step": 41430
    },
    {
      "epoch": 0.9430398470746194,
      "grad_norm": 0.1826171875,
      "learning_rate": 1.1629203046628275e-06,
      "loss": 0.4131,
      "step": 41440
    },
    {
      "epoch": 0.9432674146052841,
      "grad_norm": 1.2578125,
      "learning_rate": 1.158276054244845e-06,
      "loss": 0.6401,
      "step": 41450
    },
    {
      "epoch": 0.9434949821359488,
      "grad_norm": 440.0,
      "learning_rate": 1.1536318038268625e-06,
      "loss": 0.5336,
      "step": 41460
    },
    {
      "epoch": 0.9437225496666136,
      "grad_norm": 0.99609375,
      "learning_rate": 1.14898755340888e-06,
      "loss": 1.025,
      "step": 41470
    },
    {
      "epoch": 0.9439501171972783,
      "grad_norm": 135.0,
      "learning_rate": 1.1443433029908974e-06,
      "loss": 0.9362,
      "step": 41480
    },
    {
      "epoch": 0.9441776847279431,
      "grad_norm": 0.0004825592041015625,
      "learning_rate": 1.139699052572915e-06,
      "loss": 0.351,
      "step": 41490
    },
    {
      "epoch": 0.9444052522586077,
      "grad_norm": 55.75,
      "learning_rate": 1.1350548021549324e-06,
      "loss": 0.2591,
      "step": 41500
    },
    {
      "epoch": 0.9446328197892725,
      "grad_norm": 153.0,
      "learning_rate": 1.1304105517369496e-06,
      "loss": 0.8268,
      "step": 41510
    },
    {
      "epoch": 0.9448603873199372,
      "grad_norm": 732.0,
      "learning_rate": 1.1257663013189671e-06,
      "loss": 1.0877,
      "step": 41520
    },
    {
      "epoch": 0.9450879548506019,
      "grad_norm": 155.0,
      "learning_rate": 1.1211220509009848e-06,
      "loss": 1.7367,
      "step": 41530
    },
    {
      "epoch": 0.9453155223812666,
      "grad_norm": 92.5,
      "learning_rate": 1.116477800483002e-06,
      "loss": 0.9412,
      "step": 41540
    },
    {
      "epoch": 0.9455430899119314,
      "grad_norm": 33.0,
      "learning_rate": 1.1118335500650196e-06,
      "loss": 0.4394,
      "step": 41550
    },
    {
      "epoch": 0.9457706574425961,
      "grad_norm": 0.021484375,
      "learning_rate": 1.107189299647037e-06,
      "loss": 0.8693,
      "step": 41560
    },
    {
      "epoch": 0.9459982249732608,
      "grad_norm": 1464.0,
      "learning_rate": 1.1025450492290545e-06,
      "loss": 0.7364,
      "step": 41570
    },
    {
      "epoch": 0.9462257925039256,
      "grad_norm": 59.75,
      "learning_rate": 1.097900798811072e-06,
      "loss": 0.7507,
      "step": 41580
    },
    {
      "epoch": 0.9464533600345902,
      "grad_norm": 0.057861328125,
      "learning_rate": 1.0932565483930895e-06,
      "loss": 1.0329,
      "step": 41590
    },
    {
      "epoch": 0.946680927565255,
      "grad_norm": 214.0,
      "learning_rate": 1.088612297975107e-06,
      "loss": 0.2563,
      "step": 41600
    },
    {
      "epoch": 0.9469084950959197,
      "grad_norm": 223.0,
      "learning_rate": 1.0839680475571242e-06,
      "loss": 1.4936,
      "step": 41610
    },
    {
      "epoch": 0.9471360626265845,
      "grad_norm": 41.25,
      "learning_rate": 1.079323797139142e-06,
      "loss": 0.3199,
      "step": 41620
    },
    {
      "epoch": 0.9473636301572491,
      "grad_norm": 30.875,
      "learning_rate": 1.0746795467211594e-06,
      "loss": 0.7364,
      "step": 41630
    },
    {
      "epoch": 0.9475911976879139,
      "grad_norm": 0.055908203125,
      "learning_rate": 1.0700352963031769e-06,
      "loss": 0.1981,
      "step": 41640
    },
    {
      "epoch": 0.9478187652185787,
      "grad_norm": 186.0,
      "learning_rate": 1.0653910458851941e-06,
      "loss": 0.2064,
      "step": 41650
    },
    {
      "epoch": 0.9480463327492433,
      "grad_norm": 488.0,
      "learning_rate": 1.0607467954672116e-06,
      "loss": 1.3921,
      "step": 41660
    },
    {
      "epoch": 0.9482739002799081,
      "grad_norm": 198.0,
      "learning_rate": 1.056102545049229e-06,
      "loss": 1.7618,
      "step": 41670
    },
    {
      "epoch": 0.9485014678105728,
      "grad_norm": 79.5,
      "learning_rate": 1.0514582946312466e-06,
      "loss": 0.6623,
      "step": 41680
    },
    {
      "epoch": 0.9487290353412375,
      "grad_norm": 10.3125,
      "learning_rate": 1.046814044213264e-06,
      "loss": 0.453,
      "step": 41690
    },
    {
      "epoch": 0.9489566028719022,
      "grad_norm": 9.5,
      "learning_rate": 1.0421697937952815e-06,
      "loss": 0.752,
      "step": 41700
    },
    {
      "epoch": 0.949184170402567,
      "grad_norm": 0.150390625,
      "learning_rate": 1.037525543377299e-06,
      "loss": 0.7399,
      "step": 41710
    },
    {
      "epoch": 0.9494117379332316,
      "grad_norm": 106.0,
      "learning_rate": 1.0328812929593165e-06,
      "loss": 0.9454,
      "step": 41720
    },
    {
      "epoch": 0.9496393054638964,
      "grad_norm": 0.2890625,
      "learning_rate": 1.028237042541334e-06,
      "loss": 0.6121,
      "step": 41730
    },
    {
      "epoch": 0.9498668729945612,
      "grad_norm": 71.0,
      "learning_rate": 1.0235927921233514e-06,
      "loss": 1.1535,
      "step": 41740
    },
    {
      "epoch": 0.9500944405252258,
      "grad_norm": 6.375,
      "learning_rate": 1.0189485417053687e-06,
      "loss": 1.9655,
      "step": 41750
    },
    {
      "epoch": 0.9503220080558906,
      "grad_norm": 1.6796875,
      "learning_rate": 1.0143042912873862e-06,
      "loss": 1.015,
      "step": 41760
    },
    {
      "epoch": 0.9505495755865553,
      "grad_norm": 13.5625,
      "learning_rate": 1.0096600408694039e-06,
      "loss": 0.7088,
      "step": 41770
    },
    {
      "epoch": 0.9507771431172201,
      "grad_norm": 75.5,
      "learning_rate": 1.0050157904514213e-06,
      "loss": 0.3242,
      "step": 41780
    },
    {
      "epoch": 0.9510047106478847,
      "grad_norm": 154.0,
      "learning_rate": 1.0003715400334386e-06,
      "loss": 0.4388,
      "step": 41790
    },
    {
      "epoch": 0.9512322781785495,
      "grad_norm": 101.5,
      "learning_rate": 9.95727289615456e-07,
      "loss": 0.3259,
      "step": 41800
    },
    {
      "epoch": 0.9514598457092142,
      "grad_norm": 33.75,
      "learning_rate": 9.910830391974736e-07,
      "loss": 0.5351,
      "step": 41810
    },
    {
      "epoch": 0.9516874132398789,
      "grad_norm": 0.12255859375,
      "learning_rate": 9.86438788779491e-07,
      "loss": 1.1308,
      "step": 41820
    },
    {
      "epoch": 0.9519149807705437,
      "grad_norm": 39.25,
      "learning_rate": 9.817945383615085e-07,
      "loss": 0.2565,
      "step": 41830
    },
    {
      "epoch": 0.9521425483012084,
      "grad_norm": 0.0308837890625,
      "learning_rate": 9.77150287943526e-07,
      "loss": 0.3183,
      "step": 41840
    },
    {
      "epoch": 0.9523701158318731,
      "grad_norm": 0.50390625,
      "learning_rate": 9.725060375255435e-07,
      "loss": 0.4233,
      "step": 41850
    },
    {
      "epoch": 0.9525976833625378,
      "grad_norm": 0.06689453125,
      "learning_rate": 9.67861787107561e-07,
      "loss": 0.3665,
      "step": 41860
    },
    {
      "epoch": 0.9528252508932026,
      "grad_norm": 171.0,
      "learning_rate": 9.632175366895784e-07,
      "loss": 0.5755,
      "step": 41870
    },
    {
      "epoch": 0.9530528184238672,
      "grad_norm": 212.0,
      "learning_rate": 9.58573286271596e-07,
      "loss": 1.2677,
      "step": 41880
    },
    {
      "epoch": 0.953280385954532,
      "grad_norm": 157.0,
      "learning_rate": 9.539290358536132e-07,
      "loss": 0.8514,
      "step": 41890
    },
    {
      "epoch": 0.9535079534851967,
      "grad_norm": 34.25,
      "learning_rate": 9.492847854356308e-07,
      "loss": 0.9848,
      "step": 41900
    },
    {
      "epoch": 0.9537355210158615,
      "grad_norm": 3.046875,
      "learning_rate": 9.446405350176483e-07,
      "loss": 0.2756,
      "step": 41910
    },
    {
      "epoch": 0.9539630885465262,
      "grad_norm": 430.0,
      "learning_rate": 9.399962845996657e-07,
      "loss": 2.0585,
      "step": 41920
    },
    {
      "epoch": 0.9541906560771909,
      "grad_norm": 7.15625,
      "learning_rate": 9.353520341816831e-07,
      "loss": 0.784,
      "step": 41930
    },
    {
      "epoch": 0.9544182236078557,
      "grad_norm": 2.96875,
      "learning_rate": 9.307077837637006e-07,
      "loss": 0.7663,
      "step": 41940
    },
    {
      "epoch": 0.9546457911385203,
      "grad_norm": 223.0,
      "learning_rate": 9.260635333457181e-07,
      "loss": 1.0669,
      "step": 41950
    },
    {
      "epoch": 0.9548733586691851,
      "grad_norm": 0.00150299072265625,
      "learning_rate": 9.214192829277354e-07,
      "loss": 1.0868,
      "step": 41960
    },
    {
      "epoch": 0.9551009261998498,
      "grad_norm": 200.0,
      "learning_rate": 9.16775032509753e-07,
      "loss": 1.2098,
      "step": 41970
    },
    {
      "epoch": 0.9553284937305145,
      "grad_norm": 79.0,
      "learning_rate": 9.121307820917705e-07,
      "loss": 1.1021,
      "step": 41980
    },
    {
      "epoch": 0.9555560612611792,
      "grad_norm": 210.0,
      "learning_rate": 9.07486531673788e-07,
      "loss": 0.9072,
      "step": 41990
    },
    {
      "epoch": 0.955783628791844,
      "grad_norm": 254.0,
      "learning_rate": 9.028422812558053e-07,
      "loss": 0.6898,
      "step": 42000
    },
    {
      "epoch": 0.9560111963225087,
      "grad_norm": 203.0,
      "learning_rate": 8.981980308378228e-07,
      "loss": 0.6469,
      "step": 42010
    },
    {
      "epoch": 0.9562387638531734,
      "grad_norm": 0.013671875,
      "learning_rate": 8.935537804198403e-07,
      "loss": 0.2524,
      "step": 42020
    },
    {
      "epoch": 0.9564663313838382,
      "grad_norm": 1.5078125,
      "learning_rate": 8.889095300018579e-07,
      "loss": 1.061,
      "step": 42030
    },
    {
      "epoch": 0.9566938989145028,
      "grad_norm": 247.0,
      "learning_rate": 8.842652795838752e-07,
      "loss": 1.2864,
      "step": 42040
    },
    {
      "epoch": 0.9569214664451676,
      "grad_norm": 73.5,
      "learning_rate": 8.796210291658927e-07,
      "loss": 1.1352,
      "step": 42050
    },
    {
      "epoch": 0.9571490339758323,
      "grad_norm": 64.5,
      "learning_rate": 8.749767787479102e-07,
      "loss": 0.5545,
      "step": 42060
    },
    {
      "epoch": 0.9573766015064971,
      "grad_norm": 6.28125,
      "learning_rate": 8.703325283299276e-07,
      "loss": 1.3192,
      "step": 42070
    },
    {
      "epoch": 0.9576041690371617,
      "grad_norm": 202.0,
      "learning_rate": 8.656882779119451e-07,
      "loss": 0.5621,
      "step": 42080
    },
    {
      "epoch": 0.9578317365678265,
      "grad_norm": 3.078125,
      "learning_rate": 8.610440274939625e-07,
      "loss": 0.531,
      "step": 42090
    },
    {
      "epoch": 0.9580593040984913,
      "grad_norm": 0.00830078125,
      "learning_rate": 8.5639977707598e-07,
      "loss": 0.681,
      "step": 42100
    },
    {
      "epoch": 0.9582868716291559,
      "grad_norm": 66.5,
      "learning_rate": 8.517555266579974e-07,
      "loss": 0.3645,
      "step": 42110
    },
    {
      "epoch": 0.9585144391598207,
      "grad_norm": 0.007720947265625,
      "learning_rate": 8.47111276240015e-07,
      "loss": 0.0488,
      "step": 42120
    },
    {
      "epoch": 0.9587420066904854,
      "grad_norm": 0.056640625,
      "learning_rate": 8.424670258220325e-07,
      "loss": 0.5117,
      "step": 42130
    },
    {
      "epoch": 0.9589695742211501,
      "grad_norm": 5.40625,
      "learning_rate": 8.378227754040498e-07,
      "loss": 0.6899,
      "step": 42140
    },
    {
      "epoch": 0.9591971417518148,
      "grad_norm": 0.0037689208984375,
      "learning_rate": 8.331785249860673e-07,
      "loss": 0.6219,
      "step": 42150
    },
    {
      "epoch": 0.9594247092824796,
      "grad_norm": 0.1796875,
      "learning_rate": 8.285342745680848e-07,
      "loss": 0.8517,
      "step": 42160
    },
    {
      "epoch": 0.9596522768131444,
      "grad_norm": 169.0,
      "learning_rate": 8.238900241501023e-07,
      "loss": 0.6809,
      "step": 42170
    },
    {
      "epoch": 0.959879844343809,
      "grad_norm": 53.75,
      "learning_rate": 8.192457737321196e-07,
      "loss": 0.7981,
      "step": 42180
    },
    {
      "epoch": 0.9601074118744738,
      "grad_norm": 79.5,
      "learning_rate": 8.146015233141371e-07,
      "loss": 0.4688,
      "step": 42190
    },
    {
      "epoch": 0.9603349794051385,
      "grad_norm": 0.006072998046875,
      "learning_rate": 8.099572728961547e-07,
      "loss": 0.1908,
      "step": 42200
    },
    {
      "epoch": 0.9605625469358032,
      "grad_norm": 145.0,
      "learning_rate": 8.053130224781721e-07,
      "loss": 0.9836,
      "step": 42210
    },
    {
      "epoch": 0.9607901144664679,
      "grad_norm": 0.005584716796875,
      "learning_rate": 8.006687720601896e-07,
      "loss": 0.6546,
      "step": 42220
    },
    {
      "epoch": 0.9610176819971327,
      "grad_norm": 0.006683349609375,
      "learning_rate": 7.96024521642207e-07,
      "loss": 0.7382,
      "step": 42230
    },
    {
      "epoch": 0.9612452495277973,
      "grad_norm": 0.003143310546875,
      "learning_rate": 7.913802712242245e-07,
      "loss": 0.7653,
      "step": 42240
    },
    {
      "epoch": 0.9614728170584621,
      "grad_norm": 188.0,
      "learning_rate": 7.867360208062419e-07,
      "loss": 1.1141,
      "step": 42250
    },
    {
      "epoch": 0.9617003845891269,
      "grad_norm": 0.03759765625,
      "learning_rate": 7.820917703882594e-07,
      "loss": 0.3988,
      "step": 42260
    },
    {
      "epoch": 0.9619279521197915,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.774475199702769e-07,
      "loss": 0.2041,
      "step": 42270
    },
    {
      "epoch": 0.9621555196504563,
      "grad_norm": 0.0019378662109375,
      "learning_rate": 7.728032695522942e-07,
      "loss": 0.3761,
      "step": 42280
    },
    {
      "epoch": 0.962383087181121,
      "grad_norm": 20.5,
      "learning_rate": 7.681590191343118e-07,
      "loss": 0.3932,
      "step": 42290
    },
    {
      "epoch": 0.9626106547117858,
      "grad_norm": 336.0,
      "learning_rate": 7.635147687163293e-07,
      "loss": 0.9872,
      "step": 42300
    },
    {
      "epoch": 0.9628382222424504,
      "grad_norm": 159.0,
      "learning_rate": 7.588705182983468e-07,
      "loss": 0.4259,
      "step": 42310
    },
    {
      "epoch": 0.9630657897731152,
      "grad_norm": 99.0,
      "learning_rate": 7.542262678803641e-07,
      "loss": 1.3817,
      "step": 42320
    },
    {
      "epoch": 0.9632933573037799,
      "grad_norm": 1.078125,
      "learning_rate": 7.495820174623816e-07,
      "loss": 0.745,
      "step": 42330
    },
    {
      "epoch": 0.9635209248344446,
      "grad_norm": 0.01336669921875,
      "learning_rate": 7.449377670443991e-07,
      "loss": 2.0588,
      "step": 42340
    },
    {
      "epoch": 0.9637484923651094,
      "grad_norm": 0.0135498046875,
      "learning_rate": 7.402935166264165e-07,
      "loss": 0.9216,
      "step": 42350
    },
    {
      "epoch": 0.9639760598957741,
      "grad_norm": 185.0,
      "learning_rate": 7.35649266208434e-07,
      "loss": 0.5483,
      "step": 42360
    },
    {
      "epoch": 0.9642036274264388,
      "grad_norm": 22.625,
      "learning_rate": 7.310050157904515e-07,
      "loss": 1.2616,
      "step": 42370
    },
    {
      "epoch": 0.9644311949571035,
      "grad_norm": 0.0015716552734375,
      "learning_rate": 7.26360765372469e-07,
      "loss": 1.3654,
      "step": 42380
    },
    {
      "epoch": 0.9646587624877683,
      "grad_norm": 156.0,
      "learning_rate": 7.217165149544864e-07,
      "loss": 0.5145,
      "step": 42390
    },
    {
      "epoch": 0.9648863300184329,
      "grad_norm": 262.0,
      "learning_rate": 7.170722645365038e-07,
      "loss": 0.9576,
      "step": 42400
    },
    {
      "epoch": 0.9651138975490977,
      "grad_norm": 0.003997802734375,
      "learning_rate": 7.124280141185213e-07,
      "loss": 1.2964,
      "step": 42410
    },
    {
      "epoch": 0.9653414650797624,
      "grad_norm": 128.0,
      "learning_rate": 7.077837637005387e-07,
      "loss": 0.4682,
      "step": 42420
    },
    {
      "epoch": 0.9655690326104271,
      "grad_norm": 540.0,
      "learning_rate": 7.031395132825562e-07,
      "loss": 1.5978,
      "step": 42430
    },
    {
      "epoch": 0.9657966001410919,
      "grad_norm": 83.0,
      "learning_rate": 6.984952628645738e-07,
      "loss": 1.1033,
      "step": 42440
    },
    {
      "epoch": 0.9660241676717566,
      "grad_norm": 43.0,
      "learning_rate": 6.938510124465912e-07,
      "loss": 1.1169,
      "step": 42450
    },
    {
      "epoch": 0.9662517352024214,
      "grad_norm": 0.0133056640625,
      "learning_rate": 6.892067620286086e-07,
      "loss": 0.4807,
      "step": 42460
    },
    {
      "epoch": 0.966479302733086,
      "grad_norm": 206.0,
      "learning_rate": 6.845625116106261e-07,
      "loss": 1.3346,
      "step": 42470
    },
    {
      "epoch": 0.9667068702637508,
      "grad_norm": 0.0008544921875,
      "learning_rate": 6.799182611926436e-07,
      "loss": 1.2885,
      "step": 42480
    },
    {
      "epoch": 0.9669344377944155,
      "grad_norm": 140.0,
      "learning_rate": 6.75274010774661e-07,
      "loss": 0.9859,
      "step": 42490
    },
    {
      "epoch": 0.9671620053250802,
      "grad_norm": 40.0,
      "learning_rate": 6.706297603566784e-07,
      "loss": 0.4546,
      "step": 42500
    },
    {
      "epoch": 0.9673895728557449,
      "grad_norm": 84.5,
      "learning_rate": 6.65985509938696e-07,
      "loss": 1.7257,
      "step": 42510
    },
    {
      "epoch": 0.9676171403864097,
      "grad_norm": 82.5,
      "learning_rate": 6.613412595207135e-07,
      "loss": 0.6211,
      "step": 42520
    },
    {
      "epoch": 0.9678447079170744,
      "grad_norm": 229.0,
      "learning_rate": 6.566970091027309e-07,
      "loss": 0.5329,
      "step": 42530
    },
    {
      "epoch": 0.9680722754477391,
      "grad_norm": 0.0010528564453125,
      "learning_rate": 6.520527586847483e-07,
      "loss": 1.2191,
      "step": 42540
    },
    {
      "epoch": 0.9682998429784039,
      "grad_norm": 0.0021209716796875,
      "learning_rate": 6.474085082667658e-07,
      "loss": 0.398,
      "step": 42550
    },
    {
      "epoch": 0.9685274105090685,
      "grad_norm": 60.25,
      "learning_rate": 6.427642578487833e-07,
      "loss": 0.7799,
      "step": 42560
    },
    {
      "epoch": 0.9687549780397333,
      "grad_norm": 656.0,
      "learning_rate": 6.381200074308007e-07,
      "loss": 1.6435,
      "step": 42570
    },
    {
      "epoch": 0.968982545570398,
      "grad_norm": 0.00347900390625,
      "learning_rate": 6.334757570128181e-07,
      "loss": 0.5347,
      "step": 42580
    },
    {
      "epoch": 0.9692101131010628,
      "grad_norm": 12.375,
      "learning_rate": 6.288315065948357e-07,
      "loss": 0.4244,
      "step": 42590
    },
    {
      "epoch": 0.9694376806317274,
      "grad_norm": 0.0205078125,
      "learning_rate": 6.241872561768531e-07,
      "loss": 2.2514,
      "step": 42600
    },
    {
      "epoch": 0.9696652481623922,
      "grad_norm": 0.000682830810546875,
      "learning_rate": 6.195430057588706e-07,
      "loss": 0.6696,
      "step": 42610
    },
    {
      "epoch": 0.969892815693057,
      "grad_norm": 19.375,
      "learning_rate": 6.148987553408881e-07,
      "loss": 0.8169,
      "step": 42620
    },
    {
      "epoch": 0.9701203832237216,
      "grad_norm": 177.0,
      "learning_rate": 6.102545049229055e-07,
      "loss": 0.8437,
      "step": 42630
    },
    {
      "epoch": 0.9703479507543864,
      "grad_norm": 82.0,
      "learning_rate": 6.05610254504923e-07,
      "loss": 0.9561,
      "step": 42640
    },
    {
      "epoch": 0.9705755182850511,
      "grad_norm": 0.017333984375,
      "learning_rate": 6.009660040869404e-07,
      "loss": 0.9154,
      "step": 42650
    },
    {
      "epoch": 0.9708030858157158,
      "grad_norm": 356.0,
      "learning_rate": 5.963217536689579e-07,
      "loss": 1.2073,
      "step": 42660
    },
    {
      "epoch": 0.9710306533463805,
      "grad_norm": 53.5,
      "learning_rate": 5.916775032509753e-07,
      "loss": 0.8474,
      "step": 42670
    },
    {
      "epoch": 0.9712582208770453,
      "grad_norm": 106.5,
      "learning_rate": 5.870332528329928e-07,
      "loss": 0.2919,
      "step": 42680
    },
    {
      "epoch": 0.9714857884077099,
      "grad_norm": 37.25,
      "learning_rate": 5.823890024150103e-07,
      "loss": 0.9097,
      "step": 42690
    },
    {
      "epoch": 0.9717133559383747,
      "grad_norm": 99.5,
      "learning_rate": 5.777447519970277e-07,
      "loss": 0.4928,
      "step": 42700
    },
    {
      "epoch": 0.9719409234690395,
      "grad_norm": 360.0,
      "learning_rate": 5.731005015790453e-07,
      "loss": 1.2384,
      "step": 42710
    },
    {
      "epoch": 0.9721684909997041,
      "grad_norm": 153.0,
      "learning_rate": 5.684562511610626e-07,
      "loss": 0.5017,
      "step": 42720
    },
    {
      "epoch": 0.9723960585303689,
      "grad_norm": 94.0,
      "learning_rate": 5.638120007430801e-07,
      "loss": 1.1818,
      "step": 42730
    },
    {
      "epoch": 0.9726236260610336,
      "grad_norm": 0.008544921875,
      "learning_rate": 5.591677503250976e-07,
      "loss": 1.1104,
      "step": 42740
    },
    {
      "epoch": 0.9728511935916984,
      "grad_norm": 612.0,
      "learning_rate": 5.545234999071151e-07,
      "loss": 0.7861,
      "step": 42750
    },
    {
      "epoch": 0.973078761122363,
      "grad_norm": 0.02783203125,
      "learning_rate": 5.498792494891325e-07,
      "loss": 0.6705,
      "step": 42760
    },
    {
      "epoch": 0.9733063286530278,
      "grad_norm": 0.004058837890625,
      "learning_rate": 5.452349990711499e-07,
      "loss": 0.1353,
      "step": 42770
    },
    {
      "epoch": 0.9735338961836925,
      "grad_norm": 169.0,
      "learning_rate": 5.405907486531675e-07,
      "loss": 1.8099,
      "step": 42780
    },
    {
      "epoch": 0.9737614637143572,
      "grad_norm": 17.75,
      "learning_rate": 5.359464982351849e-07,
      "loss": 0.6096,
      "step": 42790
    },
    {
      "epoch": 0.973989031245022,
      "grad_norm": 400.0,
      "learning_rate": 5.313022478172024e-07,
      "loss": 1.3651,
      "step": 42800
    },
    {
      "epoch": 0.9742165987756867,
      "grad_norm": 45.75,
      "learning_rate": 5.266579973992198e-07,
      "loss": 0.9068,
      "step": 42810
    },
    {
      "epoch": 0.9744441663063514,
      "grad_norm": 0.0035552978515625,
      "learning_rate": 5.220137469812372e-07,
      "loss": 0.7249,
      "step": 42820
    },
    {
      "epoch": 0.9746717338370161,
      "grad_norm": 5.625,
      "learning_rate": 5.173694965632548e-07,
      "loss": 0.2589,
      "step": 42830
    },
    {
      "epoch": 0.9748993013676809,
      "grad_norm": 7.90625,
      "learning_rate": 5.127252461452722e-07,
      "loss": 1.095,
      "step": 42840
    },
    {
      "epoch": 0.9751268688983455,
      "grad_norm": 0.000583648681640625,
      "learning_rate": 5.080809957272896e-07,
      "loss": 0.6279,
      "step": 42850
    },
    {
      "epoch": 0.9753544364290103,
      "grad_norm": 166.0,
      "learning_rate": 5.034367453093071e-07,
      "loss": 0.6815,
      "step": 42860
    },
    {
      "epoch": 0.975582003959675,
      "grad_norm": 130.0,
      "learning_rate": 4.987924948913246e-07,
      "loss": 0.7295,
      "step": 42870
    },
    {
      "epoch": 0.9758095714903398,
      "grad_norm": 0.93359375,
      "learning_rate": 4.941482444733421e-07,
      "loss": 0.2115,
      "step": 42880
    },
    {
      "epoch": 0.9760371390210045,
      "grad_norm": 95.0,
      "learning_rate": 4.895039940553594e-07,
      "loss": 2.0283,
      "step": 42890
    },
    {
      "epoch": 0.9762647065516692,
      "grad_norm": 243.0,
      "learning_rate": 4.84859743637377e-07,
      "loss": 0.6729,
      "step": 42900
    },
    {
      "epoch": 0.976492274082334,
      "grad_norm": 125.0,
      "learning_rate": 4.802154932193944e-07,
      "loss": 0.4388,
      "step": 42910
    },
    {
      "epoch": 0.9767198416129986,
      "grad_norm": 28.0,
      "learning_rate": 4.7557124280141193e-07,
      "loss": 0.3676,
      "step": 42920
    },
    {
      "epoch": 0.9769474091436634,
      "grad_norm": 382.0,
      "learning_rate": 4.7092699238342936e-07,
      "loss": 0.659,
      "step": 42930
    },
    {
      "epoch": 0.9771749766743281,
      "grad_norm": 324.0,
      "learning_rate": 4.6628274196544684e-07,
      "loss": 1.2889,
      "step": 42940
    },
    {
      "epoch": 0.9774025442049928,
      "grad_norm": 170.0,
      "learning_rate": 4.6163849154746426e-07,
      "loss": 0.552,
      "step": 42950
    },
    {
      "epoch": 0.9776301117356575,
      "grad_norm": 205.0,
      "learning_rate": 4.569942411294817e-07,
      "loss": 1.1317,
      "step": 42960
    },
    {
      "epoch": 0.9778576792663223,
      "grad_norm": 182.0,
      "learning_rate": 4.523499907114992e-07,
      "loss": 0.6163,
      "step": 42970
    },
    {
      "epoch": 0.978085246796987,
      "grad_norm": 262.0,
      "learning_rate": 4.4770574029351664e-07,
      "loss": 0.9328,
      "step": 42980
    },
    {
      "epoch": 0.9783128143276517,
      "grad_norm": 45.5,
      "learning_rate": 4.430614898755341e-07,
      "loss": 1.3236,
      "step": 42990
    },
    {
      "epoch": 0.9785403818583165,
      "grad_norm": 276.0,
      "learning_rate": 4.3841723945755155e-07,
      "loss": 1.0827,
      "step": 43000
    },
    {
      "epoch": 0.9787679493889812,
      "grad_norm": 58.25,
      "learning_rate": 4.337729890395691e-07,
      "loss": 1.3059,
      "step": 43010
    },
    {
      "epoch": 0.9789955169196459,
      "grad_norm": 248.0,
      "learning_rate": 4.291287386215865e-07,
      "loss": 1.2348,
      "step": 43020
    },
    {
      "epoch": 0.9792230844503106,
      "grad_norm": 143.0,
      "learning_rate": 4.2448448820360393e-07,
      "loss": 0.2925,
      "step": 43030
    },
    {
      "epoch": 0.9794506519809754,
      "grad_norm": 0.0242919921875,
      "learning_rate": 4.1984023778562146e-07,
      "loss": 0.44,
      "step": 43040
    },
    {
      "epoch": 0.97967821951164,
      "grad_norm": 116.5,
      "learning_rate": 4.151959873676389e-07,
      "loss": 0.5218,
      "step": 43050
    },
    {
      "epoch": 0.9799057870423048,
      "grad_norm": 75.0,
      "learning_rate": 4.1055173694965637e-07,
      "loss": 0.5091,
      "step": 43060
    },
    {
      "epoch": 0.9801333545729696,
      "grad_norm": 233.0,
      "learning_rate": 4.059074865316738e-07,
      "loss": 0.6416,
      "step": 43070
    },
    {
      "epoch": 0.9803609221036342,
      "grad_norm": 123.5,
      "learning_rate": 4.012632361136913e-07,
      "loss": 0.7359,
      "step": 43080
    },
    {
      "epoch": 0.980588489634299,
      "grad_norm": 0.01171875,
      "learning_rate": 3.9661898569570875e-07,
      "loss": 0.4958,
      "step": 43090
    },
    {
      "epoch": 0.9808160571649637,
      "grad_norm": 316.0,
      "learning_rate": 3.9197473527772623e-07,
      "loss": 0.5802,
      "step": 43100
    },
    {
      "epoch": 0.9810436246956284,
      "grad_norm": 206.0,
      "learning_rate": 3.8733048485974365e-07,
      "loss": 0.3825,
      "step": 43110
    },
    {
      "epoch": 0.9812711922262931,
      "grad_norm": 0.3203125,
      "learning_rate": 3.826862344417611e-07,
      "loss": 0.4393,
      "step": 43120
    },
    {
      "epoch": 0.9814987597569579,
      "grad_norm": 0.0140380859375,
      "learning_rate": 3.780419840237786e-07,
      "loss": 0.8786,
      "step": 43130
    },
    {
      "epoch": 0.9817263272876225,
      "grad_norm": 272.0,
      "learning_rate": 3.7339773360579604e-07,
      "loss": 0.824,
      "step": 43140
    },
    {
      "epoch": 0.9819538948182873,
      "grad_norm": 7.152557373046875e-05,
      "learning_rate": 3.687534831878135e-07,
      "loss": 1.7853,
      "step": 43150
    },
    {
      "epoch": 0.9821814623489521,
      "grad_norm": 31.0,
      "learning_rate": 3.64109232769831e-07,
      "loss": 1.4823,
      "step": 43160
    },
    {
      "epoch": 0.9824090298796168,
      "grad_norm": 121.0,
      "learning_rate": 3.5946498235184847e-07,
      "loss": 0.1354,
      "step": 43170
    },
    {
      "epoch": 0.9826365974102815,
      "grad_norm": 0.6640625,
      "learning_rate": 3.548207319338659e-07,
      "loss": 0.4764,
      "step": 43180
    },
    {
      "epoch": 0.9828641649409462,
      "grad_norm": 0.004364013671875,
      "learning_rate": 3.501764815158833e-07,
      "loss": 1.0185,
      "step": 43190
    },
    {
      "epoch": 0.983091732471611,
      "grad_norm": 124.0,
      "learning_rate": 3.4553223109790085e-07,
      "loss": 0.5175,
      "step": 43200
    },
    {
      "epoch": 0.9833193000022756,
      "grad_norm": 36.75,
      "learning_rate": 3.408879806799183e-07,
      "loss": 0.275,
      "step": 43210
    },
    {
      "epoch": 0.9835468675329404,
      "grad_norm": 292.0,
      "learning_rate": 3.3624373026193576e-07,
      "loss": 0.7722,
      "step": 43220
    },
    {
      "epoch": 0.9837744350636052,
      "grad_norm": 0.04052734375,
      "learning_rate": 3.315994798439532e-07,
      "loss": 0.7943,
      "step": 43230
    },
    {
      "epoch": 0.9840020025942698,
      "grad_norm": 282.0,
      "learning_rate": 3.269552294259707e-07,
      "loss": 0.6405,
      "step": 43240
    },
    {
      "epoch": 0.9842295701249346,
      "grad_norm": 0.00830078125,
      "learning_rate": 3.2231097900798814e-07,
      "loss": 0.4154,
      "step": 43250
    },
    {
      "epoch": 0.9844571376555993,
      "grad_norm": 402.0,
      "learning_rate": 3.1766672859000556e-07,
      "loss": 0.6708,
      "step": 43260
    },
    {
      "epoch": 0.984684705186264,
      "grad_norm": 252.0,
      "learning_rate": 3.1302247817202304e-07,
      "loss": 0.3828,
      "step": 43270
    },
    {
      "epoch": 0.9849122727169287,
      "grad_norm": 49.75,
      "learning_rate": 3.083782277540405e-07,
      "loss": 0.7748,
      "step": 43280
    },
    {
      "epoch": 0.9851398402475935,
      "grad_norm": 9.875,
      "learning_rate": 3.03733977336058e-07,
      "loss": 0.8022,
      "step": 43290
    },
    {
      "epoch": 0.9853674077782582,
      "grad_norm": 49.25,
      "learning_rate": 2.990897269180754e-07,
      "loss": 0.5513,
      "step": 43300
    },
    {
      "epoch": 0.9855949753089229,
      "grad_norm": 218.0,
      "learning_rate": 2.944454765000929e-07,
      "loss": 1.4295,
      "step": 43310
    },
    {
      "epoch": 0.9858225428395877,
      "grad_norm": 89.0,
      "learning_rate": 2.898012260821104e-07,
      "loss": 0.7273,
      "step": 43320
    },
    {
      "epoch": 0.9860501103702524,
      "grad_norm": 165.0,
      "learning_rate": 2.851569756641278e-07,
      "loss": 0.2631,
      "step": 43330
    },
    {
      "epoch": 0.9862776779009171,
      "grad_norm": 0.0245361328125,
      "learning_rate": 2.805127252461453e-07,
      "loss": 0.8145,
      "step": 43340
    },
    {
      "epoch": 0.9865052454315818,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.7586847482816277e-07,
      "loss": 1.5433,
      "step": 43350
    },
    {
      "epoch": 0.9867328129622466,
      "grad_norm": 94.0,
      "learning_rate": 2.7122422441018024e-07,
      "loss": 0.6862,
      "step": 43360
    },
    {
      "epoch": 0.9869603804929112,
      "grad_norm": 0.39453125,
      "learning_rate": 2.665799739921977e-07,
      "loss": 1.0692,
      "step": 43370
    },
    {
      "epoch": 0.987187948023576,
      "grad_norm": 166.0,
      "learning_rate": 2.6193572357421515e-07,
      "loss": 0.7135,
      "step": 43380
    },
    {
      "epoch": 0.9874155155542407,
      "grad_norm": 0.035888671875,
      "learning_rate": 2.5729147315623257e-07,
      "loss": 1.3545,
      "step": 43390
    },
    {
      "epoch": 0.9876430830849054,
      "grad_norm": 51.75,
      "learning_rate": 2.5264722273825005e-07,
      "loss": 1.6251,
      "step": 43400
    },
    {
      "epoch": 0.9878706506155702,
      "grad_norm": 137.0,
      "learning_rate": 2.4800297232026753e-07,
      "loss": 0.9526,
      "step": 43410
    },
    {
      "epoch": 0.9880982181462349,
      "grad_norm": 35.5,
      "learning_rate": 2.43358721902285e-07,
      "loss": 0.8263,
      "step": 43420
    },
    {
      "epoch": 0.9883257856768997,
      "grad_norm": 96.0,
      "learning_rate": 2.387144714843025e-07,
      "loss": 1.305,
      "step": 43430
    },
    {
      "epoch": 0.9885533532075643,
      "grad_norm": 205.0,
      "learning_rate": 2.3407022106631994e-07,
      "loss": 0.5553,
      "step": 43440
    },
    {
      "epoch": 0.9887809207382291,
      "grad_norm": 106.5,
      "learning_rate": 2.2942597064833736e-07,
      "loss": 0.4371,
      "step": 43450
    },
    {
      "epoch": 0.9890084882688938,
      "grad_norm": 0.306640625,
      "learning_rate": 2.2478172023035482e-07,
      "loss": 1.1876,
      "step": 43460
    },
    {
      "epoch": 0.9892360557995585,
      "grad_norm": 114.5,
      "learning_rate": 2.201374698123723e-07,
      "loss": 0.6701,
      "step": 43470
    },
    {
      "epoch": 0.9894636233302232,
      "grad_norm": 34.25,
      "learning_rate": 2.1549321939438977e-07,
      "loss": 0.6967,
      "step": 43480
    },
    {
      "epoch": 0.989691190860888,
      "grad_norm": 0.008056640625,
      "learning_rate": 2.1084896897640723e-07,
      "loss": 0.9611,
      "step": 43490
    },
    {
      "epoch": 0.9899187583915527,
      "grad_norm": 400.0,
      "learning_rate": 2.062047185584247e-07,
      "loss": 0.7535,
      "step": 43500
    },
    {
      "epoch": 0.9901463259222174,
      "grad_norm": 154.0,
      "learning_rate": 2.0156046814044216e-07,
      "loss": 0.4265,
      "step": 43510
    },
    {
      "epoch": 0.9903738934528822,
      "grad_norm": 14.3125,
      "learning_rate": 1.9691621772245963e-07,
      "loss": 0.5422,
      "step": 43520
    },
    {
      "epoch": 0.9906014609835468,
      "grad_norm": 1.4765625,
      "learning_rate": 1.9227196730447706e-07,
      "loss": 0.5955,
      "step": 43530
    },
    {
      "epoch": 0.9908290285142116,
      "grad_norm": 58.75,
      "learning_rate": 1.8762771688649454e-07,
      "loss": 0.9521,
      "step": 43540
    },
    {
      "epoch": 0.9910565960448763,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.82983466468512e-07,
      "loss": 1.6394,
      "step": 43550
    },
    {
      "epoch": 0.991284163575541,
      "grad_norm": 0.0017242431640625,
      "learning_rate": 1.7833921605052947e-07,
      "loss": 0.5429,
      "step": 43560
    },
    {
      "epoch": 0.9915117311062057,
      "grad_norm": 0.0047607421875,
      "learning_rate": 1.7369496563254692e-07,
      "loss": 1.2209,
      "step": 43570
    },
    {
      "epoch": 0.9917392986368705,
      "grad_norm": 2.09375,
      "learning_rate": 1.690507152145644e-07,
      "loss": 0.4713,
      "step": 43580
    },
    {
      "epoch": 0.9919668661675353,
      "grad_norm": 0.00543212890625,
      "learning_rate": 1.6440646479658185e-07,
      "loss": 1.0367,
      "step": 43590
    },
    {
      "epoch": 0.9921944336981999,
      "grad_norm": 233.0,
      "learning_rate": 1.5976221437859933e-07,
      "loss": 0.2743,
      "step": 43600
    },
    {
      "epoch": 0.9924220012288647,
      "grad_norm": 210.0,
      "learning_rate": 1.5511796396061678e-07,
      "loss": 1.5769,
      "step": 43610
    },
    {
      "epoch": 0.9926495687595294,
      "grad_norm": 100.0,
      "learning_rate": 1.5047371354263423e-07,
      "loss": 1.1105,
      "step": 43620
    },
    {
      "epoch": 0.9928771362901941,
      "grad_norm": 53.0,
      "learning_rate": 1.4582946312465169e-07,
      "loss": 0.5386,
      "step": 43630
    },
    {
      "epoch": 0.9931047038208588,
      "grad_norm": 67.5,
      "learning_rate": 1.4118521270666916e-07,
      "loss": 0.8195,
      "step": 43640
    },
    {
      "epoch": 0.9933322713515236,
      "grad_norm": 0.46484375,
      "learning_rate": 1.3654096228868662e-07,
      "loss": 0.156,
      "step": 43650
    },
    {
      "epoch": 0.9935598388821882,
      "grad_norm": 336.0,
      "learning_rate": 1.3189671187070407e-07,
      "loss": 1.1726,
      "step": 43660
    },
    {
      "epoch": 0.993787406412853,
      "grad_norm": 159.0,
      "learning_rate": 1.2725246145272155e-07,
      "loss": 0.8899,
      "step": 43670
    },
    {
      "epoch": 0.9940149739435178,
      "grad_norm": 157.0,
      "learning_rate": 1.22608211034739e-07,
      "loss": 1.5587,
      "step": 43680
    },
    {
      "epoch": 0.9942425414741825,
      "grad_norm": 512.0,
      "learning_rate": 1.1796396061675648e-07,
      "loss": 1.1514,
      "step": 43690
    },
    {
      "epoch": 0.9944701090048472,
      "grad_norm": 0.0028839111328125,
      "learning_rate": 1.1331971019877392e-07,
      "loss": 1.0759,
      "step": 43700
    },
    {
      "epoch": 0.9946976765355119,
      "grad_norm": 243.0,
      "learning_rate": 1.086754597807914e-07,
      "loss": 1.0491,
      "step": 43710
    },
    {
      "epoch": 0.9949252440661767,
      "grad_norm": 123.0,
      "learning_rate": 1.0403120936280886e-07,
      "loss": 0.841,
      "step": 43720
    },
    {
      "epoch": 0.9951528115968413,
      "grad_norm": 0.0179443359375,
      "learning_rate": 9.938695894482632e-08,
      "loss": 0.4865,
      "step": 43730
    },
    {
      "epoch": 0.9953803791275061,
      "grad_norm": 182.0,
      "learning_rate": 9.474270852684376e-08,
      "loss": 0.4718,
      "step": 43740
    },
    {
      "epoch": 0.9956079466581708,
      "grad_norm": 85.0,
      "learning_rate": 9.009845810886124e-08,
      "loss": 0.2899,
      "step": 43750
    },
    {
      "epoch": 0.9958355141888355,
      "grad_norm": 33.0,
      "learning_rate": 8.54542076908787e-08,
      "loss": 0.1927,
      "step": 43760
    },
    {
      "epoch": 0.9960630817195003,
      "grad_norm": 0.2890625,
      "learning_rate": 8.080995727289617e-08,
      "loss": 0.857,
      "step": 43770
    },
    {
      "epoch": 0.996290649250165,
      "grad_norm": 0.0033721923828125,
      "learning_rate": 7.616570685491362e-08,
      "loss": 0.8845,
      "step": 43780
    },
    {
      "epoch": 0.9965182167808297,
      "grad_norm": 80.0,
      "learning_rate": 7.152145643693109e-08,
      "loss": 0.3362,
      "step": 43790
    },
    {
      "epoch": 0.9967457843114944,
      "grad_norm": 0.0162353515625,
      "learning_rate": 6.687720601894855e-08,
      "loss": 0.3574,
      "step": 43800
    },
    {
      "epoch": 0.9969733518421592,
      "grad_norm": 112.5,
      "learning_rate": 6.2232955600966e-08,
      "loss": 0.4992,
      "step": 43810
    },
    {
      "epoch": 0.9972009193728238,
      "grad_norm": 0.036865234375,
      "learning_rate": 5.7588705182983464e-08,
      "loss": 0.9337,
      "step": 43820
    },
    {
      "epoch": 0.9974284869034886,
      "grad_norm": 268.0,
      "learning_rate": 5.2944454765000936e-08,
      "loss": 0.5788,
      "step": 43830
    },
    {
      "epoch": 0.9976560544341533,
      "grad_norm": 17.875,
      "learning_rate": 4.830020434701839e-08,
      "loss": 0.6825,
      "step": 43840
    },
    {
      "epoch": 0.9978836219648181,
      "grad_norm": 126.5,
      "learning_rate": 4.365595392903586e-08,
      "loss": 1.3572,
      "step": 43850
    },
    {
      "epoch": 0.9981111894954828,
      "grad_norm": 56.5,
      "learning_rate": 3.901170351105332e-08,
      "loss": 0.431,
      "step": 43860
    },
    {
      "epoch": 0.9983387570261475,
      "grad_norm": 64.5,
      "learning_rate": 3.4367453093070784e-08,
      "loss": 0.5223,
      "step": 43870
    },
    {
      "epoch": 0.9985663245568123,
      "grad_norm": 19.125,
      "learning_rate": 2.9723202675088246e-08,
      "loss": 0.4116,
      "step": 43880
    },
    {
      "epoch": 0.9987938920874769,
      "grad_norm": 2.375,
      "learning_rate": 2.5078952257105708e-08,
      "loss": 0.2104,
      "step": 43890
    },
    {
      "epoch": 0.9990214596181417,
      "grad_norm": 0.001373291015625,
      "learning_rate": 2.043470183912317e-08,
      "loss": 0.8277,
      "step": 43900
    },
    {
      "epoch": 0.9992490271488064,
      "grad_norm": 25.375,
      "learning_rate": 1.5790451421140628e-08,
      "loss": 0.6373,
      "step": 43910
    },
    {
      "epoch": 0.9994765946794711,
      "grad_norm": 346.0,
      "learning_rate": 1.1146201003158092e-08,
      "loss": 0.9354,
      "step": 43920
    },
    {
      "epoch": 0.9997041622101358,
      "grad_norm": 199.0,
      "learning_rate": 6.501950585175554e-09,
      "loss": 1.3832,
      "step": 43930
    },
    {
      "epoch": 0.9999317297408006,
      "grad_norm": 0.0037078857421875,
      "learning_rate": 1.8577001671930154e-09,
      "loss": 1.4164,
      "step": 43940
    }
  ],
  "logging_steps": 10,
  "max_steps": 43943,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.407608319990349e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}