{
  "best_global_step": 400,
  "best_metric": 0.9433377981185913,
  "best_model_checkpoint": "/data6/jongbinwon/evolving_MoE/checkpoints/sft_llama31_8b_sni_ours-c_61797_ffn_only_5ep/checkpoint-400",
  "epoch": 5.0,
  "eval_steps": 100,
  "global_step": 2290,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 1.8063647478818894,
      "epoch": 0.02185792349726776,
      "grad_norm": 3.423154830932617,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 1.4558660507202148,
      "mean_token_accuracy": 0.47703990414738656,
      "num_tokens": 35903.0,
      "step": 10
    },
    {
      "entropy": 1.75661009401083,
      "epoch": 0.04371584699453552,
      "grad_norm": 5.560386657714844,
      "learning_rate": 5.507246376811595e-06,
      "loss": 1.4596789360046387,
      "mean_token_accuracy": 0.49832420609891415,
      "num_tokens": 73449.0,
      "step": 20
    },
    {
      "entropy": 1.7712477460503577,
      "epoch": 0.06557377049180328,
      "grad_norm": 10.560691833496094,
      "learning_rate": 8.405797101449275e-06,
      "loss": 1.3317112922668457,
      "mean_token_accuracy": 0.5486307129263878,
      "num_tokens": 110133.0,
      "step": 30
    },
    {
      "entropy": 1.7841636031866073,
      "epoch": 0.08743169398907104,
      "grad_norm": 2.2995595932006836,
      "learning_rate": 1.1304347826086957e-05,
      "loss": 1.2257501602172851,
      "mean_token_accuracy": 0.6805846616625786,
      "num_tokens": 145855.0,
      "step": 40
    },
    {
      "entropy": 1.8792877070605756,
      "epoch": 0.1092896174863388,
      "grad_norm": 5.942349910736084,
      "learning_rate": 1.420289855072464e-05,
      "loss": 1.1201908111572265,
      "mean_token_accuracy": 0.6831882126629353,
      "num_tokens": 184759.0,
      "step": 50
    },
    {
      "entropy": 2.0755775086581707,
      "epoch": 0.13114754098360656,
      "grad_norm": 2.814479112625122,
      "learning_rate": 1.710144927536232e-05,
      "loss": 0.8766800880432128,
      "mean_token_accuracy": 0.72079848498106,
      "num_tokens": 221627.0,
      "step": 60
    },
    {
      "entropy": 2.1253911912441255,
      "epoch": 0.15300546448087432,
      "grad_norm": 3.9029574394226074,
      "learning_rate": 2e-05,
      "loss": 1.0101543426513673,
      "mean_token_accuracy": 0.6912351328879595,
      "num_tokens": 255149.0,
      "step": 70
    },
    {
      "entropy": 2.175362668186426,
      "epoch": 0.17486338797814208,
      "grad_norm": 0.565113365650177,
      "learning_rate": 1.9998999619099485e-05,
      "loss": 0.9229906082153321,
      "mean_token_accuracy": 0.7883817024528981,
      "num_tokens": 301738.0,
      "step": 80
    },
    {
      "entropy": 2.2245839439332484,
      "epoch": 0.19672131147540983,
      "grad_norm": 1.5411109924316406,
      "learning_rate": 1.999599867655032e-05,
      "loss": 0.7247772216796875,
      "mean_token_accuracy": 0.7564309105277062,
      "num_tokens": 343764.0,
      "step": 90
    },
    {
      "entropy": 2.2699677906930447,
      "epoch": 0.2185792349726776,
      "grad_norm": 0.1435852348804474,
      "learning_rate": 1.9990997772769632e-05,
      "loss": 0.7856242656707764,
      "mean_token_accuracy": 0.6836250949651003,
      "num_tokens": 379571.0,
      "step": 100
    },
    {
      "epoch": 0.2185792349726776,
      "eval_entropy": 2.2037881575524807,
      "eval_loss": 0.9982572793960571,
      "eval_mean_token_accuracy": 0.757006504223682,
      "eval_num_tokens": 379571.0,
      "eval_runtime": 25.2898,
      "eval_samples_per_second": 20.245,
      "eval_steps_per_second": 10.123,
      "step": 100
    },
    {
      "entropy": 2.1564709596335887,
      "epoch": 0.24043715846994534,
      "grad_norm": 2.1890692710876465,
      "learning_rate": 1.9983997908319148e-05,
      "loss": 0.7731066703796386,
      "mean_token_accuracy": 0.7627312574535609,
      "num_tokens": 433289.0,
      "step": 110
    },
    {
      "entropy": 2.247432677447796,
      "epoch": 0.26229508196721313,
      "grad_norm": 1.639736294746399,
      "learning_rate": 1.9975000483705003e-05,
      "loss": 0.6999035835266113,
      "mean_token_accuracy": 0.789754918217659,
      "num_tokens": 471536.0,
      "step": 120
    },
    {
      "entropy": 2.159890849888325,
      "epoch": 0.28415300546448086,
      "grad_norm": 6.709820747375488,
      "learning_rate": 1.9964007299097545e-05,
      "loss": 0.7218995094299316,
      "mean_token_accuracy": 0.7591581210494042,
      "num_tokens": 506833.0,
      "step": 130
    },
    {
      "entropy": 2.2273385375738144,
      "epoch": 0.30601092896174864,
      "grad_norm": 4.0001444816589355,
      "learning_rate": 1.9951020553971163e-05,
      "loss": 0.5942524909973145,
      "mean_token_accuracy": 0.7525342235341668,
      "num_tokens": 544522.0,
      "step": 140
    },
    {
      "entropy": 2.1546551048755647,
      "epoch": 0.32786885245901637,
      "grad_norm": 2.4917547702789307,
      "learning_rate": 1.9936042846664212e-05,
      "loss": 0.9668840408325196,
      "mean_token_accuracy": 0.7837827190756798,
      "num_tokens": 585059.0,
      "step": 150
    },
    {
      "entropy": 2.1303168818354608,
      "epoch": 0.34972677595628415,
      "grad_norm": 4.402857780456543,
      "learning_rate": 1.991907717385915e-05,
      "loss": 0.6431563377380372,
      "mean_token_accuracy": 0.7454070471227169,
      "num_tokens": 616372.0,
      "step": 160
    },
    {
      "entropy": 2.197571408748627,
      "epoch": 0.37158469945355194,
      "grad_norm": 2.195585012435913,
      "learning_rate": 1.9900126929982996e-05,
      "loss": 0.8557025909423828,
      "mean_token_accuracy": 0.7525063119828701,
      "num_tokens": 651102.0,
      "step": 170
    },
    {
      "entropy": 2.107708531618118,
      "epoch": 0.39344262295081966,
      "grad_norm": 3.916335105895996,
      "learning_rate": 1.987919590652815e-05,
      "loss": 0.8322604179382325,
      "mean_token_accuracy": 0.7549931760877371,
      "num_tokens": 688543.0,
      "step": 180
    },
    {
      "entropy": 2.118951064348221,
      "epoch": 0.41530054644808745,
      "grad_norm": 2.6407856941223145,
      "learning_rate": 1.985628829129383e-05,
      "loss": 0.7598269939422607,
      "mean_token_accuracy": 0.7544480536133051,
      "num_tokens": 722067.0,
      "step": 190
    },
    {
      "entropy": 2.0949587907642124,
      "epoch": 0.4371584699453552,
      "grad_norm": 3.0593597888946533,
      "learning_rate": 1.983140866754819e-05,
      "loss": 0.627412748336792,
      "mean_token_accuracy": 0.7927730124443769,
      "num_tokens": 760510.0,
      "step": 200
    },
    {
      "epoch": 0.4371584699453552,
      "eval_entropy": 2.1543811270967126,
      "eval_loss": 0.9629567861557007,
      "eval_mean_token_accuracy": 0.7506587785901502,
      "eval_num_tokens": 760510.0,
      "eval_runtime": 25.2887,
      "eval_samples_per_second": 20.246,
      "eval_steps_per_second": 10.123,
      "step": 200
    },
    {
      "entropy": 2.0893077090382577,
      "epoch": 0.45901639344262296,
      "grad_norm": 5.467627048492432,
      "learning_rate": 1.9804562013111307e-05,
      "loss": 0.7501600742340088,
      "mean_token_accuracy": 0.7809621814638377,
      "num_tokens": 800890.0,
      "step": 210
    },
    {
      "entropy": 2.1575353160500526,
      "epoch": 0.4808743169398907,
      "grad_norm": 3.708895683288574,
      "learning_rate": 1.9775753699359254e-05,
      "loss": 0.7001654624938964,
      "mean_token_accuracy": 0.8030995633453131,
      "num_tokens": 841418.0,
      "step": 220
    },
    {
      "entropy": 2.1907149985432626,
      "epoch": 0.5027322404371585,
      "grad_norm": 5.7744269371032715,
      "learning_rate": 1.9744989490149403e-05,
      "loss": 0.9803537368774414,
      "mean_token_accuracy": 0.7649548169225454,
      "num_tokens": 876345.0,
      "step": 230
    },
    {
      "entropy": 2.1935453712940216,
      "epoch": 0.5245901639344263,
      "grad_norm": 1.9867186546325684,
      "learning_rate": 1.971227554066721e-05,
      "loss": 0.9978780746459961,
      "mean_token_accuracy": 0.74650296010077,
      "num_tokens": 911261.0,
      "step": 240
    },
    {
      "entropy": 2.1268230110406874,
      "epoch": 0.546448087431694,
      "grad_norm": 5.917712211608887,
      "learning_rate": 1.967761839619473e-05,
      "loss": 0.8284941673278808,
      "mean_token_accuracy": 0.8314751997590065,
      "num_tokens": 946757.0,
      "step": 250
    },
    {
      "entropy": 2.0225247129797936,
      "epoch": 0.5683060109289617,
      "grad_norm": 4.7960052490234375,
      "learning_rate": 1.964102499080104e-05,
      "loss": 0.6664062023162842,
      "mean_token_accuracy": 0.7690637439489365,
      "num_tokens": 986979.0,
      "step": 260
    },
    {
      "entropy": 2.0265661850571632,
      "epoch": 0.5901639344262295,
      "grad_norm": 5.520472049713135,
      "learning_rate": 1.9602502645954908e-05,
      "loss": 0.6990304470062256,
      "mean_token_accuracy": 0.7301983077079057,
      "num_tokens": 1030774.0,
      "step": 270
    },
    {
      "entropy": 2.1217927753925325,
      "epoch": 0.6120218579234973,
      "grad_norm": 2.748990535736084,
      "learning_rate": 1.9562059069059936e-05,
      "loss": 0.6773673057556152,
      "mean_token_accuracy": 0.7711487539112568,
      "num_tokens": 1074459.0,
      "step": 280
    },
    {
      "entropy": 1.9649823542684317,
      "epoch": 0.6338797814207651,
      "grad_norm": 2.781599283218384,
      "learning_rate": 1.9519702351912504e-05,
      "loss": 0.7069565773010253,
      "mean_token_accuracy": 0.7780966762453317,
      "num_tokens": 1119672.0,
      "step": 290
    },
    {
      "entropy": 2.0320102006196974,
      "epoch": 0.6557377049180327,
      "grad_norm": 1.9452710151672363,
      "learning_rate": 1.9475440969082778e-05,
      "loss": 0.6044183254241944,
      "mean_token_accuracy": 0.8210622433573007,
      "num_tokens": 1154144.0,
      "step": 300
    },
    {
      "epoch": 0.6557377049180327,
      "eval_entropy": 2.0770468888804317,
      "eval_loss": 0.9566176533699036,
      "eval_mean_token_accuracy": 0.7631950126960874,
      "eval_num_tokens": 1154144.0,
      "eval_runtime": 25.2625,
      "eval_samples_per_second": 20.267,
      "eval_steps_per_second": 10.134,
      "step": 300
    },
    {
      "entropy": 2.0678520128130913,
      "epoch": 0.6775956284153005,
      "grad_norm": 1.895296335220337,
      "learning_rate": 1.9429283776219164e-05,
      "loss": 0.7932215213775635,
      "mean_token_accuracy": 0.7864639211446047,
      "num_tokens": 1184770.0,
      "step": 310
    },
    {
      "entropy": 2.160663002729416,
      "epoch": 0.6994535519125683,
      "grad_norm": 2.3100788593292236,
      "learning_rate": 1.938124000827649e-05,
      "loss": 0.9294171333312988,
      "mean_token_accuracy": 0.7919486667960882,
      "num_tokens": 1216799.0,
      "step": 320
    },
    {
      "entropy": 2.1491317108273504,
      "epoch": 0.7213114754098361,
      "grad_norm": 3.916989803314209,
      "learning_rate": 1.9331319277668324e-05,
      "loss": 0.5258510112762451,
      "mean_token_accuracy": 0.7600123845040798,
      "num_tokens": 1255660.0,
      "step": 330
    },
    {
      "entropy": 2.070452833175659,
      "epoch": 0.7431693989071039,
      "grad_norm": 3.346385955810547,
      "learning_rate": 1.927953157234376e-05,
      "loss": 0.698235559463501,
      "mean_token_accuracy": 0.796322438493371,
      "num_tokens": 1292285.0,
      "step": 340
    },
    {
      "entropy": 2.0435832902789115,
      "epoch": 0.7650273224043715,
      "grad_norm": 1.684852123260498,
      "learning_rate": 1.9225887253789048e-05,
      "loss": 0.5733598709106446,
      "mean_token_accuracy": 0.8418302100151778,
      "num_tokens": 1344845.0,
      "step": 350
    },
    {
      "entropy": 2.1818724423646927,
      "epoch": 0.7868852459016393,
      "grad_norm": 3.2415714263916016,
      "learning_rate": 1.9170397054954532e-05,
      "loss": 0.8297238349914551,
      "mean_token_accuracy": 0.7598258547484875,
      "num_tokens": 1376803.0,
      "step": 360
    },
    {
      "entropy": 2.1174234583973885,
      "epoch": 0.8087431693989071,
      "grad_norm": 6.760610580444336,
      "learning_rate": 1.9113072078107227e-05,
      "loss": 0.7774213790893555,
      "mean_token_accuracy": 0.7755776561796666,
      "num_tokens": 1416000.0,
      "step": 370
    },
    {
      "entropy": 2.0920872397720816,
      "epoch": 0.8306010928961749,
      "grad_norm": 1.603777289390564,
      "learning_rate": 1.905392379260953e-05,
      "loss": 0.6159489631652832,
      "mean_token_accuracy": 0.8068475790321827,
      "num_tokens": 1456210.0,
      "step": 380
    },
    {
      "entropy": 2.0747187957167625,
      "epoch": 0.8524590163934426,
      "grad_norm": 11.510518074035645,
      "learning_rate": 1.899296403262446e-05,
      "loss": 0.7138397216796875,
      "mean_token_accuracy": 0.7663105923682452,
      "num_tokens": 1493407.0,
      "step": 390
    },
    {
      "entropy": 1.9438473023474216,
      "epoch": 0.8743169398907104,
      "grad_norm": 3.9798667430877686,
      "learning_rate": 1.893020499474793e-05,
      "loss": 0.5529248237609863,
      "mean_token_accuracy": 0.7778996028006077,
      "num_tokens": 1533619.0,
      "step": 400
    },
    {
      "epoch": 0.8743169398907104,
      "eval_entropy": 2.032593308482319,
      "eval_loss": 0.9433377981185913,
      "eval_mean_token_accuracy": 0.7655589742353186,
      "eval_num_tokens": 1533619.0,
      "eval_runtime": 25.276,
      "eval_samples_per_second": 20.256,
      "eval_steps_per_second": 10.128,
      "step": 400
    },
    {
      "entropy": 1.9707405015826225,
      "epoch": 0.8961748633879781,
      "grad_norm": 1.7734041213989258,
      "learning_rate": 1.8865659235568508e-05,
      "loss": 1.0756511688232422,
      "mean_token_accuracy": 0.7205761078745127,
      "num_tokens": 1575976.0,
      "step": 410
    },
    {
      "entropy": 1.9950819492340088,
      "epoch": 0.9180327868852459,
      "grad_norm": 1.9045618772506714,
      "learning_rate": 1.879933966915514e-05,
      "loss": 0.663206148147583,
      "mean_token_accuracy": 0.7859252873808146,
      "num_tokens": 1621647.0,
      "step": 420
    },
    {
      "entropy": 2.0255909264087677,
      "epoch": 0.9398907103825137,
      "grad_norm": 6.2514214515686035,
      "learning_rate": 1.873125956447333e-05,
      "loss": 0.6743417263031006,
      "mean_token_accuracy": 0.8094865296036005,
      "num_tokens": 1658612.0,
      "step": 430
    },
    {
      "entropy": 2.109029750525951,
      "epoch": 0.9617486338797814,
      "grad_norm": 5.623356342315674,
      "learning_rate": 1.8661432542730365e-05,
      "loss": 0.7329175472259521,
      "mean_token_accuracy": 0.74499936029315,
      "num_tokens": 1691663.0,
      "step": 440
    },
    {
      "entropy": 2.137598206847906,
      "epoch": 0.9836065573770492,
      "grad_norm": 3.081730842590332,
      "learning_rate": 1.858987257465003e-05,
      "loss": 0.6648739814758301,
      "mean_token_accuracy": 0.8063705991953611,
      "num_tokens": 1735524.0,
      "step": 450
    },
    {
      "entropy": 2.1458014014520144,
      "epoch": 1.0043715846994536,
      "grad_norm": 3.006117105484009,
      "learning_rate": 1.8516593977677384e-05,
      "loss": 0.6919145584106445,
      "mean_token_accuracy": 0.8407357433124593,
      "num_tokens": 1768193.0,
      "step": 460
    },
    {
      "entropy": 2.110413059592247,
      "epoch": 1.0262295081967212,
      "grad_norm": 0.3206161856651306,
      "learning_rate": 1.8441611413114187e-05,
      "loss": 0.5173742294311523,
      "mean_token_accuracy": 0.8401550747454166,
      "num_tokens": 1804593.0,
      "step": 470
    },
    {
      "entropy": 2.0955183535814284,
      "epoch": 1.048087431693989,
      "grad_norm": 3.1438255310058594,
      "learning_rate": 1.836493988318554e-05,
      "loss": 0.5835023880004883,
      "mean_token_accuracy": 0.8145145986229181,
      "num_tokens": 1837965.0,
      "step": 480
    },
    {
      "entropy": 2.0786552384495733,
      "epoch": 1.0699453551912568,
      "grad_norm": 5.301725387573242,
      "learning_rate": 1.8286594728038268e-05,
      "loss": 0.5727744102478027,
      "mean_token_accuracy": 0.8251623082906008,
      "num_tokens": 1897279.0,
      "step": 490
    },
    {
      "entropy": 2.0536697883158923,
      "epoch": 1.0918032786885246,
      "grad_norm": 1.8960548639297485,
      "learning_rate": 1.8206591622671746e-05,
      "loss": 0.4378913402557373,
      "mean_token_accuracy": 0.8633764263242483,
      "num_tokens": 1936074.0,
      "step": 500
    },
    {
      "epoch": 1.0918032786885246,
      "eval_entropy": 2.0147718605585396,
      "eval_loss": 1.0390677452087402,
      "eval_mean_token_accuracy": 0.730600222479552,
      "eval_num_tokens": 1936074.0,
      "eval_runtime": 25.2597,
      "eval_samples_per_second": 20.269,
      "eval_steps_per_second": 10.135,
      "step": 500
    },
    {
      "entropy": 2.032060332596302,
      "epoch": 1.1136612021857923,
      "grad_norm": 10.848620414733887,
      "learning_rate": 1.8124946573801686e-05,
      "loss": 0.6254462718963623,
      "mean_token_accuracy": 0.8352152071893215,
      "num_tokens": 1970651.0,
      "step": 510
    },
    {
      "entropy": 1.856139574199915,
      "epoch": 1.1355191256830601,
      "grad_norm": 4.950552463531494,
      "learning_rate": 1.8041675916657597e-05,
      "loss": 0.5790982723236084,
      "mean_token_accuracy": 0.8180347349494695,
      "num_tokens": 2013248.0,
      "step": 520
    },
    {
      "entropy": 1.9065284475684166,
      "epoch": 1.157377049180328,
      "grad_norm": 4.5021891593933105,
      "learning_rate": 1.7956796311714474e-05,
      "loss": 0.564539098739624,
      "mean_token_accuracy": 0.8569660175591707,
      "num_tokens": 2052332.0,
      "step": 530
    },
    {
      "entropy": 1.9384179085493087,
      "epoch": 1.1792349726775957,
      "grad_norm": 13.089709281921387,
      "learning_rate": 1.787032474135944e-05,
      "loss": 0.7635960578918457,
      "mean_token_accuracy": 0.7828190468251706,
      "num_tokens": 2088694.0,
      "step": 540
    },
    {
      "entropy": 1.9600365862250329,
      "epoch": 1.2010928961748635,
      "grad_norm": 2.0104820728302,
      "learning_rate": 1.778227850649398e-05,
      "loss": 0.6327505588531495,
      "mean_token_accuracy": 0.7942470081150532,
      "num_tokens": 2125379.0,
      "step": 550
    },
    {
      "entropy": 1.9059402167797088,
      "epoch": 1.222950819672131,
      "grad_norm": 2.1560399532318115,
      "learning_rate": 1.769267522307244e-05,
      "loss": 0.6111732959747315,
      "mean_token_accuracy": 0.8242439936846495,
      "num_tokens": 2165069.0,
      "step": 560
    },
    {
      "entropy": 1.946674208343029,
      "epoch": 1.2448087431693988,
      "grad_norm": 0.47851449251174927,
      "learning_rate": 1.760153281857749e-05,
      "loss": 0.4375136852264404,
      "mean_token_accuracy": 0.8364114321768283,
      "num_tokens": 2202898.0,
      "step": 570
    },
    {
      "entropy": 2.031615723669529,
      "epoch": 1.2666666666666666,
      "grad_norm": 6.617761135101318,
      "learning_rate": 1.7508869528433264e-05,
      "loss": 0.829713249206543,
      "mean_token_accuracy": 0.7643794182687997,
      "num_tokens": 2240453.0,
      "step": 580
    },
    {
      "entropy": 1.9364936739206313,
      "epoch": 1.2885245901639344,
      "grad_norm": 7.973677158355713,
      "learning_rate": 1.74147038923569e-05,
      "loss": 0.437787389755249,
      "mean_token_accuracy": 0.8334135215729475,
      "num_tokens": 2282118.0,
      "step": 590
    },
    {
      "entropy": 1.994447159767151,
      "epoch": 1.3103825136612022,
      "grad_norm": 3.4388744831085205,
      "learning_rate": 1.731905475064915e-05,
      "loss": 0.5179662704467773,
      "mean_token_accuracy": 0.8467293042689562,
      "num_tokens": 2320520.0,
      "step": 600
    },
    {
      "epoch": 1.3103825136612022,
      "eval_entropy": 1.9500198299065232,
      "eval_loss": 1.0030851364135742,
      "eval_mean_token_accuracy": 0.757350493222475,
      "eval_num_tokens": 2320520.0,
      "eval_runtime": 25.2503,
      "eval_samples_per_second": 20.277,
      "eval_steps_per_second": 10.138,
      "step": 600
    },
    {
      "entropy": 1.907582050561905,
      "epoch": 1.33224043715847,
      "grad_norm": 3.3789825439453125,
      "learning_rate": 1.7221941240424922e-05,
      "loss": 0.4778538703918457,
      "mean_token_accuracy": 0.8434304721653462,
      "num_tokens": 2355096.0,
      "step": 610
    },
    {
      "entropy": 1.8661066718399524,
      "epoch": 1.3540983606557377,
      "grad_norm": 2.7720794677734375,
      "learning_rate": 1.7123382791784376e-05,
      "loss": 0.4396005630493164,
      "mean_token_accuracy": 0.8284852344542741,
      "num_tokens": 2409391.0,
      "step": 620
    },
    {
      "entropy": 1.806156025826931,
      "epoch": 1.3759562841530055,
      "grad_norm": 3.8762011528015137,
      "learning_rate": 1.7023399123925438e-05,
      "loss": 0.5669230461120606,
      "mean_token_accuracy": 0.828336314111948,
      "num_tokens": 2452196.0,
      "step": 630
    },
    {
      "entropy": 1.8862054750323296,
      "epoch": 1.3978142076502733,
      "grad_norm": 3.880317211151123,
      "learning_rate": 1.6922010241198445e-05,
      "loss": 0.4151432514190674,
      "mean_token_accuracy": 0.8618706040084362,
      "num_tokens": 2484719.0,
      "step": 640
    },
    {
      "entropy": 1.901587300747633,
      "epoch": 1.419672131147541,
      "grad_norm": 5.403715133666992,
      "learning_rate": 1.681923642910375e-05,
      "loss": 0.5961950778961181,
      "mean_token_accuracy": 0.8228115990757943,
      "num_tokens": 2523804.0,
      "step": 650
    },
    {
      "entropy": 1.8603504352271556,
      "epoch": 1.4415300546448089,
      "grad_norm": 2.0737757682800293,
      "learning_rate": 1.6715098250233095e-05,
      "loss": 0.5909753799438476,
      "mean_token_accuracy": 0.8104891542345285,
      "num_tokens": 2565708.0,
      "step": 660
    },
    {
      "entropy": 2.000732755661011,
      "epoch": 1.4633879781420764,
      "grad_norm": 4.641599655151367,
      "learning_rate": 1.660961654015552e-05,
      "loss": 0.5969127655029297,
      "mean_token_accuracy": 0.8471055120229721,
      "num_tokens": 2600157.0,
      "step": 670
    },
    {
      "entropy": 1.9003810167312623,
      "epoch": 1.4852459016393442,
      "grad_norm": 2.3957788944244385,
      "learning_rate": 1.650281240324864e-05,
      "loss": 0.4810887336730957,
      "mean_token_accuracy": 0.8310797978192568,
      "num_tokens": 2634747.0,
      "step": 680
    },
    {
      "entropy": 2.019043204188347,
      "epoch": 1.507103825136612,
      "grad_norm": 5.3165059089660645,
      "learning_rate": 1.6394707208476184e-05,
      "loss": 0.657526683807373,
      "mean_token_accuracy": 0.7817666057497263,
      "num_tokens": 2669560.0,
      "step": 690
    },
    {
      "entropy": 1.939988984167576,
      "epoch": 1.5289617486338798,
      "grad_norm": 3.616457462310791,
      "learning_rate": 1.6285322585112577e-05,
      "loss": 0.8010999679565429,
      "mean_token_accuracy": 0.7895378671586514,
      "num_tokens": 2712811.0,
      "step": 700
    },
    {
      "epoch": 1.5289617486338798,
      "eval_entropy": 1.9267194443382323,
      "eval_loss": 1.0051671266555786,
      "eval_mean_token_accuracy": 0.7411219369969331,
      "eval_num_tokens": 2712811.0,
      "eval_runtime": 25.2755,
      "eval_samples_per_second": 20.257,
      "eval_steps_per_second": 10.128,
      "step": 700
    },
    {
      "entropy": 1.9218722447752952,
      "epoch": 1.5508196721311476,
      "grad_norm": 3.080705165863037,
      "learning_rate": 1.6174680418415426e-05,
      "loss": 0.6289849758148194,
      "mean_token_accuracy": 0.882987642288208,
      "num_tokens": 2746923.0,
      "step": 710
    },
    {
      "entropy": 1.8692668572068214,
      "epoch": 1.5726775956284151,
      "grad_norm": 8.707109451293945,
      "learning_rate": 1.6062802845246796e-05,
      "loss": 0.5447452068328857,
      "mean_token_accuracy": 0.8194241803139448,
      "num_tokens": 2782316.0,
      "step": 720
    },
    {
      "entropy": 1.9840392470359802,
      "epoch": 1.594535519125683,
      "grad_norm": 5.839150428771973,
      "learning_rate": 1.5949712249644168e-05,
      "loss": 0.5748161792755127,
      "mean_token_accuracy": 0.8199563117697835,
      "num_tokens": 2812008.0,
      "step": 730
    },
    {
      "entropy": 1.877804670482874,
      "epoch": 1.6163934426229507,
      "grad_norm": 2.8095271587371826,
      "learning_rate": 1.5835431258341914e-05,
      "loss": 0.3862598896026611,
      "mean_token_accuracy": 0.8174357455223799,
      "num_tokens": 2851938.0,
      "step": 740
    },
    {
      "entropy": 1.8835655882954598,
      "epoch": 1.6382513661202185,
      "grad_norm": 11.515563011169434,
      "learning_rate": 1.571998273624424e-05,
      "loss": 0.5016937732696534,
      "mean_token_accuracy": 0.849822697788477,
      "num_tokens": 2885244.0,
      "step": 750
    },
    {
      "entropy": 1.7773355908691884,
      "epoch": 1.6601092896174863,
      "grad_norm": 6.528411388397217,
      "learning_rate": 1.5603389781850435e-05,
      "loss": 0.3264592170715332,
      "mean_token_accuracy": 0.8357654266059399,
      "num_tokens": 2926241.0,
      "step": 760
    },
    {
      "entropy": 1.8843942552804946,
      "epoch": 1.681967213114754,
      "grad_norm": 6.097729682922363,
      "learning_rate": 1.5485675722633445e-05,
      "loss": 0.6704532623291015,
      "mean_token_accuracy": 0.7745772052556277,
      "num_tokens": 2958842.0,
      "step": 770
    },
    {
      "entropy": 1.8674893841147422,
      "epoch": 1.7038251366120218,
      "grad_norm": 5.29870080947876,
      "learning_rate": 1.5366864110372586e-05,
      "loss": 0.6037456512451171,
      "mean_token_accuracy": 0.8349868554621935,
      "num_tokens": 2995790.0,
      "step": 780
    },
    {
      "entropy": 1.9335614942014216,
      "epoch": 1.7256830601092896,
      "grad_norm": 7.609033584594727,
      "learning_rate": 1.5246978716441392e-05,
      "loss": 0.478175687789917,
      "mean_token_accuracy": 0.8572088401764631,
      "num_tokens": 3034539.0,
      "step": 790
    },
    {
      "entropy": 1.9231948465108872,
      "epoch": 1.7475409836065574,
      "grad_norm": 3.5952892303466797,
      "learning_rate": 1.5126043527051523e-05,
      "loss": 0.5622974872589112,
      "mean_token_accuracy": 0.8431862521916628,
      "num_tokens": 3070216.0,
      "step": 800
    },
    {
      "epoch": 1.7475409836065574,
      "eval_entropy": 1.9174917968921363,
      "eval_loss": 0.9968341588973999,
      "eval_mean_token_accuracy": 0.7547392580891028,
      "eval_num_tokens": 3070216.0,
      "eval_runtime": 25.2412,
      "eval_samples_per_second": 20.284,
      "eval_steps_per_second": 10.142,
      "step": 800
    },
    {
      "entropy": 2.000046853721142,
      "epoch": 1.7693989071038252,
      "grad_norm": 5.004440784454346,
      "learning_rate": 1.5004082738453723e-05,
      "loss": 0.5045398235321045,
      "mean_token_accuracy": 0.8353351525962353,
      "num_tokens": 3104720.0,
      "step": 810
    },
    {
      "entropy": 1.932582141458988,
      "epoch": 1.791256830601093,
      "grad_norm": 4.402656555175781,
      "learning_rate": 1.4881120752096685e-05,
      "loss": 0.5935905933380127,
      "mean_token_accuracy": 0.7998956862837077,
      "num_tokens": 3142794.0,
      "step": 820
    },
    {
      "entropy": 1.9920608803629876,
      "epoch": 1.8131147540983608,
      "grad_norm": 10.046414375305176,
      "learning_rate": 1.4757182169744948e-05,
      "loss": 0.6299263000488281,
      "mean_token_accuracy": 0.824649827927351,
      "num_tokens": 3178323.0,
      "step": 830
    },
    {
      "entropy": 2.0002417393028735,
      "epoch": 1.8349726775956285,
      "grad_norm": 2.1592979431152344,
      "learning_rate": 1.463229178855663e-05,
      "loss": 0.42464284896850585,
      "mean_token_accuracy": 0.8876678507775069,
      "num_tokens": 3210741.0,
      "step": 840
    },
    {
      "entropy": 1.9218681126832962,
      "epoch": 1.8568306010928963,
      "grad_norm": 3.5906593799591064,
      "learning_rate": 1.4506474596122132e-05,
      "loss": 0.5482982635498047,
      "mean_token_accuracy": 0.8235908593982458,
      "num_tokens": 3253021.0,
      "step": 850
    },
    {
      "entropy": 1.8942916221916675,
      "epoch": 1.8786885245901639,
      "grad_norm": 2.9451968669891357,
      "learning_rate": 1.4379755765464709e-05,
      "loss": 0.6192903995513916,
      "mean_token_accuracy": 0.8629166536033154,
      "num_tokens": 3292097.0,
      "step": 860
    },
    {
      "entropy": 1.9666990574449301,
      "epoch": 1.9005464480874317,
      "grad_norm": 3.6715590953826904,
      "learning_rate": 1.4252160650003944e-05,
      "loss": 0.47508945465087893,
      "mean_token_accuracy": 0.8053756408393383,
      "num_tokens": 3330520.0,
      "step": 870
    },
    {
      "entropy": 1.9449772492051125,
      "epoch": 1.9224043715846995,
      "grad_norm": 6.027235507965088,
      "learning_rate": 1.4123714778483142e-05,
      "loss": 0.5430461883544921,
      "mean_token_accuracy": 0.8390930660068989,
      "num_tokens": 3367361.0,
      "step": 880
    },
    {
      "entropy": 1.9635326191782951,
      "epoch": 1.9442622950819672,
      "grad_norm": 3.413417100906372,
      "learning_rate": 1.3994443849861623e-05,
      "loss": 0.6275568962097168,
      "mean_token_accuracy": 0.8400135699659586,
      "num_tokens": 3410634.0,
      "step": 890
    },
    {
      "entropy": 1.899067870527506,
      "epoch": 1.966120218579235,
      "grad_norm": 4.991271018981934,
      "learning_rate": 1.3864373728172986e-05,
      "loss": 0.4935734272003174,
      "mean_token_accuracy": 0.8543500185012818,
      "num_tokens": 3461846.0,
      "step": 900
    },
    {
      "epoch": 1.966120218579235,
      "eval_entropy": 1.9543956331908703,
      "eval_loss": 0.9799688458442688,
      "eval_mean_token_accuracy": 0.7691519922809675,
      "eval_num_tokens": 3461846.0,
      "eval_runtime": 25.2895,
      "eval_samples_per_second": 20.246,
      "eval_steps_per_second": 10.123,
      "step": 900
    },
    {
      "entropy": 1.932174527645111,
      "epoch": 1.9879781420765026,
      "grad_norm": 4.647815227508545,
      "learning_rate": 1.3733530437350324e-05,
      "loss": 0.5729746341705322,
      "mean_token_accuracy": 0.8431255351752043,
      "num_tokens": 3508119.0,
      "step": 910
    },
    {
      "entropy": 1.9605209450972707,
      "epoch": 2.0087431693989073,
      "grad_norm": 5.845576763153076,
      "learning_rate": 1.3601940156019456e-05,
      "loss": 0.4193246364593506,
      "mean_token_accuracy": 0.8327035539244351,
      "num_tokens": 3539359.0,
      "step": 920
    },
    {
      "entropy": 2.0226026967167856,
      "epoch": 2.030601092896175,
      "grad_norm": 2.6995978355407715,
      "learning_rate": 1.346962921226121e-05,
      "loss": 0.37759575843811033,
      "mean_token_accuracy": 0.8809590727090836,
      "num_tokens": 3568555.0,
      "step": 930
    },
    {
      "entropy": 1.8960323825478553,
      "epoch": 2.0524590163934424,
      "grad_norm": 2.899524450302124,
      "learning_rate": 1.3336624078343801e-05,
      "loss": 0.5012131214141846,
      "mean_token_accuracy": 0.877274863421917,
      "num_tokens": 3602954.0,
      "step": 940
    },
    {
      "entropy": 1.7720212742686272,
      "epoch": 2.07431693989071,
      "grad_norm": 1.3420737981796265,
      "learning_rate": 1.3202951365426352e-05,
      "loss": 0.2880971908569336,
      "mean_token_accuracy": 0.8983731687068939,
      "num_tokens": 3641856.0,
      "step": 950
    },
    {
      "entropy": 1.7504506237804889,
      "epoch": 2.096174863387978,
      "grad_norm": 6.308726787567139,
      "learning_rate": 1.3068637818234652e-05,
      "loss": 0.5575133323669433,
      "mean_token_accuracy": 0.8421003326773644,
      "num_tokens": 3680686.0,
      "step": 960
    },
    {
      "entropy": 1.7622926771640777,
      "epoch": 2.1180327868852458,
      "grad_norm": 6.427295684814453,
      "learning_rate": 1.2933710309710162e-05,
      "loss": 0.37821121215820314,
      "mean_token_accuracy": 0.9191356211900711,
      "num_tokens": 3723403.0,
      "step": 970
    },
    {
      "entropy": 1.7686702385544777,
      "epoch": 2.1398907103825136,
      "grad_norm": 6.796016693115234,
      "learning_rate": 1.2798195835633373e-05,
      "loss": 0.45128607749938965,
      "mean_token_accuracy": 0.906242086738348,
      "num_tokens": 3756322.0,
      "step": 980
    },
    {
      "entropy": 1.7671647615730763,
      "epoch": 2.1617486338797813,
      "grad_norm": 4.9459733963012695,
      "learning_rate": 1.2662121509222607e-05,
      "loss": 0.36589665412902833,
      "mean_token_accuracy": 0.8796804122626781,
      "num_tokens": 3794193.0,
      "step": 990
    },
    {
      "entropy": 1.7161696419119834,
      "epoch": 2.183606557377049,
      "grad_norm": 9.006570816040039,
      "learning_rate": 1.2525514555709308e-05,
      "loss": 0.5742815971374512,
      "mean_token_accuracy": 0.8753333643078804,
      "num_tokens": 3830821.0,
      "step": 1000
    },
    {
      "epoch": 2.183606557377049,
      "eval_entropy": 1.743996654637158,
      "eval_loss": 1.1557646989822388,
      "eval_mean_token_accuracy": 0.7421538368216716,
      "eval_num_tokens": 3830821.0,
      "eval_runtime": 25.2766,
      "eval_samples_per_second": 20.256,
      "eval_steps_per_second": 10.128,
      "step": 1000
    },
    {
      "entropy": 1.7266477182507516,
      "epoch": 2.205464480874317,
      "grad_norm": 0.14489249885082245,
      "learning_rate": 1.2388402306890903e-05,
      "loss": 0.41260638236999514,
      "mean_token_accuracy": 0.9061720766127109,
      "num_tokens": 3874976.0,
      "step": 1010
    },
    {
      "entropy": 1.7216939367353916,
      "epoch": 2.2273224043715847,
      "grad_norm": 1.9707283973693848,
      "learning_rate": 1.2250812195662387e-05,
      "loss": 0.35822205543518065,
      "mean_token_accuracy": 0.8634335152804852,
      "num_tokens": 3905246.0,
      "step": 1020
    },
    {
      "entropy": 1.6640645649284125,
      "epoch": 2.2491803278688525,
      "grad_norm": 5.178168296813965,
      "learning_rate": 1.2112771750527632e-05,
      "loss": 0.40171108245849607,
      "mean_token_accuracy": 0.8396187137812376,
      "num_tokens": 3942601.0,
      "step": 1030
    },
    {
      "entropy": 1.7783586084842682,
      "epoch": 2.2710382513661203,
      "grad_norm": 11.327513694763184,
      "learning_rate": 1.1974308590091601e-05,
      "loss": 0.2785263299942017,
      "mean_token_accuracy": 0.870139553397894,
      "num_tokens": 3984000.0,
      "step": 1040
    },
    {
      "entropy": 1.6852504387497902,
      "epoch": 2.292896174863388,
      "grad_norm": 9.063831329345703,
      "learning_rate": 1.1835450417534517e-05,
      "loss": 0.5509501934051514,
      "mean_token_accuracy": 0.8589577071368695,
      "num_tokens": 4019526.0,
      "step": 1050
    },
    {
      "entropy": 1.6793502882122993,
      "epoch": 2.314754098360656,
      "grad_norm": 2.575770378112793,
      "learning_rate": 1.1696225015069128e-05,
      "loss": 0.3649094343185425,
      "mean_token_accuracy": 0.8885138727724552,
      "num_tokens": 4057765.0,
      "step": 1060
    },
    {
      "entropy": 1.7490564540028573,
      "epoch": 2.3366120218579236,
      "grad_norm": 14.544415473937988,
      "learning_rate": 1.1556660238382125e-05,
      "loss": 0.4366298675537109,
      "mean_token_accuracy": 0.8778704255819321,
      "num_tokens": 4093759.0,
      "step": 1070
    },
    {
      "entropy": 1.7166971430182456,
      "epoch": 2.3584699453551914,
      "grad_norm": 10.271318435668945,
      "learning_rate": 1.1416784011060904e-05,
      "loss": 0.25078210830688474,
      "mean_token_accuracy": 0.903883146867156,
      "num_tokens": 4127540.0,
      "step": 1080
    },
    {
      "entropy": 1.6313612181693316,
      "epoch": 2.380327868852459,
      "grad_norm": 3.0819733142852783,
      "learning_rate": 1.1276624319006722e-05,
      "loss": 0.33878474235534667,
      "mean_token_accuracy": 0.8705951869487762,
      "num_tokens": 4167769.0,
      "step": 1090
    },
    {
      "entropy": 1.6783426776528358,
      "epoch": 2.402185792349727,
      "grad_norm": 8.976804733276367,
      "learning_rate": 1.1136209204835364e-05,
      "loss": 0.33856441974639895,
      "mean_token_accuracy": 0.8913385391235351,
      "num_tokens": 4212055.0,
      "step": 1100
    },
    {
      "epoch": 2.402185792349727,
      "eval_entropy": 1.7138912174850702,
      "eval_loss": 1.1432414054870605,
      "eval_mean_token_accuracy": 0.7529156694072299,
      "eval_num_tokens": 4212055.0,
      "eval_runtime": 25.2663,
      "eval_samples_per_second": 20.264,
      "eval_steps_per_second": 10.132,
      "step": 1100
    },
    {
      "entropy": 1.676077627390623,
      "epoch": 2.4240437158469943,
      "grad_norm": 3.6485555171966553,
      "learning_rate": 1.0995566762266505e-05,
      "loss": 0.43614993095397947,
      "mean_token_accuracy": 0.8534008748829365,
      "num_tokens": 4253039.0,
      "step": 1110
    },
    {
      "entropy": 1.785607959330082,
      "epoch": 2.445901639344262,
      "grad_norm": 5.675464630126953,
      "learning_rate": 1.0854725130502815e-05,
      "loss": 0.4932537078857422,
      "mean_token_accuracy": 0.8849537067115307,
      "num_tokens": 4289265.0,
      "step": 1120
    },
    {
      "entropy": 1.6909036561846733,
      "epoch": 2.46775956284153,
      "grad_norm": 2.5323069095611572,
      "learning_rate": 1.0713712488599975e-05,
      "loss": 0.29662504196166994,
      "mean_token_accuracy": 0.906553142517805,
      "num_tokens": 4330290.0,
      "step": 1130
    },
    {
      "entropy": 1.7376374121755362,
      "epoch": 2.4896174863387976,
      "grad_norm": 8.417685508728027,
      "learning_rate": 1.057255704982872e-05,
      "loss": 0.2720386505126953,
      "mean_token_accuracy": 0.9060880910605192,
      "num_tokens": 4374908.0,
      "step": 1140
    },
    {
      "entropy": 1.8001229599118234,
      "epoch": 2.5114754098360654,
      "grad_norm": 5.350477695465088,
      "learning_rate": 1.0431287056030048e-05,
      "loss": 0.3160851240158081,
      "mean_token_accuracy": 0.9162062905728817,
      "num_tokens": 4415082.0,
      "step": 1150
    },
    {
      "entropy": 1.796884122490883,
      "epoch": 2.533333333333333,
      "grad_norm": 6.009316444396973,
      "learning_rate": 1.0289930771964678e-05,
      "loss": 0.34149200916290284,
      "mean_token_accuracy": 0.8666319936513901,
      "num_tokens": 4451123.0,
      "step": 1160
    },
    {
      "entropy": 1.7506244897842407,
      "epoch": 2.555191256830601,
      "grad_norm": 4.113317966461182,
      "learning_rate": 1.0148516479657955e-05,
      "loss": 0.46685094833374025,
      "mean_token_accuracy": 0.8790101282298565,
      "num_tokens": 4490088.0,
      "step": 1170
    },
    {
      "entropy": 1.745325656235218,
      "epoch": 2.577049180327869,
      "grad_norm": 0.6159476637840271,
      "learning_rate": 1.0007072472741303e-05,
      "loss": 0.45745162963867186,
      "mean_token_accuracy": 0.8782992571592331,
      "num_tokens": 4531211.0,
      "step": 1180
    },
    {
      "entropy": 1.755560566484928,
      "epoch": 2.5989071038251366,
      "grad_norm": 8.604616165161133,
      "learning_rate": 9.865627050791319e-06,
      "loss": 0.47327699661254885,
      "mean_token_accuracy": 0.8790083147585392,
      "num_tokens": 4573006.0,
      "step": 1190
    },
    {
      "entropy": 1.7254169069230556,
      "epoch": 2.6207650273224044,
      "grad_norm": 4.2413859367370605,
      "learning_rate": 9.724208513667723e-06,
      "loss": 0.39214675426483153,
      "mean_token_accuracy": 0.8431641358882189,
      "num_tokens": 4618083.0,
      "step": 1200
    },
    {
      "epoch": 2.6207650273224044,
      "eval_entropy": 1.7866586744785309,
      "eval_loss": 1.0981754064559937,
      "eval_mean_token_accuracy": 0.7460045842453837,
      "eval_num_tokens": 4618083.0,
      "eval_runtime": 25.3227,
      "eval_samples_per_second": 20.219,
      "eval_steps_per_second": 10.109,
      "step": 1200
    },
    {
      "entropy": 1.7824625477194787,
      "epoch": 2.642622950819672,
      "grad_norm": 7.563173294067383,
      "learning_rate": 9.582845155851219e-06,
      "loss": 0.5630898475646973,
      "mean_token_accuracy": 0.8865630835294723,
      "num_tokens": 4657609.0,
      "step": 1210
    },
    {
      "entropy": 1.7146922402083873,
      "epoch": 2.66448087431694,
      "grad_norm": 9.422724723815918,
      "learning_rate": 9.441565260782441e-06,
      "loss": 0.5290060997009277,
      "mean_token_accuracy": 0.8360320836305618,
      "num_tokens": 4697085.0,
      "step": 1220
    },
    {
      "entropy": 1.806973297894001,
      "epoch": 2.6863387978142077,
      "grad_norm": 4.4218854904174805,
      "learning_rate": 9.300397095203124e-06,
      "loss": 0.22829580307006836,
      "mean_token_accuracy": 0.9107354000210762,
      "num_tokens": 4728928.0,
      "step": 1230
    },
    {
      "entropy": 1.7696783646941185,
      "epoch": 2.7081967213114755,
      "grad_norm": 3.3204195499420166,
      "learning_rate": 9.15936890350059e-06,
      "loss": 0.34404170513153076,
      "mean_token_accuracy": 0.8584935840219259,
      "num_tokens": 4759301.0,
      "step": 1240
    },
    {
      "entropy": 1.7514332398772239,
      "epoch": 2.7300546448087433,
      "grad_norm": 4.120963096618652,
      "learning_rate": 9.018508902056716e-06,
      "loss": 0.42660813331604003,
      "mean_token_accuracy": 0.8891314808279276,
      "num_tokens": 4792127.0,
      "step": 1250
    },
    {
      "entropy": 1.7041171003133058,
      "epoch": 2.751912568306011,
      "grad_norm": 6.271151542663574,
      "learning_rate": 8.877845273602526e-06,
      "loss": 0.4269681930541992,
      "mean_token_accuracy": 0.8639719765633345,
      "num_tokens": 4840954.0,
      "step": 1260
    },
    {
      "entropy": 1.7781821444630623,
      "epoch": 2.773770491803279,
      "grad_norm": 9.52706527709961,
      "learning_rate": 8.737406161579478e-06,
      "loss": 0.4761667251586914,
      "mean_token_accuracy": 0.8543274749070406,
      "num_tokens": 4878725.0,
      "step": 1270
    },
    {
      "entropy": 1.7140948161482812,
      "epoch": 2.7956284153005466,
      "grad_norm": 7.590384006500244,
      "learning_rate": 8.597219664508646e-06,
      "loss": 0.3446077346801758,
      "mean_token_accuracy": 0.8726995788514614,
      "num_tokens": 4919138.0,
      "step": 1280
    },
    {
      "entropy": 1.8007852733135223,
      "epoch": 2.8174863387978144,
      "grad_norm": 3.562910795211792,
      "learning_rate": 8.457313830368862e-06,
      "loss": 0.43190617561340333,
      "mean_token_accuracy": 0.8488910354673862,
      "num_tokens": 4962188.0,
      "step": 1290
    },
    {
      "entropy": 1.7394358351826669,
      "epoch": 2.839344262295082,
      "grad_norm": 4.378752708435059,
      "learning_rate": 8.317716650984997e-06,
      "loss": 0.3554516315460205,
      "mean_token_accuracy": 0.900595785677433,
      "num_tokens": 5003129.0,
      "step": 1300
    },
    {
      "epoch": 2.839344262295082,
      "eval_entropy": 1.7719783522188663,
      "eval_loss": 1.0918433666229248,
      "eval_mean_token_accuracy": 0.7487576644634828,
      "eval_num_tokens": 5003129.0,
      "eval_runtime": 25.2737,
      "eval_samples_per_second": 20.258,
      "eval_steps_per_second": 10.129,
      "step": 1300
    },
    {
      "entropy": 1.697381130978465,
      "epoch": 2.86120218579235,
      "grad_norm": 5.49855375289917,
      "learning_rate": 8.17845605642745e-06,
      "loss": 0.2979349374771118,
      "mean_token_accuracy": 0.8746489737182855,
      "num_tokens": 5047985.0,
      "step": 1310
    },
    {
      "entropy": 1.7068176820874215,
      "epoch": 2.8830601092896178,
      "grad_norm": 4.538719654083252,
      "learning_rate": 8.039559909424028e-06,
      "loss": 0.4271398067474365,
      "mean_token_accuracy": 0.8372387941926718,
      "num_tokens": 5086478.0,
      "step": 1320
    },
    {
      "entropy": 1.7037811130285263,
      "epoch": 2.904918032786885,
      "grad_norm": 8.465956687927246,
      "learning_rate": 7.901055999785251e-06,
      "loss": 0.5102831840515136,
      "mean_token_accuracy": 0.8660007566213608,
      "num_tokens": 5125267.0,
      "step": 1330
    },
    {
      "entropy": 1.683811515569687,
      "epoch": 2.926775956284153,
      "grad_norm": 11.026825904846191,
      "learning_rate": 7.762972038844287e-06,
      "loss": 0.453878116607666,
      "mean_token_accuracy": 0.8719288915395736,
      "num_tokens": 5160129.0,
      "step": 1340
    },
    {
      "entropy": 1.712484674155712,
      "epoch": 2.9486338797814207,
      "grad_norm": 7.742736339569092,
      "learning_rate": 7.625335653912578e-06,
      "loss": 0.34582009315490725,
      "mean_token_accuracy": 0.8842464782297611,
      "num_tokens": 5194331.0,
      "step": 1350
    },
    {
      "entropy": 1.7974103182554244,
      "epoch": 2.9704918032786884,
      "grad_norm": 16.226985931396484,
      "learning_rate": 7.488174382752256e-06,
      "loss": 0.5330884456634521,
      "mean_token_accuracy": 0.8534679703414441,
      "num_tokens": 5244510.0,
      "step": 1360
    },
    {
      "entropy": 1.7614209264516831,
      "epoch": 2.9923497267759562,
      "grad_norm": 3.4853780269622803,
      "learning_rate": 7.351515668066524e-06,
      "loss": 0.353696346282959,
      "mean_token_accuracy": 0.9209138300269842,
      "num_tokens": 5280186.0,
      "step": 1370
    },
    {
      "entropy": 1.6901170454527203,
      "epoch": 3.0131147540983605,
      "grad_norm": 3.082343339920044,
      "learning_rate": 7.2153868520089884e-06,
      "loss": 0.46216726303100586,
      "mean_token_accuracy": 0.9046591481095866,
      "num_tokens": 5312685.0,
      "step": 1380
    },
    {
      "entropy": 1.6913618817925453,
      "epoch": 3.0349726775956283,
      "grad_norm": 3.6766366958618164,
      "learning_rate": 7.0798151707131705e-06,
      "loss": 0.31811132431030276,
      "mean_token_accuracy": 0.9293436668813229,
      "num_tokens": 5355921.0,
      "step": 1390
    },
    {
      "entropy": 1.6231368098407983,
      "epoch": 3.056830601092896,
      "grad_norm": 5.460843086242676,
      "learning_rate": 6.944827748843187e-06,
      "loss": 0.2959195613861084,
      "mean_token_accuracy": 0.9214666485786438,
      "num_tokens": 5396513.0,
      "step": 1400
    },
    {
      "epoch": 3.056830601092896,
      "eval_entropy": 1.6358041786588728,
      "eval_loss": 1.2711889743804932,
      "eval_mean_token_accuracy": 0.7361913194181398,
      "eval_num_tokens": 5396513.0,
      "eval_runtime": 25.2843,
      "eval_samples_per_second": 20.25,
      "eval_steps_per_second": 10.125,
      "step": 1400
    },
    {
      "entropy": 1.6227923020720483,
      "epoch": 3.078688524590164,
      "grad_norm": 5.745507717132568,
      "learning_rate": 6.810451594166778e-06,
      "loss": 0.3128030776977539,
      "mean_token_accuracy": 0.923306229710579,
      "num_tokens": 5439874.0,
      "step": 1410
    },
    {
      "entropy": 1.5370216600596904,
      "epoch": 3.1005464480874316,
      "grad_norm": 3.301435708999634,
      "learning_rate": 6.676713592151663e-06,
      "loss": 0.1619083046913147,
      "mean_token_accuracy": 0.9161941222846508,
      "num_tokens": 5482643.0,
      "step": 1420
    },
    {
      "entropy": 1.5821660324931144,
      "epoch": 3.1224043715846994,
      "grad_norm": 6.687673568725586,
      "learning_rate": 6.5436405005864215e-06,
      "loss": 0.2582197666168213,
      "mean_token_accuracy": 0.9377869054675102,
      "num_tokens": 5520901.0,
      "step": 1430
    },
    {
      "entropy": 1.598063986003399,
      "epoch": 3.144262295081967,
      "grad_norm": 1.1611300706863403,
      "learning_rate": 6.4112589442268855e-06,
      "loss": 0.22129406929016113,
      "mean_token_accuracy": 0.9037676326930523,
      "num_tokens": 5554607.0,
      "step": 1440
    },
    {
      "entropy": 1.5651034712791443,
      "epoch": 3.166120218579235,
      "grad_norm": 14.291632652282715,
      "learning_rate": 6.279595409469169e-06,
      "loss": 0.2598989963531494,
      "mean_token_accuracy": 0.9124826833605766,
      "num_tokens": 5597787.0,
      "step": 1450
    },
    {
      "entropy": 1.6206938516348601,
      "epoch": 3.1879781420765028,
      "grad_norm": 2.9516093730926514,
      "learning_rate": 6.148676239050366e-06,
      "loss": 0.3642241954803467,
      "mean_token_accuracy": 0.9069154664874077,
      "num_tokens": 5637024.0,
      "step": 1460
    },
    {
      "entropy": 1.551618180796504,
      "epoch": 3.2098360655737705,
      "grad_norm": 7.661910533905029,
      "learning_rate": 6.0185276267779965e-06,
      "loss": 0.19039158821105956,
      "mean_token_accuracy": 0.9062366366386414,
      "num_tokens": 5677024.0,
      "step": 1470
    },
    {
      "entropy": 1.5639437273144723,
      "epoch": 3.2316939890710383,
      "grad_norm": 5.479608058929443,
      "learning_rate": 5.889175612289251e-06,
      "loss": 0.29235665798187255,
      "mean_token_accuracy": 0.9100384078919888,
      "num_tokens": 5717446.0,
      "step": 1480
    },
    {
      "entropy": 1.5362057872116566,
      "epoch": 3.253551912568306,
      "grad_norm": 10.538714408874512,
      "learning_rate": 5.7606460758410655e-06,
      "loss": 0.26177852153778075,
      "mean_token_accuracy": 0.9071271993219853,
      "num_tokens": 5760098.0,
      "step": 1490
    },
    {
      "entropy": 1.5072948724031447,
      "epoch": 3.275409836065574,
      "grad_norm": 10.498205184936523,
      "learning_rate": 5.632964733132142e-06,
      "loss": 0.3608083724975586,
      "mean_token_accuracy": 0.9086304809898138,
      "num_tokens": 5794405.0,
      "step": 1500
    },
    {
      "epoch": 3.275409836065574,
      "eval_entropy": 1.5782789755612612,
      "eval_loss": 1.2997822761535645,
      "eval_mean_token_accuracy": 0.74409809790086,
      "eval_num_tokens": 5794405.0,
      "eval_runtime": 25.2677,
      "eval_samples_per_second": 20.263,
      "eval_steps_per_second": 10.131,
      "step": 1500
    },
    {
      "entropy": 1.5180404711514712,
      "epoch": 3.2972677595628417,
      "grad_norm": 3.2218379974365234,
      "learning_rate": 5.506157130157788e-06,
      "loss": 0.17872884273529052,
      "mean_token_accuracy": 0.9064229533076287,
      "num_tokens": 5834688.0,
      "step": 1510
    },
    {
      "entropy": 1.534475613385439,
      "epoch": 3.3191256830601095,
      "grad_norm": 11.32108211517334,
      "learning_rate": 5.380248638098821e-06,
      "loss": 0.286358642578125,
      "mean_token_accuracy": 0.9199724584817887,
      "num_tokens": 5878545.0,
      "step": 1520
    },
    {
      "entropy": 1.6077978804707527,
      "epoch": 3.3409836065573773,
      "grad_norm": 6.171378135681152,
      "learning_rate": 5.255264448245366e-06,
      "loss": 0.2198193073272705,
      "mean_token_accuracy": 0.9341178297996521,
      "num_tokens": 5918379.0,
      "step": 1530
    },
    {
      "entropy": 1.6092828884720802,
      "epoch": 3.362841530054645,
      "grad_norm": 5.977477550506592,
      "learning_rate": 5.1312295669567085e-06,
      "loss": 0.3700289487838745,
      "mean_token_accuracy": 0.8991808846592904,
      "num_tokens": 5950574.0,
      "step": 1540
    },
    {
      "entropy": 1.5695540212094783,
      "epoch": 3.384699453551913,
      "grad_norm": 0.7643852829933167,
      "learning_rate": 5.0081688106581e-06,
      "loss": 0.19067635536193847,
      "mean_token_accuracy": 0.9202431552112103,
      "num_tokens": 5987585.0,
      "step": 1550
    },
    {
      "entropy": 1.5746611192822457,
      "epoch": 3.40655737704918,
      "grad_norm": 5.357200622558594,
      "learning_rate": 4.886106800875572e-06,
      "loss": 0.26554481983184813,
      "mean_token_accuracy": 0.8943272069096565,
      "num_tokens": 6024581.0,
      "step": 1560
    },
    {
      "entropy": 1.591781959682703,
      "epoch": 3.428415300546448,
      "grad_norm": 11.075048446655273,
      "learning_rate": 4.765067959309782e-06,
      "loss": 0.2532616138458252,
      "mean_token_accuracy": 0.9050811067223549,
      "num_tokens": 6054472.0,
      "step": 1570
    },
    {
      "entropy": 1.6494382977485658,
      "epoch": 3.4502732240437157,
      "grad_norm": 6.1315693855285645,
      "learning_rate": 4.6450765029497945e-06,
      "loss": 0.2048194408416748,
      "mean_token_accuracy": 0.9233898296952248,
      "num_tokens": 6092006.0,
      "step": 1580
    },
    {
      "entropy": 1.5534474547952413,
      "epoch": 3.4721311475409835,
      "grad_norm": 4.249480724334717,
      "learning_rate": 4.526156439227848e-06,
      "loss": 0.3019617795944214,
      "mean_token_accuracy": 0.9094546407461166,
      "num_tokens": 6130677.0,
      "step": 1590
    },
    {
      "entropy": 1.616061332821846,
      "epoch": 3.4939890710382513,
      "grad_norm": 15.559029579162598,
      "learning_rate": 4.408331561216022e-06,
      "loss": 0.15720347166061402,
      "mean_token_accuracy": 0.9568640798330307,
      "num_tokens": 6168361.0,
      "step": 1600
    },
    {
      "epoch": 3.4939890710382513,
      "eval_entropy": 1.5903846183791757,
      "eval_loss": 1.3212494850158691,
      "eval_mean_token_accuracy": 0.7415331557858735,
      "eval_num_tokens": 6168361.0,
      "eval_runtime": 25.2585,
      "eval_samples_per_second": 20.27,
      "eval_steps_per_second": 10.135,
      "step": 1600
    },
    {
      "entropy": 1.5285045128315686,
      "epoch": 3.515846994535519,
      "grad_norm": 7.225104331970215,
      "learning_rate": 4.291625442865836e-06,
      "loss": 0.27741916179656984,
      "mean_token_accuracy": 0.9309419341385364,
      "num_tokens": 6217528.0,
      "step": 1610
    },
    {
      "entropy": 1.5996034383773803,
      "epoch": 3.537704918032787,
      "grad_norm": 8.290458679199219,
      "learning_rate": 4.176061434291637e-06,
      "loss": 0.3423781156539917,
      "mean_token_accuracy": 0.8994432087987662,
      "num_tokens": 6253094.0,
      "step": 1620
    },
    {
      "entropy": 1.6044300124049187,
      "epoch": 3.5595628415300546,
      "grad_norm": 7.356506824493408,
      "learning_rate": 4.061662657098824e-06,
      "loss": 0.3168341875076294,
      "mean_token_accuracy": 0.9347146339714527,
      "num_tokens": 6292894.0,
      "step": 1630
    },
    {
      "entropy": 1.5532973259687424,
      "epoch": 3.5814207650273224,
      "grad_norm": 10.509337425231934,
      "learning_rate": 3.948451999757744e-06,
      "loss": 0.3533791542053223,
      "mean_token_accuracy": 0.9503577001392841,
      "num_tokens": 6329002.0,
      "step": 1640
    },
    {
      "entropy": 1.5191884014755488,
      "epoch": 3.60327868852459,
      "grad_norm": 4.559752941131592,
      "learning_rate": 3.836452113024271e-06,
      "loss": 0.22230336666107178,
      "mean_token_accuracy": 0.9372093766927719,
      "num_tokens": 6368568.0,
      "step": 1650
    },
    {
      "entropy": 1.6334565032273531,
      "epoch": 3.625136612021858,
      "grad_norm": 8.851140975952148,
      "learning_rate": 3.725685405407904e-06,
      "loss": 0.18717575073242188,
      "mean_token_accuracy": 0.9109462410211563,
      "num_tokens": 6400868.0,
      "step": 1660
    },
    {
      "entropy": 1.5916602939367295,
      "epoch": 3.646994535519126,
      "grad_norm": 3.6515164375305176,
      "learning_rate": 3.6161740386883916e-06,
      "loss": 0.29880824089050295,
      "mean_token_accuracy": 0.8781519372016191,
      "num_tokens": 6434878.0,
      "step": 1670
    },
    {
      "entropy": 1.5508679054677486,
      "epoch": 3.6688524590163936,
      "grad_norm": 5.768099784851074,
      "learning_rate": 3.5079399234816636e-06,
      "loss": 0.34449656009674073,
      "mean_token_accuracy": 0.9187621869146824,
      "num_tokens": 6482039.0,
      "step": 1680
    },
    {
      "entropy": 1.6093040138483048,
      "epoch": 3.6907103825136613,
      "grad_norm": 10.039252281188965,
      "learning_rate": 3.4010047148560522e-06,
      "loss": 0.3076228141784668,
      "mean_token_accuracy": 0.9134237676858902,
      "num_tokens": 6519513.0,
      "step": 1690
    },
    {
      "entropy": 1.5371469020843507,
      "epoch": 3.7125683060109287,
      "grad_norm": 4.985898971557617,
      "learning_rate": 3.2953898079996092e-06,
      "loss": 0.35710718631744387,
      "mean_token_accuracy": 0.922921521216631,
      "num_tokens": 6556344.0,
      "step": 1700
    },
    {
      "epoch": 3.7125683060109287,
      "eval_entropy": 1.5996102751232684,
      "eval_loss": 1.310859203338623,
      "eval_mean_token_accuracy": 0.7403237737016752,
      "eval_num_tokens": 6556344.0,
      "eval_runtime": 25.2734,
      "eval_samples_per_second": 20.258,
      "eval_steps_per_second": 10.129,
      "step": 1700
    },
    {
      "entropy": 1.5552573069930076,
      "epoch": 3.7344262295081965,
      "grad_norm": 5.772418975830078,
      "learning_rate": 3.191116333939469e-06,
      "loss": 0.18078577518463135,
      "mean_token_accuracy": 0.9119489409029484,
      "num_tokens": 6593300.0,
      "step": 1710
    },
    {
      "entropy": 1.5171513885259629,
      "epoch": 3.7562841530054643,
      "grad_norm": 14.933680534362793,
      "learning_rate": 3.088205155313997e-06,
      "loss": 0.21673390865325928,
      "mean_token_accuracy": 0.9187557071447372,
      "num_tokens": 6640279.0,
      "step": 1720
    },
    {
      "entropy": 1.6221169739961625,
      "epoch": 3.778142076502732,
      "grad_norm": 3.3472113609313965,
      "learning_rate": 2.986676862198714e-06,
      "loss": 0.17331513166427612,
      "mean_token_accuracy": 0.928995156288147,
      "num_tokens": 6676193.0,
      "step": 1730
    },
    {
      "entropy": 1.6270448446273804,
      "epoch": 3.8,
      "grad_norm": 6.625114440917969,
      "learning_rate": 2.8865517679866728e-06,
      "loss": 0.28036065101623536,
      "mean_token_accuracy": 0.8961473934352397,
      "num_tokens": 6714066.0,
      "step": 1740
    },
    {
      "entropy": 1.5236708283424378,
      "epoch": 3.8218579234972676,
      "grad_norm": 8.20154094696045,
      "learning_rate": 2.7878499053242593e-06,
      "loss": 0.3181121826171875,
      "mean_token_accuracy": 0.9385631062090397,
      "num_tokens": 6754143.0,
      "step": 1750
    },
    {
      "entropy": 1.548745909333229,
      "epoch": 3.8437158469945354,
      "grad_norm": 8.136096000671387,
      "learning_rate": 2.6905910221031186e-06,
      "loss": 0.14712284803390502,
      "mean_token_accuracy": 0.9373674474656581,
      "num_tokens": 6789949.0,
      "step": 1760
    },
    {
      "entropy": 1.6179724402725697,
      "epoch": 3.865573770491803,
      "grad_norm": 8.492207527160645,
      "learning_rate": 2.59479457750909e-06,
      "loss": 0.22144730091094972,
      "mean_token_accuracy": 0.9098515301942826,
      "num_tokens": 6823239.0,
      "step": 1770
    },
    {
      "entropy": 1.5828871674835683,
      "epoch": 3.887431693989071,
      "grad_norm": 4.814517021179199,
      "learning_rate": 2.500479738128879e-06,
      "loss": 0.19235601425170898,
      "mean_token_accuracy": 0.9154853902757167,
      "num_tokens": 6864003.0,
      "step": 1780
    },
    {
      "entropy": 1.5566079840064049,
      "epoch": 3.9092896174863387,
      "grad_norm": 1.330918550491333,
      "learning_rate": 2.4076653741152687e-06,
      "loss": 0.1576427102088928,
      "mean_token_accuracy": 0.9358552634716034,
      "num_tokens": 6893305.0,
      "step": 1790
    },
    {
      "entropy": 1.5818458020687103,
      "epoch": 3.9311475409836065,
      "grad_norm": 7.912631511688232,
      "learning_rate": 2.3163700554116765e-06,
      "loss": 0.3152014255523682,
      "mean_token_accuracy": 0.9117202915251255,
      "num_tokens": 6931798.0,
      "step": 1800
    },
    {
      "epoch": 3.9311475409836065,
      "eval_entropy": 1.5581459198147058,
      "eval_loss": 1.3899604082107544,
      "eval_mean_token_accuracy": 0.7383179743774235,
      "eval_num_tokens": 6931798.0,
      "eval_runtime": 25.2597,
      "eval_samples_per_second": 20.269,
      "eval_steps_per_second": 10.135,
      "step": 1800
    },
    {
      "entropy": 1.522872880101204,
      "epoch": 3.9530054644808743,
      "grad_norm": 9.055262565612793,
      "learning_rate": 2.2266120480367202e-06,
      "loss": 0.15156043767929078,
      "mean_token_accuracy": 0.9299404136836529,
      "num_tokens": 6966554.0,
      "step": 1810
    },
    {
      "entropy": 1.6150698214769363,
      "epoch": 3.974863387978142,
      "grad_norm": 3.513589859008789,
      "learning_rate": 2.1384093104296624e-06,
      "loss": 0.32263479232788084,
      "mean_token_accuracy": 0.9067349448800087,
      "num_tokens": 7000411.0,
      "step": 1820
    },
    {
      "entropy": 1.5076138753443957,
      "epoch": 3.99672131147541,
      "grad_norm": 4.524299144744873,
      "learning_rate": 2.0517794898573075e-06,
      "loss": 0.21262218952178955,
      "mean_token_accuracy": 0.9192514590919018,
      "num_tokens": 7047052.0,
      "step": 1830
    },
    {
      "entropy": 1.5324510986867703,
      "epoch": 4.017486338797815,
      "grad_norm": 11.481642723083496,
      "learning_rate": 1.9667399188832448e-06,
      "loss": 0.11275914907455445,
      "mean_token_accuracy": 0.9351930963365656,
      "num_tokens": 7078248.0,
      "step": 1840
    },
    {
      "entropy": 1.548930251598358,
      "epoch": 4.039344262295082,
      "grad_norm": 3.9623751640319824,
      "learning_rate": 1.883307611899986e-06,
      "loss": 0.15659793615341186,
      "mean_token_accuracy": 0.9287457197904587,
      "num_tokens": 7124139.0,
      "step": 1850
    },
    {
      "entropy": 1.4768053125590086,
      "epoch": 4.06120218579235,
      "grad_norm": 4.408417701721191,
      "learning_rate": 1.8014992617248184e-06,
      "loss": 0.12531669139862062,
      "mean_token_accuracy": 0.9682689242064952,
      "num_tokens": 7164293.0,
      "step": 1860
    },
    {
      "entropy": 1.4974569860845803,
      "epoch": 4.083060109289617,
      "grad_norm": 6.734787464141846,
      "learning_rate": 1.7213312362599376e-06,
      "loss": 0.09787633419036865,
      "mean_token_accuracy": 0.9446013942360878,
      "num_tokens": 7205071.0,
      "step": 1870
    },
    {
      "entropy": 1.4866872608661652,
      "epoch": 4.104918032786885,
      "grad_norm": 1.529787540435791,
      "learning_rate": 1.6428195752176512e-06,
      "loss": 0.21155085563659667,
      "mean_token_accuracy": 0.9456927172839642,
      "num_tokens": 7236779.0,
      "step": 1880
    },
    {
      "entropy": 1.467564422264695,
      "epoch": 4.126775956284153,
      "grad_norm": 10.446721076965332,
      "learning_rate": 1.565979986911189e-06,
      "loss": 0.30939359664916993,
      "mean_token_accuracy": 0.90986697524786,
      "num_tokens": 7272026.0,
      "step": 1890
    },
    {
      "entropy": 1.4936597064137458,
      "epoch": 4.14863387978142,
      "grad_norm": 1.104892373085022,
      "learning_rate": 1.4908278451118629e-06,
      "loss": 0.24086241722106932,
      "mean_token_accuracy": 0.9398242756724358,
      "num_tokens": 7310448.0,
      "step": 1900
    },
    {
      "epoch": 4.14863387978142,
      "eval_entropy": 1.508402063511312,
      "eval_loss": 1.5420618057250977,
      "eval_mean_token_accuracy": 0.7262968142749742,
      "eval_num_tokens": 7310448.0,
      "eval_runtime": 25.2826,
      "eval_samples_per_second": 20.251,
      "eval_steps_per_second": 10.126,
      "step": 1900
    },
    {
      "entropy": 1.526249935477972,
      "epoch": 4.170491803278688,
      "grad_norm": 7.124038219451904,
      "learning_rate": 1.4173781859731294e-06,
      "loss": 0.20966436862945556,
      "mean_token_accuracy": 0.9498579859733581,
      "num_tokens": 7349305.0,
      "step": 1910
    },
    {
      "entropy": 1.5114058703184128,
      "epoch": 4.192349726775956,
      "grad_norm": 5.5153279304504395,
      "learning_rate": 1.3456457050222227e-06,
      "loss": 0.13045576810836793,
      "mean_token_accuracy": 0.9261387333273887,
      "num_tokens": 7378315.0,
      "step": 1920
    },
    {
      "entropy": 1.5217763639986515,
      "epoch": 4.214207650273224,
      "grad_norm": 10.250426292419434,
      "learning_rate": 1.2756447542199146e-06,
      "loss": 0.16847167015075684,
      "mean_token_accuracy": 0.9444297157227993,
      "num_tokens": 7416219.0,
      "step": 1930
    },
    {
      "entropy": 1.4267330523580313,
      "epoch": 4.2360655737704915,
      "grad_norm": 1.7956056594848633,
      "learning_rate": 1.207389339089049e-06,
      "loss": 0.14541631937026978,
      "mean_token_accuracy": 0.9462509617209435,
      "num_tokens": 7455837.0,
      "step": 1940
    },
    {
      "entropy": 1.4471466541290283,
      "epoch": 4.257923497267759,
      "grad_norm": 12.950838088989258,
      "learning_rate": 1.140893115912358e-06,
      "loss": 0.25396480560302737,
      "mean_token_accuracy": 0.9296876281499863,
      "num_tokens": 7491173.0,
      "step": 1950
    },
    {
      "entropy": 1.480190312117338,
      "epoch": 4.279781420765027,
      "grad_norm": 18.943504333496094,
      "learning_rate": 1.0761693890001624e-06,
      "loss": 0.12710440158843994,
      "mean_token_accuracy": 0.9423264317214489,
      "num_tokens": 7532165.0,
      "step": 1960
    },
    {
      "entropy": 1.4895855821669102,
      "epoch": 4.301639344262295,
      "grad_norm": 15.379076957702637,
      "learning_rate": 1.0132311080285061e-06,
      "loss": 0.13325932025909423,
      "mean_token_accuracy": 0.9429835468530655,
      "num_tokens": 7578121.0,
      "step": 1970
    },
    {
      "entropy": 1.4661264941096306,
      "epoch": 4.323497267759563,
      "grad_norm": 9.257176399230957,
      "learning_rate": 9.520908654482308e-07,
      "loss": 0.18194167613983153,
      "mean_token_accuracy": 0.9348225675523281,
      "num_tokens": 7615493.0,
      "step": 1980
    },
    {
      "entropy": 1.487341583520174,
      "epoch": 4.3453551912568305,
      "grad_norm": 8.9894380569458,
      "learning_rate": 8.927608939655185e-07,
      "loss": 0.28666565418243406,
      "mean_token_accuracy": 0.9525417745113373,
      "num_tokens": 7657601.0,
      "step": 1990
    },
    {
      "entropy": 1.4775039009749888,
      "epoch": 4.367213114754098,
      "grad_norm": 6.236198902130127,
      "learning_rate": 8.352530640944312e-07,
      "loss": 0.3593610286712646,
      "mean_token_accuracy": 0.9542292051017285,
      "num_tokens": 7695168.0,
      "step": 2000
    },
    {
      "epoch": 4.367213114754098,
      "eval_entropy": 1.4882210597861558,
      "eval_loss": 1.5714366436004639,
      "eval_mean_token_accuracy": 0.7317562090465799,
      "eval_num_tokens": 7695168.0,
      "eval_runtime": 25.2945,
      "eval_samples_per_second": 20.242,
      "eval_steps_per_second": 10.121,
      "step": 2000
    },
    {
      "entropy": 1.4730473149567842,
      "epoch": 4.389071038251366,
      "grad_norm": 12.652424812316895,
      "learning_rate": 7.795788817818972e-07,
      "loss": 0.0923178493976593,
      "mean_token_accuracy": 0.9668491549789906,
      "num_tokens": 7733619.0,
      "step": 2010
    },
    {
      "entropy": 1.5186619710177183,
      "epoch": 4.410928961748634,
      "grad_norm": 1.527182698249817,
      "learning_rate": 7.257494861056402e-07,
      "loss": 0.23416218757629395,
      "mean_token_accuracy": 0.9748448759317399,
      "num_tokens": 7771339.0,
      "step": 2020
    },
    {
      "entropy": 1.4672299686819315,
      "epoch": 4.432786885245902,
      "grad_norm": 10.213470458984375,
      "learning_rate": 6.737756470455259e-07,
      "loss": 0.20572996139526367,
      "mean_token_accuracy": 0.9412943184375763,
      "num_tokens": 7814745.0,
      "step": 2030
    },
    {
      "entropy": 1.4385718286037446,
      "epoch": 4.454644808743169,
      "grad_norm": 6.762119293212891,
      "learning_rate": 6.236677633287403e-07,
      "loss": 0.2968504190444946,
      "mean_token_accuracy": 0.945137070864439,
      "num_tokens": 7850365.0,
      "step": 2040
    },
    {
      "entropy": 1.48858273178339,
      "epoch": 4.476502732240437,
      "grad_norm": 1.248301386833191,
      "learning_rate": 5.754358603492505e-07,
      "loss": 0.08900911211967469,
      "mean_token_accuracy": 0.9440581537783146,
      "num_tokens": 7892276.0,
      "step": 2050
    },
    {
      "entropy": 1.439606611058116,
      "epoch": 4.498360655737705,
      "grad_norm": 6.034962177276611,
      "learning_rate": 5.290895881619618e-07,
      "loss": 0.17720189094543456,
      "mean_token_accuracy": 0.951624260097742,
      "num_tokens": 7930929.0,
      "step": 2060
    },
    {
      "entropy": 1.4914564713835716,
      "epoch": 4.520218579234973,
      "grad_norm": 10.574368476867676,
      "learning_rate": 4.846382195519772e-07,
      "loss": 0.10065207481384278,
      "mean_token_accuracy": 0.9822921209037304,
      "num_tokens": 7971276.0,
      "step": 2070
    },
    {
      "entropy": 1.4798044867813587,
      "epoch": 4.5420765027322405,
      "grad_norm": 1.7470002174377441,
      "learning_rate": 4.420906481793252e-07,
      "loss": 0.0717387318611145,
      "mean_token_accuracy": 0.9487549245357514,
      "num_tokens": 8004006.0,
      "step": 2080
    },
    {
      "entropy": 1.513572445511818,
      "epoch": 4.563934426229508,
      "grad_norm": 8.526530265808105,
      "learning_rate": 4.014553867995641e-07,
      "loss": 0.17615694999694825,
      "mean_token_accuracy": 0.9310802616178989,
      "num_tokens": 8042109.0,
      "step": 2090
    },
    {
      "entropy": 1.4915996737778188,
      "epoch": 4.585792349726776,
      "grad_norm": 3.867147207260132,
      "learning_rate": 3.6274056556056425e-07,
      "loss": 0.38680167198181153,
      "mean_token_accuracy": 0.9497635766863823,
      "num_tokens": 8079012.0,
      "step": 2100
    },
    {
      "epoch": 4.585792349726776,
      "eval_entropy": 1.4877252136357129,
      "eval_loss": 1.5858014822006226,
      "eval_mean_token_accuracy": 0.730545868864283,
      "eval_num_tokens": 8079012.0,
      "eval_runtime": 25.2797,
      "eval_samples_per_second": 20.253,
      "eval_steps_per_second": 10.127,
      "step": 2100
    },
    {
      "entropy": 1.5772681519389153,
      "epoch": 4.607650273224044,
      "grad_norm": 8.74162769317627,
      "learning_rate": 3.259539303758752e-07,
      "loss": 0.17112650871276855,
      "mean_token_accuracy": 0.9037679031491279,
      "num_tokens": 8113891.0,
      "step": 2110
    },
    {
      "entropy": 1.450429831445217,
      "epoch": 4.629508196721312,
      "grad_norm": 3.486109972000122,
      "learning_rate": 2.911028413749395e-07,
      "loss": 0.12542186975479125,
      "mean_token_accuracy": 0.9641172751784325,
      "num_tokens": 8158656.0,
      "step": 2120
    },
    {
      "entropy": 1.457353810966015,
      "epoch": 4.651366120218579,
      "grad_norm": 5.5825090408325195,
      "learning_rate": 2.581942714305219e-07,
      "loss": 0.11810015439987183,
      "mean_token_accuracy": 0.9155681878328323,
      "num_tokens": 8208969.0,
      "step": 2130
    },
    {
      "entropy": 1.4100075215101242,
      "epoch": 4.673224043715847,
      "grad_norm": 10.298120498657227,
      "learning_rate": 2.2723480476358529e-07,
      "loss": 0.20408346652984619,
      "mean_token_accuracy": 0.9427890285849572,
      "num_tokens": 8248876.0,
      "step": 2140
    },
    {
      "entropy": 1.4350176867097617,
      "epoch": 4.695081967213115,
      "grad_norm": 1.2466769218444824,
      "learning_rate": 1.9823063562596224e-07,
      "loss": 0.07048531770706176,
      "mean_token_accuracy": 0.9548317432403565,
      "num_tokens": 8284270.0,
      "step": 2150
    },
    {
      "entropy": 1.4886561691761018,
      "epoch": 4.716939890710383,
      "grad_norm": 7.229863166809082,
      "learning_rate": 1.7118756706101748e-07,
      "loss": 0.1607893705368042,
      "mean_token_accuracy": 0.9643361568450928,
      "num_tokens": 8317245.0,
      "step": 2160
    },
    {
      "entropy": 1.4858718007802962,
      "epoch": 4.738797814207651,
      "grad_norm": 7.945173740386963,
      "learning_rate": 1.4611100974260994e-07,
      "loss": 0.2605283737182617,
      "mean_token_accuracy": 0.9339787639677525,
      "num_tokens": 8359492.0,
      "step": 2170
    },
    {
      "entropy": 1.4697067685425282,
      "epoch": 4.760655737704918,
      "grad_norm": 2.577718496322632,
      "learning_rate": 1.230059808925377e-07,
      "loss": 0.23986785411834716,
      "mean_token_accuracy": 0.9224813591688872,
      "num_tokens": 8392450.0,
      "step": 2180
    },
    {
      "entropy": 1.4250585563480853,
      "epoch": 4.782513661202186,
      "grad_norm": 3.4977567195892334,
      "learning_rate": 1.0187710327671297e-07,
      "loss": 0.1858036160469055,
      "mean_token_accuracy": 0.9554606847465038,
      "num_tokens": 8436208.0,
      "step": 2190
    },
    {
      "entropy": 1.4355518952012063,
      "epoch": 4.804371584699454,
      "grad_norm": 2.248436450958252,
      "learning_rate": 8.272860428026108e-08,
      "loss": 0.27360637187957765,
      "mean_token_accuracy": 0.9597639583051205,
      "num_tokens": 8474217.0,
      "step": 2200
    },
    {
      "epoch": 4.804371584699454,
      "eval_entropy": 1.48676719609648,
      "eval_loss": 1.57476806640625,
      "eval_mean_token_accuracy": 0.733519333647564,
      "eval_num_tokens": 8474217.0,
      "eval_runtime": 25.278,
      "eval_samples_per_second": 20.255,
      "eval_steps_per_second": 10.127,
      "step": 2200
    },
    {
      "entropy": 1.4184951461851596,
      "epoch": 4.826229508196722,
      "grad_norm": 3.388068675994873,
      "learning_rate": 6.556431506171357e-08,
      "loss": 0.2510469198226929,
      "mean_token_accuracy": 0.9551761075854301,
      "num_tokens": 8512910.0,
      "step": 2210
    },
    {
      "entropy": 1.5154303699731826,
      "epoch": 4.848087431693989,
      "grad_norm": 8.467205047607422,
      "learning_rate": 5.038766978649356e-08,
      "loss": 0.1511959910392761,
      "mean_token_accuracy": 0.9454753026366234,
      "num_tokens": 8550362.0,
      "step": 2220
    },
    {
      "entropy": 1.474125138297677,
      "epoch": 4.869945355191257,
      "grad_norm": 4.428479194641113,
      "learning_rate": 3.7201704939813233e-08,
      "loss": 0.1717713952064514,
      "mean_token_accuracy": 0.9650007724761963,
      "num_tokens": 8593901.0,
      "step": 2230
    },
    {
      "entropy": 1.507462577521801,
      "epoch": 4.891803278688524,
      "grad_norm": 4.86669921875,
      "learning_rate": 2.6009058719150825e-08,
      "loss": 0.15636818408966063,
      "mean_token_accuracy": 0.9328071504831315,
      "num_tokens": 8632919.0,
      "step": 2240
    },
    {
      "entropy": 1.5307415947318077,
      "epoch": 4.913661202185793,
      "grad_norm": 14.530187606811523,
      "learning_rate": 1.681197050640626e-08,
      "loss": 0.3196465730667114,
      "mean_token_accuracy": 0.9131122671067715,
      "num_tokens": 8672929.0,
      "step": 2250
    },
    {
      "entropy": 1.4384046517312528,
      "epoch": 4.93551912568306,
      "grad_norm": 8.218527793884277,
      "learning_rate": 9.61228041985951e-09,
      "loss": 0.21683952808380128,
      "mean_token_accuracy": 0.9353933550417424,
      "num_tokens": 8713146.0,
      "step": 2260
    },
    {
      "entropy": 1.4315903328359127,
      "epoch": 4.9573770491803275,
      "grad_norm": 16.356935501098633,
      "learning_rate": 4.411428945998442e-09,
      "loss": 0.11141870021820069,
      "mean_token_accuracy": 0.9473440609872341,
      "num_tokens": 8743042.0,
      "step": 2270
    },
    {
      "entropy": 1.508497305959463,
      "epoch": 4.979234972677595,
      "grad_norm": 4.874918460845947,
      "learning_rate": 1.2104566513204685e-09,
      "loss": 0.281640362739563,
      "mean_token_accuracy": 0.919170456379652,
      "num_tokens": 8781700.0,
      "step": 2280
    },
    {
      "entropy": 1.4428076344101053,
      "epoch": 5.0,
      "grad_norm": 6.22572660446167,
      "learning_rate": 1.0003974135752715e-11,
      "loss": 0.1342082738876343,
      "mean_token_accuracy": 0.9626582837418506,
      "num_tokens": 8817875.0,
      "step": 2290
    },
    {
      "epoch": 5.0,
      "eval_entropy": 1.4876198212150484,
      "eval_loss": 1.570472002029419,
      "eval_mean_token_accuracy": 0.7331945998594165,
      "eval_num_tokens": 8817875.0,
      "eval_runtime": 25.2565,
      "eval_samples_per_second": 20.272,
      "eval_steps_per_second": 10.136,
      "step": 2290
    },
    {
      "epoch": 5.0,
      "step": 2290,
      "total_flos": 3.98562812212224e+17,
      "train_loss": 0.445265986773645,
      "train_runtime": 5773.6536,
      "train_samples_per_second": 3.17,
      "train_steps_per_second": 0.397
    }
  ],
  "logging_steps": 10,
  "max_steps": 2290,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.98562812212224e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}