{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9999392232734348,
  "eval_steps": 500,
  "global_step": 74040,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00040517817710338124,
      "grad_norm": 0.34768396615982056,
      "learning_rate": 4e-05,
      "loss": 1.8546,
      "step": 10
    },
    {
      "epoch": 0.0008103563542067625,
      "grad_norm": 0.5508710741996765,
      "learning_rate": 8e-05,
      "loss": 1.7537,
      "step": 20
    },
    {
      "epoch": 0.0012155345313101436,
      "grad_norm": 0.6456859707832336,
      "learning_rate": 0.00012,
      "loss": 1.6036,
      "step": 30
    },
    {
      "epoch": 0.001620712708413525,
      "grad_norm": 0.770844578742981,
      "learning_rate": 0.00016,
      "loss": 1.4788,
      "step": 40
    },
    {
      "epoch": 0.002025890885516906,
      "grad_norm": 0.6708967089653015,
      "learning_rate": 0.0002,
      "loss": 1.3791,
      "step": 50
    },
    {
      "epoch": 0.002431069062620287,
      "grad_norm": 0.6494114398956299,
      "learning_rate": 0.00019999999098587267,
      "loss": 1.3573,
      "step": 60
    },
    {
      "epoch": 0.0028362472397236685,
      "grad_norm": 0.5639660358428955,
      "learning_rate": 0.00019999996394349232,
      "loss": 1.3102,
      "step": 70
    },
    {
      "epoch": 0.00324142541682705,
      "grad_norm": 0.5362613201141357,
      "learning_rate": 0.00019999991887286375,
      "loss": 1.3205,
      "step": 80
    },
    {
      "epoch": 0.003646603593930431,
      "grad_norm": 0.5595932006835938,
      "learning_rate": 0.00019999985577399517,
      "loss": 1.3452,
      "step": 90
    },
    {
      "epoch": 0.004051781771033812,
      "grad_norm": 0.564537525177002,
      "learning_rate": 0.00019999977464689797,
      "loss": 1.3299,
      "step": 100
    },
    {
      "epoch": 0.004456959948137194,
      "grad_norm": 0.528636634349823,
      "learning_rate": 0.00019999967549158668,
      "loss": 1.3109,
      "step": 110
    },
    {
      "epoch": 0.004862138125240574,
      "grad_norm": 0.44926753640174866,
      "learning_rate": 0.00019999955830807923,
      "loss": 1.3173,
      "step": 120
    },
    {
      "epoch": 0.005267316302343956,
      "grad_norm": 0.5326265096664429,
      "learning_rate": 0.00019999942309639678,
      "loss": 1.3198,
      "step": 130
    },
    {
      "epoch": 0.005672494479447337,
      "grad_norm": 0.4632180333137512,
      "learning_rate": 0.00019999926985656366,
      "loss": 1.3377,
      "step": 140
    },
    {
      "epoch": 0.006077672656550718,
      "grad_norm": 0.5136409401893616,
      "learning_rate": 0.00019999909858860751,
      "loss": 1.3149,
      "step": 150
    },
    {
      "epoch": 0.0064828508336541,
      "grad_norm": 0.4948781728744507,
      "learning_rate": 0.00019999890929255917,
      "loss": 1.3349,
      "step": 160
    },
    {
      "epoch": 0.00688802901075748,
      "grad_norm": 0.5366935729980469,
      "learning_rate": 0.00019999870196845283,
      "loss": 1.2808,
      "step": 170
    },
    {
      "epoch": 0.007293207187860862,
      "grad_norm": 0.5862165689468384,
      "learning_rate": 0.0001999984766163258,
      "loss": 1.2879,
      "step": 180
    },
    {
      "epoch": 0.007698385364964243,
      "grad_norm": 0.5177245140075684,
      "learning_rate": 0.0001999982332362188,
      "loss": 1.2896,
      "step": 190
    },
    {
      "epoch": 0.008103563542067625,
      "grad_norm": 0.5481498837471008,
      "learning_rate": 0.00019999797182817562,
      "loss": 1.256,
      "step": 200
    },
    {
      "epoch": 0.008508741719171006,
      "grad_norm": 0.5290907025337219,
      "learning_rate": 0.00019999769239224344,
      "loss": 1.3018,
      "step": 210
    },
    {
      "epoch": 0.008913919896274387,
      "grad_norm": 0.520854115486145,
      "learning_rate": 0.0001999973949284726,
      "loss": 1.299,
      "step": 220
    },
    {
      "epoch": 0.009319098073377767,
      "grad_norm": 0.49972298741340637,
      "learning_rate": 0.00019999707943691678,
      "loss": 1.2585,
      "step": 230
    },
    {
      "epoch": 0.009724276250481148,
      "grad_norm": 0.4811578094959259,
      "learning_rate": 0.00019999674591763278,
      "loss": 1.3088,
      "step": 240
    },
    {
      "epoch": 0.01012945442758453,
      "grad_norm": 0.49944010376930237,
      "learning_rate": 0.00019999639437068075,
      "loss": 1.2788,
      "step": 250
    },
    {
      "epoch": 0.010534632604687911,
      "grad_norm": 0.6002933979034424,
      "learning_rate": 0.00019999602479612417,
      "loss": 1.2981,
      "step": 260
    },
    {
      "epoch": 0.010939810781791293,
      "grad_norm": 0.5613688826560974,
      "learning_rate": 0.0001999956371940295,
      "loss": 1.2806,
      "step": 270
    },
    {
      "epoch": 0.011344988958894674,
      "grad_norm": 0.5869052410125732,
      "learning_rate": 0.00019999523156446672,
      "loss": 1.2802,
      "step": 280
    },
    {
      "epoch": 0.011750167135998055,
      "grad_norm": 0.5404346585273743,
      "learning_rate": 0.000199994807907509,
      "loss": 1.294,
      "step": 290
    },
    {
      "epoch": 0.012155345313101437,
      "grad_norm": 0.5316123366355896,
      "learning_rate": 0.0001999943662232326,
      "loss": 1.2604,
      "step": 300
    },
    {
      "epoch": 0.012560523490204818,
      "grad_norm": 0.5258374810218811,
      "learning_rate": 0.00019999390651171723,
      "loss": 1.2688,
      "step": 310
    },
    {
      "epoch": 0.0129657016673082,
      "grad_norm": 0.5717278122901917,
      "learning_rate": 0.00019999342877304575,
      "loss": 1.291,
      "step": 320
    },
    {
      "epoch": 0.01337087984441158,
      "grad_norm": 0.5478302240371704,
      "learning_rate": 0.00019999293300730427,
      "loss": 1.2213,
      "step": 330
    },
    {
      "epoch": 0.01377605802151496,
      "grad_norm": 0.5620000958442688,
      "learning_rate": 0.0001999924192145822,
      "loss": 1.2902,
      "step": 340
    },
    {
      "epoch": 0.014181236198618342,
      "grad_norm": 0.610001266002655,
      "learning_rate": 0.00019999188739497212,
      "loss": 1.2846,
      "step": 350
    },
    {
      "epoch": 0.014586414375721724,
      "grad_norm": 0.5536356568336487,
      "learning_rate": 0.00019999133754856996,
      "loss": 1.2905,
      "step": 360
    },
    {
      "epoch": 0.014991592552825105,
      "grad_norm": 0.6234183311462402,
      "learning_rate": 0.00019999076967547482,
      "loss": 1.2878,
      "step": 370
    },
    {
      "epoch": 0.015396770729928486,
      "grad_norm": 0.4934855103492737,
      "learning_rate": 0.00019999018377578906,
      "loss": 1.2656,
      "step": 380
    },
    {
      "epoch": 0.015801948907031866,
      "grad_norm": 0.4936909079551697,
      "learning_rate": 0.00019998957984961835,
      "loss": 1.2677,
      "step": 390
    },
    {
      "epoch": 0.01620712708413525,
      "grad_norm": 0.5147895812988281,
      "learning_rate": 0.00019998895789707154,
      "loss": 1.2961,
      "step": 400
    },
    {
      "epoch": 0.01661230526123863,
      "grad_norm": 0.5227635502815247,
      "learning_rate": 0.0001999883179182608,
      "loss": 1.2884,
      "step": 410
    },
    {
      "epoch": 0.017017483438342012,
      "grad_norm": 0.5173860788345337,
      "learning_rate": 0.00019998765991330143,
      "loss": 1.2473,
      "step": 420
    },
    {
      "epoch": 0.01742266161544539,
      "grad_norm": 0.5594643950462341,
      "learning_rate": 0.0001999869838823121,
      "loss": 1.2397,
      "step": 430
    },
    {
      "epoch": 0.017827839792548775,
      "grad_norm": 0.5018565058708191,
      "learning_rate": 0.0001999862898254147,
      "loss": 1.2541,
      "step": 440
    },
    {
      "epoch": 0.018233017969652154,
      "grad_norm": 0.5900639891624451,
      "learning_rate": 0.0001999855777427343,
      "loss": 1.2513,
      "step": 450
    },
    {
      "epoch": 0.018638196146755534,
      "grad_norm": 0.5088098049163818,
      "learning_rate": 0.00019998484763439939,
      "loss": 1.2595,
      "step": 460
    },
    {
      "epoch": 0.019043374323858917,
      "grad_norm": 0.5259648561477661,
      "learning_rate": 0.00019998409950054146,
      "loss": 1.2788,
      "step": 470
    },
    {
      "epoch": 0.019448552500962297,
      "grad_norm": 0.5566315054893494,
      "learning_rate": 0.0001999833333412955,
      "loss": 1.2459,
      "step": 480
    },
    {
      "epoch": 0.01985373067806568,
      "grad_norm": 0.5249390006065369,
      "learning_rate": 0.0001999825491567995,
      "loss": 1.2741,
      "step": 490
    },
    {
      "epoch": 0.02025890885516906,
      "grad_norm": 0.627704381942749,
      "learning_rate": 0.000199981746947195,
      "loss": 1.2349,
      "step": 500
    },
    {
      "epoch": 0.020664087032272443,
      "grad_norm": 0.436422199010849,
      "learning_rate": 0.00019998092671262652,
      "loss": 1.2151,
      "step": 510
    },
    {
      "epoch": 0.021069265209375822,
      "grad_norm": 0.5417822599411011,
      "learning_rate": 0.00019998008845324193,
      "loss": 1.264,
      "step": 520
    },
    {
      "epoch": 0.021474443386479206,
      "grad_norm": 0.44335830211639404,
      "learning_rate": 0.0001999792321691924,
      "loss": 1.2445,
      "step": 530
    },
    {
      "epoch": 0.021879621563582585,
      "grad_norm": 0.5444326400756836,
      "learning_rate": 0.0001999783578606323,
      "loss": 1.2655,
      "step": 540
    },
    {
      "epoch": 0.022284799740685965,
      "grad_norm": 0.584891140460968,
      "learning_rate": 0.0001999774655277192,
      "loss": 1.3084,
      "step": 550
    },
    {
      "epoch": 0.022689977917789348,
      "grad_norm": 0.5807209610939026,
      "learning_rate": 0.00019997655517061405,
      "loss": 1.2662,
      "step": 560
    },
    {
      "epoch": 0.023095156094892728,
      "grad_norm": 0.5752170085906982,
      "learning_rate": 0.0001999756267894809,
      "loss": 1.2626,
      "step": 570
    },
    {
      "epoch": 0.02350033427199611,
      "grad_norm": 0.5122477412223816,
      "learning_rate": 0.00019997468038448717,
      "loss": 1.2276,
      "step": 580
    },
    {
      "epoch": 0.02390551244909949,
      "grad_norm": 0.5006405711174011,
      "learning_rate": 0.00019997371595580345,
      "loss": 1.2512,
      "step": 590
    },
    {
      "epoch": 0.024310690626202874,
      "grad_norm": 0.6164976358413696,
      "learning_rate": 0.0001999727335036036,
      "loss": 1.2319,
      "step": 600
    },
    {
      "epoch": 0.024715868803306253,
      "grad_norm": 0.59392911195755,
      "learning_rate": 0.00019997173302806478,
      "loss": 1.2634,
      "step": 610
    },
    {
      "epoch": 0.025121046980409636,
      "grad_norm": 0.5711776614189148,
      "learning_rate": 0.00019997071452936734,
      "loss": 1.2677,
      "step": 620
    },
    {
      "epoch": 0.025526225157513016,
      "grad_norm": 0.5695605278015137,
      "learning_rate": 0.0001999696780076949,
      "loss": 1.2343,
      "step": 630
    },
    {
      "epoch": 0.0259314033346164,
      "grad_norm": 0.5710616111755371,
      "learning_rate": 0.0001999686234632343,
      "loss": 1.2126,
      "step": 640
    },
    {
      "epoch": 0.02633658151171978,
      "grad_norm": 0.6264880299568176,
      "learning_rate": 0.00019996755089617566,
      "loss": 1.2481,
      "step": 650
    },
    {
      "epoch": 0.02674175968882316,
      "grad_norm": 0.5485592484474182,
      "learning_rate": 0.00019996646030671239,
      "loss": 1.2136,
      "step": 660
    },
    {
      "epoch": 0.02714693786592654,
      "grad_norm": 0.585409939289093,
      "learning_rate": 0.0001999653516950411,
      "loss": 1.2716,
      "step": 670
    },
    {
      "epoch": 0.02755211604302992,
      "grad_norm": 0.5780940651893616,
      "learning_rate": 0.0001999642250613616,
      "loss": 1.2476,
      "step": 680
    },
    {
      "epoch": 0.027957294220133305,
      "grad_norm": 0.5648075938224792,
      "learning_rate": 0.00019996308040587705,
      "loss": 1.2051,
      "step": 690
    },
    {
      "epoch": 0.028362472397236684,
      "grad_norm": 0.48909419775009155,
      "learning_rate": 0.00019996191772879378,
      "loss": 1.2248,
      "step": 700
    },
    {
      "epoch": 0.028767650574340067,
      "grad_norm": 0.557926595211029,
      "learning_rate": 0.0001999607370303214,
      "loss": 1.2351,
      "step": 710
    },
    {
      "epoch": 0.029172828751443447,
      "grad_norm": 0.5209690928459167,
      "learning_rate": 0.0001999595383106728,
      "loss": 1.2299,
      "step": 720
    },
    {
      "epoch": 0.02957800692854683,
      "grad_norm": 0.5277084112167358,
      "learning_rate": 0.0001999583215700641,
      "loss": 1.2848,
      "step": 730
    },
    {
      "epoch": 0.02998318510565021,
      "grad_norm": 0.6520649790763855,
      "learning_rate": 0.00019995708680871458,
      "loss": 1.2722,
      "step": 740
    },
    {
      "epoch": 0.03038836328275359,
      "grad_norm": 0.5660078525543213,
      "learning_rate": 0.00019995583402684694,
      "loss": 1.2211,
      "step": 750
    },
    {
      "epoch": 0.030793541459856973,
      "grad_norm": 0.4896627962589264,
      "learning_rate": 0.00019995456322468696,
      "loss": 1.223,
      "step": 760
    },
    {
      "epoch": 0.031198719636960352,
      "grad_norm": 0.5370994806289673,
      "learning_rate": 0.00019995327440246378,
      "loss": 1.2207,
      "step": 770
    },
    {
      "epoch": 0.03160389781406373,
      "grad_norm": 0.5011025667190552,
      "learning_rate": 0.00019995196756040975,
      "loss": 1.2538,
      "step": 780
    },
    {
      "epoch": 0.03200907599116712,
      "grad_norm": 0.5034549832344055,
      "learning_rate": 0.00019995064269876046,
      "loss": 1.2498,
      "step": 790
    },
    {
      "epoch": 0.0324142541682705,
      "grad_norm": 0.5401809811592102,
      "learning_rate": 0.00019994929981775476,
      "loss": 1.2345,
      "step": 800
    },
    {
      "epoch": 0.03281943234537388,
      "grad_norm": 0.47775983810424805,
      "learning_rate": 0.00019994793891763476,
      "loss": 1.2393,
      "step": 810
    },
    {
      "epoch": 0.03322461052247726,
      "grad_norm": 0.5061116218566895,
      "learning_rate": 0.00019994655999864582,
      "loss": 1.2287,
      "step": 820
    },
    {
      "epoch": 0.03362978869958064,
      "grad_norm": 0.5282241702079773,
      "learning_rate": 0.0001999451630610365,
      "loss": 1.281,
      "step": 830
    },
    {
      "epoch": 0.034034966876684024,
      "grad_norm": 0.4783414900302887,
      "learning_rate": 0.00019994374810505865,
      "loss": 1.2484,
      "step": 840
    },
    {
      "epoch": 0.034440145053787403,
      "grad_norm": 0.5489025115966797,
      "learning_rate": 0.00019994231513096735,
      "loss": 1.225,
      "step": 850
    },
    {
      "epoch": 0.03484532323089078,
      "grad_norm": 0.5059438347816467,
      "learning_rate": 0.000199940864139021,
      "loss": 1.2412,
      "step": 860
    },
    {
      "epoch": 0.03525050140799416,
      "grad_norm": 0.5379940867424011,
      "learning_rate": 0.00019993939512948113,
      "loss": 1.2393,
      "step": 870
    },
    {
      "epoch": 0.03565567958509755,
      "grad_norm": 0.5828728675842285,
      "learning_rate": 0.0001999379081026126,
      "loss": 1.2429,
      "step": 880
    },
    {
      "epoch": 0.03606085776220093,
      "grad_norm": 0.5324397087097168,
      "learning_rate": 0.00019993640305868352,
      "loss": 1.225,
      "step": 890
    },
    {
      "epoch": 0.03646603593930431,
      "grad_norm": 0.5943859219551086,
      "learning_rate": 0.00019993487999796515,
      "loss": 1.2352,
      "step": 900
    },
    {
      "epoch": 0.03687121411640769,
      "grad_norm": 0.4988947808742523,
      "learning_rate": 0.00019993333892073212,
      "loss": 1.198,
      "step": 910
    },
    {
      "epoch": 0.03727639229351107,
      "grad_norm": 0.5493001341819763,
      "learning_rate": 0.0001999317798272623,
      "loss": 1.2456,
      "step": 920
    },
    {
      "epoch": 0.037681570470614455,
      "grad_norm": 0.5663845539093018,
      "learning_rate": 0.00019993020271783668,
      "loss": 1.2082,
      "step": 930
    },
    {
      "epoch": 0.038086748647717834,
      "grad_norm": 0.5184051394462585,
      "learning_rate": 0.00019992860759273962,
      "loss": 1.2088,
      "step": 940
    },
    {
      "epoch": 0.038491926824821214,
      "grad_norm": 0.517812967300415,
      "learning_rate": 0.00019992699445225872,
      "loss": 1.2464,
      "step": 950
    },
    {
      "epoch": 0.038897105001924594,
      "grad_norm": 0.5713759064674377,
      "learning_rate": 0.00019992536329668478,
      "loss": 1.2191,
      "step": 960
    },
    {
      "epoch": 0.03930228317902798,
      "grad_norm": 0.5452908277511597,
      "learning_rate": 0.00019992371412631187,
      "loss": 1.2416,
      "step": 970
    },
    {
      "epoch": 0.03970746135613136,
      "grad_norm": 0.5917738676071167,
      "learning_rate": 0.0001999220469414373,
      "loss": 1.2057,
      "step": 980
    },
    {
      "epoch": 0.04011263953323474,
      "grad_norm": 0.5491656064987183,
      "learning_rate": 0.00019992036174236165,
      "loss": 1.213,
      "step": 990
    },
    {
      "epoch": 0.04051781771033812,
      "grad_norm": 0.5373522043228149,
      "learning_rate": 0.00019991865852938873,
      "loss": 1.2274,
      "step": 1000
    },
    {
      "epoch": 0.0409229958874415,
      "grad_norm": 0.5253644585609436,
      "learning_rate": 0.00019991693730282557,
      "loss": 1.2321,
      "step": 1010
    },
    {
      "epoch": 0.041328174064544886,
      "grad_norm": 0.5066225528717041,
      "learning_rate": 0.00019991519806298255,
      "loss": 1.2572,
      "step": 1020
    },
    {
      "epoch": 0.041733352241648265,
      "grad_norm": 0.5360060334205627,
      "learning_rate": 0.0001999134408101731,
      "loss": 1.2041,
      "step": 1030
    },
    {
      "epoch": 0.042138530418751645,
      "grad_norm": 0.6068186163902283,
      "learning_rate": 0.00019991166554471414,
      "loss": 1.223,
      "step": 1040
    },
    {
      "epoch": 0.042543708595855025,
      "grad_norm": 0.508084774017334,
      "learning_rate": 0.00019990987226692567,
      "loss": 1.2274,
      "step": 1050
    },
    {
      "epoch": 0.04294888677295841,
      "grad_norm": 0.4908534288406372,
      "learning_rate": 0.00019990806097713103,
      "loss": 1.2573,
      "step": 1060
    },
    {
      "epoch": 0.04335406495006179,
      "grad_norm": 0.5907722115516663,
      "learning_rate": 0.00019990623167565668,
      "loss": 1.2234,
      "step": 1070
    },
    {
      "epoch": 0.04375924312716517,
      "grad_norm": 0.613249659538269,
      "learning_rate": 0.00019990438436283246,
      "loss": 1.2534,
      "step": 1080
    },
    {
      "epoch": 0.04416442130426855,
      "grad_norm": 0.5669634938240051,
      "learning_rate": 0.00019990251903899143,
      "loss": 1.2148,
      "step": 1090
    },
    {
      "epoch": 0.04456959948137193,
      "grad_norm": 0.5174763202667236,
      "learning_rate": 0.00019990063570446984,
      "loss": 1.1888,
      "step": 1100
    },
    {
      "epoch": 0.044974777658475316,
      "grad_norm": 0.5306107401847839,
      "learning_rate": 0.00019989873435960723,
      "loss": 1.2201,
      "step": 1110
    },
    {
      "epoch": 0.045379955835578696,
      "grad_norm": 0.5860251188278198,
      "learning_rate": 0.00019989681500474636,
      "loss": 1.2564,
      "step": 1120
    },
    {
      "epoch": 0.045785134012682076,
      "grad_norm": 0.5726203918457031,
      "learning_rate": 0.00019989487764023332,
      "loss": 1.2671,
      "step": 1130
    },
    {
      "epoch": 0.046190312189785455,
      "grad_norm": 0.5739380121231079,
      "learning_rate": 0.00019989292226641733,
      "loss": 1.2316,
      "step": 1140
    },
    {
      "epoch": 0.04659549036688884,
      "grad_norm": 0.5087692737579346,
      "learning_rate": 0.0001998909488836509,
      "loss": 1.2459,
      "step": 1150
    },
    {
      "epoch": 0.04700066854399222,
      "grad_norm": 0.6039331555366516,
      "learning_rate": 0.00019988895749228984,
      "loss": 1.2213,
      "step": 1160
    },
    {
      "epoch": 0.0474058467210956,
      "grad_norm": 0.460628479719162,
      "learning_rate": 0.00019988694809269314,
      "loss": 1.1985,
      "step": 1170
    },
    {
      "epoch": 0.04781102489819898,
      "grad_norm": 0.5117699503898621,
      "learning_rate": 0.00019988492068522307,
      "loss": 1.195,
      "step": 1180
    },
    {
      "epoch": 0.04821620307530237,
      "grad_norm": 0.5307199954986572,
      "learning_rate": 0.0001998828752702451,
      "loss": 1.235,
      "step": 1190
    },
    {
      "epoch": 0.04862138125240575,
      "grad_norm": 0.6341548562049866,
      "learning_rate": 0.00019988081184812805,
      "loss": 1.2263,
      "step": 1200
    },
    {
      "epoch": 0.04902655942950913,
      "grad_norm": 0.5309761166572571,
      "learning_rate": 0.00019987873041924385,
      "loss": 1.2343,
      "step": 1210
    },
    {
      "epoch": 0.04943173760661251,
      "grad_norm": 0.5552977323532104,
      "learning_rate": 0.00019987663098396778,
      "loss": 1.2682,
      "step": 1220
    },
    {
      "epoch": 0.049836915783715886,
      "grad_norm": 0.5447342395782471,
      "learning_rate": 0.00019987451354267833,
      "loss": 1.1941,
      "step": 1230
    },
    {
      "epoch": 0.05024209396081927,
      "grad_norm": 0.5575499534606934,
      "learning_rate": 0.00019987237809575723,
      "loss": 1.2198,
      "step": 1240
    },
    {
      "epoch": 0.05064727213792265,
      "grad_norm": 0.5578951835632324,
      "learning_rate": 0.00019987022464358945,
      "loss": 1.1973,
      "step": 1250
    },
    {
      "epoch": 0.05105245031502603,
      "grad_norm": 0.5636481046676636,
      "learning_rate": 0.00019986805318656327,
      "loss": 1.2708,
      "step": 1260
    },
    {
      "epoch": 0.05145762849212941,
      "grad_norm": 0.5856820344924927,
      "learning_rate": 0.00019986586372507012,
      "loss": 1.1949,
      "step": 1270
    },
    {
      "epoch": 0.0518628066692328,
      "grad_norm": 0.4917871654033661,
      "learning_rate": 0.00019986365625950474,
      "loss": 1.2763,
      "step": 1280
    },
    {
      "epoch": 0.05226798484633618,
      "grad_norm": 0.5795226097106934,
      "learning_rate": 0.0001998614307902651,
      "loss": 1.2211,
      "step": 1290
    },
    {
      "epoch": 0.05267316302343956,
      "grad_norm": 0.5203938484191895,
      "learning_rate": 0.00019985918731775236,
      "loss": 1.1904,
      "step": 1300
    },
    {
      "epoch": 0.05307834120054294,
      "grad_norm": 0.5937249064445496,
      "learning_rate": 0.00019985692584237108,
      "loss": 1.2418,
      "step": 1310
    },
    {
      "epoch": 0.05348351937764632,
      "grad_norm": 0.6009014844894409,
      "learning_rate": 0.00019985464636452886,
      "loss": 1.2341,
      "step": 1320
    },
    {
      "epoch": 0.053888697554749704,
      "grad_norm": 0.5742254257202148,
      "learning_rate": 0.00019985234888463674,
      "loss": 1.2145,
      "step": 1330
    },
    {
      "epoch": 0.05429387573185308,
      "grad_norm": 0.5537120699882507,
      "learning_rate": 0.00019985003340310887,
      "loss": 1.1745,
      "step": 1340
    },
    {
      "epoch": 0.05469905390895646,
      "grad_norm": 0.5304623246192932,
      "learning_rate": 0.00019984769992036267,
      "loss": 1.193,
      "step": 1350
    },
    {
      "epoch": 0.05510423208605984,
      "grad_norm": 0.48964646458625793,
      "learning_rate": 0.00019984534843681888,
      "loss": 1.2083,
      "step": 1360
    },
    {
      "epoch": 0.05550941026316323,
      "grad_norm": 0.5262625217437744,
      "learning_rate": 0.00019984297895290137,
      "loss": 1.2509,
      "step": 1370
    },
    {
      "epoch": 0.05591458844026661,
      "grad_norm": 0.5625774264335632,
      "learning_rate": 0.0001998405914690374,
      "loss": 1.1764,
      "step": 1380
    },
    {
      "epoch": 0.05631976661736999,
      "grad_norm": 0.5014253258705139,
      "learning_rate": 0.0001998381859856573,
      "loss": 1.2134,
      "step": 1390
    },
    {
      "epoch": 0.05672494479447337,
      "grad_norm": 0.5085203647613525,
      "learning_rate": 0.0001998357625031948,
      "loss": 1.2327,
      "step": 1400
    },
    {
      "epoch": 0.05713012297157675,
      "grad_norm": 0.4464736878871918,
      "learning_rate": 0.00019983332102208677,
      "loss": 1.2111,
      "step": 1410
    },
    {
      "epoch": 0.057535301148680135,
      "grad_norm": 0.5147503018379211,
      "learning_rate": 0.0001998308615427734,
      "loss": 1.2404,
      "step": 1420
    },
    {
      "epoch": 0.057940479325783514,
      "grad_norm": 0.6577001214027405,
      "learning_rate": 0.0001998283840656981,
      "loss": 1.2118,
      "step": 1430
    },
    {
      "epoch": 0.058345657502886894,
      "grad_norm": 0.538983941078186,
      "learning_rate": 0.00019982588859130743,
      "loss": 1.1873,
      "step": 1440
    },
    {
      "epoch": 0.058750835679990274,
      "grad_norm": 0.5143253207206726,
      "learning_rate": 0.00019982337512005138,
      "loss": 1.237,
      "step": 1450
    },
    {
      "epoch": 0.05915601385709366,
      "grad_norm": 0.5343175530433655,
      "learning_rate": 0.00019982084365238306,
      "loss": 1.2029,
      "step": 1460
    },
    {
      "epoch": 0.05956119203419704,
      "grad_norm": 0.5228641033172607,
      "learning_rate": 0.00019981829418875885,
      "loss": 1.1894,
      "step": 1470
    },
    {
      "epoch": 0.05996637021130042,
      "grad_norm": 0.496764600276947,
      "learning_rate": 0.0001998157267296383,
      "loss": 1.2292,
      "step": 1480
    },
    {
      "epoch": 0.0603715483884038,
      "grad_norm": 0.5016512870788574,
      "learning_rate": 0.0001998131412754844,
      "loss": 1.194,
      "step": 1490
    },
    {
      "epoch": 0.06077672656550718,
      "grad_norm": 0.5040959119796753,
      "learning_rate": 0.00019981053782676319,
      "loss": 1.2095,
      "step": 1500
    },
    {
      "epoch": 0.061181904742610566,
      "grad_norm": 0.5558595657348633,
      "learning_rate": 0.000199807916383944,
      "loss": 1.2301,
      "step": 1510
    },
    {
      "epoch": 0.061587082919713945,
      "grad_norm": 0.578183114528656,
      "learning_rate": 0.00019980527694749952,
      "loss": 1.2155,
      "step": 1520
    },
    {
      "epoch": 0.061992261096817325,
      "grad_norm": 0.5872295498847961,
      "learning_rate": 0.00019980261951790552,
      "loss": 1.1917,
      "step": 1530
    },
    {
      "epoch": 0.062397439273920705,
      "grad_norm": 0.5450662970542908,
      "learning_rate": 0.00019979994409564112,
      "loss": 1.2443,
      "step": 1540
    },
    {
      "epoch": 0.06280261745102408,
      "grad_norm": 0.486433207988739,
      "learning_rate": 0.00019979725068118863,
      "loss": 1.2134,
      "step": 1550
    },
    {
      "epoch": 0.06320779562812746,
      "grad_norm": 0.5684202313423157,
      "learning_rate": 0.00019979453927503364,
      "loss": 1.2063,
      "step": 1560
    },
    {
      "epoch": 0.06361297380523084,
      "grad_norm": 0.5183430314064026,
      "learning_rate": 0.00019979180987766498,
      "loss": 1.2189,
      "step": 1570
    },
    {
      "epoch": 0.06401815198233424,
      "grad_norm": 0.5423526167869568,
      "learning_rate": 0.0001997890624895747,
      "loss": 1.2166,
      "step": 1580
    },
    {
      "epoch": 0.06442333015943762,
      "grad_norm": 0.5009813904762268,
      "learning_rate": 0.00019978629711125812,
      "loss": 1.2077,
      "step": 1590
    },
    {
      "epoch": 0.064828508336541,
      "grad_norm": 0.515937089920044,
      "learning_rate": 0.00019978351374321372,
      "loss": 1.2204,
      "step": 1600
    },
    {
      "epoch": 0.06523368651364438,
      "grad_norm": 0.5453330874443054,
      "learning_rate": 0.00019978071238594344,
      "loss": 1.2476,
      "step": 1610
    },
    {
      "epoch": 0.06563886469074776,
      "grad_norm": 0.5542542338371277,
      "learning_rate": 0.00019977789303995212,
      "loss": 1.1981,
      "step": 1620
    },
    {
      "epoch": 0.06604404286785114,
      "grad_norm": 0.5402368307113647,
      "learning_rate": 0.00019977505570574825,
      "loss": 1.2163,
      "step": 1630
    },
    {
      "epoch": 0.06644922104495452,
      "grad_norm": 0.5312523245811462,
      "learning_rate": 0.00019977220038384317,
      "loss": 1.2119,
      "step": 1640
    },
    {
      "epoch": 0.0668543992220579,
      "grad_norm": 0.5652551651000977,
      "learning_rate": 0.00019976932707475173,
      "loss": 1.2009,
      "step": 1650
    },
    {
      "epoch": 0.06725957739916127,
      "grad_norm": 0.5763120055198669,
      "learning_rate": 0.00019976643577899195,
      "loss": 1.207,
      "step": 1660
    },
    {
      "epoch": 0.06766475557626467,
      "grad_norm": 0.5200467109680176,
      "learning_rate": 0.00019976352649708504,
      "loss": 1.1778,
      "step": 1670
    },
    {
      "epoch": 0.06806993375336805,
      "grad_norm": 0.5550898909568787,
      "learning_rate": 0.00019976059922955551,
      "loss": 1.1782,
      "step": 1680
    },
    {
      "epoch": 0.06847511193047143,
      "grad_norm": 0.4978763461112976,
      "learning_rate": 0.00019975765397693113,
      "loss": 1.2398,
      "step": 1690
    },
    {
      "epoch": 0.06888029010757481,
      "grad_norm": 0.5891223549842834,
      "learning_rate": 0.00019975469073974281,
      "loss": 1.2311,
      "step": 1700
    },
    {
      "epoch": 0.06928546828467819,
      "grad_norm": 0.5025737285614014,
      "learning_rate": 0.00019975170951852484,
      "loss": 1.2006,
      "step": 1710
    },
    {
      "epoch": 0.06969064646178157,
      "grad_norm": 0.5287303328514099,
      "learning_rate": 0.0001997487103138146,
      "loss": 1.1924,
      "step": 1720
    },
    {
      "epoch": 0.07009582463888495,
      "grad_norm": 0.5320292115211487,
      "learning_rate": 0.0001997456931261529,
      "loss": 1.2182,
      "step": 1730
    },
    {
      "epoch": 0.07050100281598833,
      "grad_norm": 0.551361083984375,
      "learning_rate": 0.0001997426579560836,
      "loss": 1.2116,
      "step": 1740
    },
    {
      "epoch": 0.0709061809930917,
      "grad_norm": 0.5364453792572021,
      "learning_rate": 0.00019973960480415385,
      "loss": 1.2497,
      "step": 1750
    },
    {
      "epoch": 0.0713113591701951,
      "grad_norm": 0.5189634561538696,
      "learning_rate": 0.00019973653367091424,
      "loss": 1.2039,
      "step": 1760
    },
    {
      "epoch": 0.07171653734729848,
      "grad_norm": 0.6426394581794739,
      "learning_rate": 0.0001997334445569183,
      "loss": 1.2272,
      "step": 1770
    },
    {
      "epoch": 0.07212171552440186,
      "grad_norm": 0.5143496990203857,
      "learning_rate": 0.00019973033746272298,
      "loss": 1.215,
      "step": 1780
    },
    {
      "epoch": 0.07252689370150524,
      "grad_norm": 0.487033873796463,
      "learning_rate": 0.00019972721238888845,
      "loss": 1.2189,
      "step": 1790
    },
    {
      "epoch": 0.07293207187860862,
      "grad_norm": 0.5420389771461487,
      "learning_rate": 0.00019972406933597812,
      "loss": 1.2203,
      "step": 1800
    },
    {
      "epoch": 0.073337250055712,
      "grad_norm": 0.5032690167427063,
      "learning_rate": 0.0001997209083045586,
      "loss": 1.2256,
      "step": 1810
    },
    {
      "epoch": 0.07374242823281538,
      "grad_norm": 0.5467582941055298,
      "learning_rate": 0.00019971772929519975,
      "loss": 1.216,
      "step": 1820
    },
    {
      "epoch": 0.07414760640991876,
      "grad_norm": 0.46452221274375916,
      "learning_rate": 0.00019971453230847476,
      "loss": 1.1647,
      "step": 1830
    },
    {
      "epoch": 0.07455278458702214,
      "grad_norm": 0.5254132151603699,
      "learning_rate": 0.00019971131734495995,
      "loss": 1.2055,
      "step": 1840
    },
    {
      "epoch": 0.07495796276412553,
      "grad_norm": 0.5780434608459473,
      "learning_rate": 0.00019970808440523488,
      "loss": 1.2282,
      "step": 1850
    },
    {
      "epoch": 0.07536314094122891,
      "grad_norm": 0.5534215569496155,
      "learning_rate": 0.00019970483348988246,
      "loss": 1.2154,
      "step": 1860
    },
    {
      "epoch": 0.07576831911833229,
      "grad_norm": 0.5256699919700623,
      "learning_rate": 0.00019970156459948873,
      "loss": 1.2299,
      "step": 1870
    },
    {
      "epoch": 0.07617349729543567,
      "grad_norm": 0.6103618144989014,
      "learning_rate": 0.00019969827773464303,
      "loss": 1.2403,
      "step": 1880
    },
    {
      "epoch": 0.07657867547253905,
      "grad_norm": 0.5466075539588928,
      "learning_rate": 0.00019969497289593795,
      "loss": 1.2088,
      "step": 1890
    },
    {
      "epoch": 0.07698385364964243,
      "grad_norm": 0.5868998765945435,
      "learning_rate": 0.00019969165008396924,
      "loss": 1.1897,
      "step": 1900
    },
    {
      "epoch": 0.07738903182674581,
      "grad_norm": 0.5408425331115723,
      "learning_rate": 0.00019968830929933598,
      "loss": 1.2063,
      "step": 1910
    },
    {
      "epoch": 0.07779421000384919,
      "grad_norm": 0.5227720737457275,
      "learning_rate": 0.00019968495054264044,
      "loss": 1.2008,
      "step": 1920
    },
    {
      "epoch": 0.07819938818095257,
      "grad_norm": 0.5726022124290466,
      "learning_rate": 0.0001996815738144882,
      "loss": 1.2422,
      "step": 1930
    },
    {
      "epoch": 0.07860456635805596,
      "grad_norm": 0.6630988717079163,
      "learning_rate": 0.00019967817911548794,
      "loss": 1.1306,
      "step": 1940
    },
    {
      "epoch": 0.07900974453515934,
      "grad_norm": 0.5714316368103027,
      "learning_rate": 0.0001996747664462517,
      "loss": 1.2067,
      "step": 1950
    },
    {
      "epoch": 0.07941492271226272,
      "grad_norm": 0.5627208352088928,
      "learning_rate": 0.00019967133580739476,
      "loss": 1.1787,
      "step": 1960
    },
    {
      "epoch": 0.0798201008893661,
      "grad_norm": 0.5557634234428406,
      "learning_rate": 0.00019966788719953554,
      "loss": 1.2047,
      "step": 1970
    },
    {
      "epoch": 0.08022527906646948,
      "grad_norm": 0.5416475534439087,
      "learning_rate": 0.0001996644206232958,
      "loss": 1.2051,
      "step": 1980
    },
    {
      "epoch": 0.08063045724357286,
      "grad_norm": 0.5145570039749146,
      "learning_rate": 0.00019966093607930052,
      "loss": 1.2108,
      "step": 1990
    },
    {
      "epoch": 0.08103563542067624,
      "grad_norm": 0.5667361617088318,
      "learning_rate": 0.00019965743356817786,
      "loss": 1.2391,
      "step": 2000
    },
    {
      "epoch": 0.08144081359777962,
      "grad_norm": 0.6069298386573792,
      "learning_rate": 0.0001996539130905593,
      "loss": 1.212,
      "step": 2010
    },
    {
      "epoch": 0.081845991774883,
      "grad_norm": 0.5689290165901184,
      "learning_rate": 0.00019965037464707953,
      "loss": 1.2325,
      "step": 2020
    },
    {
      "epoch": 0.08225116995198639,
      "grad_norm": 0.49058398604393005,
      "learning_rate": 0.00019964681823837638,
      "loss": 1.2123,
      "step": 2030
    },
    {
      "epoch": 0.08265634812908977,
      "grad_norm": 0.5370051264762878,
      "learning_rate": 0.00019964324386509112,
      "loss": 1.1688,
      "step": 2040
    },
    {
      "epoch": 0.08306152630619315,
      "grad_norm": 0.5877329707145691,
      "learning_rate": 0.0001996396515278681,
      "loss": 1.2287,
      "step": 2050
    },
    {
      "epoch": 0.08346670448329653,
      "grad_norm": 0.5659486651420593,
      "learning_rate": 0.00019963604122735494,
      "loss": 1.1845,
      "step": 2060
    },
    {
      "epoch": 0.08387188266039991,
      "grad_norm": 0.6263470649719238,
      "learning_rate": 0.0001996324129642025,
      "loss": 1.2062,
      "step": 2070
    },
    {
      "epoch": 0.08427706083750329,
      "grad_norm": 0.582596480846405,
      "learning_rate": 0.000199628766739065,
      "loss": 1.2251,
      "step": 2080
    },
    {
      "epoch": 0.08468223901460667,
      "grad_norm": 0.5233246088027954,
      "learning_rate": 0.00019962510255259965,
      "loss": 1.1975,
      "step": 2090
    },
    {
      "epoch": 0.08508741719171005,
      "grad_norm": 0.5275470018386841,
      "learning_rate": 0.00019962142040546714,
      "loss": 1.2223,
      "step": 2100
    },
    {
      "epoch": 0.08549259536881343,
      "grad_norm": 0.49118420481681824,
      "learning_rate": 0.00019961772029833123,
      "loss": 1.1785,
      "step": 2110
    },
    {
      "epoch": 0.08589777354591682,
      "grad_norm": 0.5409901738166809,
      "learning_rate": 0.00019961400223185902,
      "loss": 1.2035,
      "step": 2120
    },
    {
      "epoch": 0.0863029517230202,
      "grad_norm": 0.5661318302154541,
      "learning_rate": 0.00019961026620672083,
      "loss": 1.2021,
      "step": 2130
    },
    {
      "epoch": 0.08670812990012358,
      "grad_norm": 0.60009765625,
      "learning_rate": 0.00019960651222359016,
      "loss": 1.165,
      "step": 2140
    },
    {
      "epoch": 0.08711330807722696,
      "grad_norm": 0.5232530832290649,
      "learning_rate": 0.0001996027402831438,
      "loss": 1.1934,
      "step": 2150
    },
    {
      "epoch": 0.08751848625433034,
      "grad_norm": 0.6539116501808167,
      "learning_rate": 0.00019959895038606176,
      "loss": 1.2244,
      "step": 2160
    },
    {
      "epoch": 0.08792366443143372,
      "grad_norm": 0.5635285973548889,
      "learning_rate": 0.00019959514253302732,
      "loss": 1.1921,
      "step": 2170
    },
    {
      "epoch": 0.0883288426085371,
      "grad_norm": 0.574511706829071,
      "learning_rate": 0.00019959131672472695,
      "loss": 1.1908,
      "step": 2180
    },
    {
      "epoch": 0.08873402078564048,
      "grad_norm": 0.5598782300949097,
      "learning_rate": 0.00019958747296185036,
      "loss": 1.1574,
      "step": 2190
    },
    {
      "epoch": 0.08913919896274386,
      "grad_norm": 0.4832444190979004,
      "learning_rate": 0.00019958361124509054,
      "loss": 1.1913,
      "step": 2200
    },
    {
      "epoch": 0.08954437713984725,
      "grad_norm": 0.4934306740760803,
      "learning_rate": 0.0001995797315751437,
      "loss": 1.197,
      "step": 2210
    },
    {
      "epoch": 0.08994955531695063,
      "grad_norm": 0.522014856338501,
      "learning_rate": 0.00019957583395270923,
      "loss": 1.2207,
      "step": 2220
    },
    {
      "epoch": 0.09035473349405401,
      "grad_norm": 0.5646122097969055,
      "learning_rate": 0.00019957191837848985,
      "loss": 1.198,
      "step": 2230
    },
    {
      "epoch": 0.09075991167115739,
      "grad_norm": 0.5438932180404663,
      "learning_rate": 0.00019956798485319143,
      "loss": 1.1853,
      "step": 2240
    },
    {
      "epoch": 0.09116508984826077,
      "grad_norm": 0.5806702971458435,
      "learning_rate": 0.00019956403337752316,
      "loss": 1.2051,
      "step": 2250
    },
    {
      "epoch": 0.09157026802536415,
      "grad_norm": 0.4922071099281311,
      "learning_rate": 0.00019956006395219737,
      "loss": 1.1915,
      "step": 2260
    },
    {
      "epoch": 0.09197544620246753,
      "grad_norm": 0.5436669588088989,
      "learning_rate": 0.00019955607657792975,
      "loss": 1.2137,
      "step": 2270
    },
    {
      "epoch": 0.09238062437957091,
      "grad_norm": 0.5298568606376648,
      "learning_rate": 0.00019955207125543908,
      "loss": 1.1859,
      "step": 2280
    },
    {
      "epoch": 0.0927858025566743,
      "grad_norm": 0.5735529661178589,
      "learning_rate": 0.00019954804798544745,
      "loss": 1.181,
      "step": 2290
    },
    {
      "epoch": 0.09319098073377768,
      "grad_norm": 0.526323676109314,
      "learning_rate": 0.00019954400676868027,
      "loss": 1.1524,
      "step": 2300
    },
    {
      "epoch": 0.09359615891088106,
      "grad_norm": 0.5396515727043152,
      "learning_rate": 0.00019953994760586602,
      "loss": 1.1858,
      "step": 2310
    },
    {
      "epoch": 0.09400133708798444,
      "grad_norm": 0.5272504091262817,
      "learning_rate": 0.00019953587049773654,
      "loss": 1.2002,
      "step": 2320
    },
    {
      "epoch": 0.09440651526508782,
      "grad_norm": 0.5551445484161377,
      "learning_rate": 0.00019953177544502685,
      "loss": 1.2269,
      "step": 2330
    },
    {
      "epoch": 0.0948116934421912,
      "grad_norm": 0.5649201273918152,
      "learning_rate": 0.0001995276624484752,
      "loss": 1.169,
      "step": 2340
    },
    {
      "epoch": 0.09521687161929458,
      "grad_norm": 0.6032335162162781,
      "learning_rate": 0.00019952353150882307,
      "loss": 1.1987,
      "step": 2350
    },
    {
      "epoch": 0.09562204979639796,
      "grad_norm": 0.5962804555892944,
      "learning_rate": 0.00019951938262681527,
      "loss": 1.2077,
      "step": 2360
    },
    {
      "epoch": 0.09602722797350134,
      "grad_norm": 0.5692049264907837,
      "learning_rate": 0.0001995152158031997,
      "loss": 1.1916,
      "step": 2370
    },
    {
      "epoch": 0.09643240615060474,
      "grad_norm": 0.6170600056648254,
      "learning_rate": 0.0001995110310387276,
      "loss": 1.2174,
      "step": 2380
    },
    {
      "epoch": 0.09683758432770811,
      "grad_norm": 0.5726680159568787,
      "learning_rate": 0.00019950682833415345,
      "loss": 1.2127,
      "step": 2390
    },
    {
      "epoch": 0.0972427625048115,
      "grad_norm": 0.6218938827514648,
      "learning_rate": 0.00019950260769023483,
      "loss": 1.1762,
      "step": 2400
    },
    {
      "epoch": 0.09764794068191487,
      "grad_norm": 0.5263229012489319,
      "learning_rate": 0.0001994983691077327,
      "loss": 1.2208,
      "step": 2410
    },
    {
      "epoch": 0.09805311885901825,
      "grad_norm": 0.5397500395774841,
      "learning_rate": 0.00019949411258741125,
      "loss": 1.1967,
      "step": 2420
    },
    {
      "epoch": 0.09845829703612163,
      "grad_norm": 0.6292558908462524,
      "learning_rate": 0.00019948983813003774,
      "loss": 1.1646,
      "step": 2430
    },
    {
      "epoch": 0.09886347521322501,
      "grad_norm": 0.5726483464241028,
      "learning_rate": 0.0001994855457363829,
      "loss": 1.2052,
      "step": 2440
    },
    {
      "epoch": 0.09926865339032839,
      "grad_norm": 0.5180758237838745,
      "learning_rate": 0.0001994812354072205,
      "loss": 1.2026,
      "step": 2450
    },
    {
      "epoch": 0.09967383156743177,
      "grad_norm": 0.5640872120857239,
      "learning_rate": 0.0001994769071433276,
      "loss": 1.2135,
      "step": 2460
    },
    {
      "epoch": 0.10007900974453517,
      "grad_norm": 0.6218693256378174,
      "learning_rate": 0.0001994725609454846,
      "loss": 1.1955,
      "step": 2470
    },
    {
      "epoch": 0.10048418792163855,
      "grad_norm": 0.6009455919265747,
      "learning_rate": 0.00019946819681447496,
      "loss": 1.2072,
      "step": 2480
    },
    {
      "epoch": 0.10088936609874193,
      "grad_norm": 0.556208610534668,
      "learning_rate": 0.0001994638147510855,
      "loss": 1.1788,
      "step": 2490
    },
    {
      "epoch": 0.1012945442758453,
      "grad_norm": 0.5331783294677734,
      "learning_rate": 0.00019945941475610623,
      "loss": 1.1997,
      "step": 2500
    },
    {
      "epoch": 0.10169972245294868,
      "grad_norm": 0.53876131772995,
      "learning_rate": 0.0001994549968303304,
      "loss": 1.1916,
      "step": 2510
    },
    {
      "epoch": 0.10210490063005206,
      "grad_norm": 0.5424575805664062,
      "learning_rate": 0.0001994505609745544,
      "loss": 1.1549,
      "step": 2520
    },
    {
      "epoch": 0.10251007880715544,
      "grad_norm": 0.5526540279388428,
      "learning_rate": 0.00019944610718957803,
      "loss": 1.1829,
      "step": 2530
    },
    {
      "epoch": 0.10291525698425882,
      "grad_norm": 0.5224103331565857,
      "learning_rate": 0.00019944163547620423,
      "loss": 1.2013,
      "step": 2540
    },
    {
      "epoch": 0.1033204351613622,
      "grad_norm": 0.5213075280189514,
      "learning_rate": 0.0001994371458352391,
      "loss": 1.1964,
      "step": 2550
    },
    {
      "epoch": 0.1037256133384656,
      "grad_norm": 0.5753841400146484,
      "learning_rate": 0.00019943263826749212,
      "loss": 1.1878,
      "step": 2560
    },
    {
      "epoch": 0.10413079151556898,
      "grad_norm": 0.47740790247917175,
      "learning_rate": 0.0001994281127737759,
      "loss": 1.2009,
      "step": 2570
    },
    {
      "epoch": 0.10453596969267236,
      "grad_norm": 0.5779164433479309,
      "learning_rate": 0.0001994235693549063,
      "loss": 1.207,
      "step": 2580
    },
    {
      "epoch": 0.10494114786977574,
      "grad_norm": 0.5346842408180237,
      "learning_rate": 0.00019941900801170238,
      "loss": 1.1832,
      "step": 2590
    },
    {
      "epoch": 0.10534632604687912,
      "grad_norm": 0.5982937216758728,
      "learning_rate": 0.00019941442874498653,
      "loss": 1.2123,
      "step": 2600
    },
    {
      "epoch": 0.1057515042239825,
      "grad_norm": 0.5167168974876404,
      "learning_rate": 0.0001994098315555843,
      "loss": 1.2042,
      "step": 2610
    },
    {
      "epoch": 0.10615668240108588,
      "grad_norm": 0.5598909854888916,
      "learning_rate": 0.00019940521644432448,
      "loss": 1.1886,
      "step": 2620
    },
    {
      "epoch": 0.10656186057818925,
      "grad_norm": 0.4887411594390869,
      "learning_rate": 0.00019940058341203908,
      "loss": 1.1956,
      "step": 2630
    },
    {
      "epoch": 0.10696703875529263,
      "grad_norm": 0.5313457250595093,
      "learning_rate": 0.0001993959324595634,
      "loss": 1.1892,
      "step": 2640
    },
    {
      "epoch": 0.10737221693239603,
      "grad_norm": 0.6050583720207214,
      "learning_rate": 0.00019939126358773583,
      "loss": 1.1821,
      "step": 2650
    },
    {
      "epoch": 0.10777739510949941,
      "grad_norm": 0.5791787505149841,
      "learning_rate": 0.0001993865767973982,
      "loss": 1.2008,
      "step": 2660
    },
    {
      "epoch": 0.10818257328660279,
      "grad_norm": 0.5392870306968689,
      "learning_rate": 0.00019938187208939535,
      "loss": 1.214,
      "step": 2670
    },
    {
      "epoch": 0.10858775146370617,
      "grad_norm": 0.519639253616333,
      "learning_rate": 0.00019937714946457555,
      "loss": 1.1907,
      "step": 2680
    },
    {
      "epoch": 0.10899292964080955,
      "grad_norm": 0.5354943871498108,
      "learning_rate": 0.00019937240892379014,
      "loss": 1.2287,
      "step": 2690
    },
    {
      "epoch": 0.10939810781791293,
      "grad_norm": 0.5471699237823486,
      "learning_rate": 0.0001993676504678938,
      "loss": 1.23,
      "step": 2700
    },
    {
      "epoch": 0.1098032859950163,
      "grad_norm": 0.5014521479606628,
      "learning_rate": 0.0001993628740977444,
      "loss": 1.2177,
      "step": 2710
    },
    {
      "epoch": 0.11020846417211969,
      "grad_norm": 0.4782751798629761,
      "learning_rate": 0.00019935807981420297,
      "loss": 1.187,
      "step": 2720
    },
    {
      "epoch": 0.11061364234922307,
      "grad_norm": 0.663466215133667,
      "learning_rate": 0.0001993532676181339,
      "loss": 1.1995,
      "step": 2730
    },
    {
      "epoch": 0.11101882052632646,
      "grad_norm": 0.5099157094955444,
      "learning_rate": 0.00019934843751040474,
      "loss": 1.1859,
      "step": 2740
    },
    {
      "epoch": 0.11142399870342984,
      "grad_norm": 0.4947831630706787,
      "learning_rate": 0.00019934358949188626,
      "loss": 1.1937,
      "step": 2750
    },
    {
      "epoch": 0.11182917688053322,
      "grad_norm": 0.5718529224395752,
      "learning_rate": 0.00019933872356345248,
      "loss": 1.2268,
      "step": 2760
    },
    {
      "epoch": 0.1122343550576366,
      "grad_norm": 0.46727344393730164,
      "learning_rate": 0.00019933383972598061,
      "loss": 1.1908,
      "step": 2770
    },
    {
      "epoch": 0.11263953323473998,
      "grad_norm": 0.5159174203872681,
      "learning_rate": 0.00019932893798035116,
      "loss": 1.1808,
      "step": 2780
    },
    {
      "epoch": 0.11304471141184336,
      "grad_norm": 0.5514495968818665,
      "learning_rate": 0.0001993240183274478,
      "loss": 1.1979,
      "step": 2790
    },
    {
      "epoch": 0.11344988958894674,
      "grad_norm": 0.4924412667751312,
      "learning_rate": 0.00019931908076815748,
      "loss": 1.2012,
      "step": 2800
    },
    {
      "epoch": 0.11385506776605012,
      "grad_norm": 0.47463729977607727,
      "learning_rate": 0.00019931412530337038,
      "loss": 1.1895,
      "step": 2810
    },
    {
      "epoch": 0.1142602459431535,
      "grad_norm": 0.5282716751098633,
      "learning_rate": 0.0001993091519339798,
      "loss": 1.1713,
      "step": 2820
    },
    {
      "epoch": 0.11466542412025689,
      "grad_norm": 0.5193506479263306,
      "learning_rate": 0.00019930416066088244,
      "loss": 1.2146,
      "step": 2830
    },
    {
      "epoch": 0.11507060229736027,
      "grad_norm": 0.532280683517456,
      "learning_rate": 0.0001992991514849781,
      "loss": 1.1973,
      "step": 2840
    },
    {
      "epoch": 0.11547578047446365,
      "grad_norm": 0.5293679237365723,
      "learning_rate": 0.00019929412440716985,
      "loss": 1.1924,
      "step": 2850
    },
    {
      "epoch": 0.11588095865156703,
      "grad_norm": 0.5664437413215637,
      "learning_rate": 0.000199289079428364,
      "loss": 1.2258,
      "step": 2860
    },
    {
      "epoch": 0.11628613682867041,
      "grad_norm": 0.5874696969985962,
      "learning_rate": 0.00019928401654947,
      "loss": 1.1591,
      "step": 2870
    },
    {
      "epoch": 0.11669131500577379,
      "grad_norm": 0.5353211164474487,
      "learning_rate": 0.00019927893577140071,
      "loss": 1.2185,
      "step": 2880
    },
    {
      "epoch": 0.11709649318287717,
      "grad_norm": 0.5592243671417236,
      "learning_rate": 0.00019927383709507203,
      "loss": 1.2019,
      "step": 2890
    },
    {
      "epoch": 0.11750167135998055,
      "grad_norm": 0.5269444584846497,
      "learning_rate": 0.00019926872052140318,
      "loss": 1.2133,
      "step": 2900
    },
    {
      "epoch": 0.11790684953708393,
      "grad_norm": 0.5460145473480225,
      "learning_rate": 0.0001992635860513166,
      "loss": 1.1835,
      "step": 2910
    },
    {
      "epoch": 0.11831202771418732,
      "grad_norm": 0.6074547171592712,
      "learning_rate": 0.00019925843368573794,
      "loss": 1.2173,
      "step": 2920
    },
    {
      "epoch": 0.1187172058912907,
      "grad_norm": 0.49032652378082275,
      "learning_rate": 0.0001992532634255961,
      "loss": 1.1957,
      "step": 2930
    },
    {
      "epoch": 0.11912238406839408,
      "grad_norm": 0.5776464343070984,
      "learning_rate": 0.00019924807527182313,
      "loss": 1.139,
      "step": 2940
    },
    {
      "epoch": 0.11952756224549746,
      "grad_norm": 0.533450722694397,
      "learning_rate": 0.00019924286922535443,
      "loss": 1.1993,
      "step": 2950
    },
    {
      "epoch": 0.11993274042260084,
      "grad_norm": 0.5598276257514954,
      "learning_rate": 0.0001992376452871285,
      "loss": 1.1881,
      "step": 2960
    },
    {
      "epoch": 0.12033791859970422,
      "grad_norm": 0.5114431381225586,
      "learning_rate": 0.00019923240345808715,
      "loss": 1.2226,
      "step": 2970
    },
    {
      "epoch": 0.1207430967768076,
      "grad_norm": 0.5370688438415527,
      "learning_rate": 0.0001992271437391754,
      "loss": 1.2121,
      "step": 2980
    },
    {
      "epoch": 0.12114827495391098,
      "grad_norm": 0.5111677646636963,
      "learning_rate": 0.0001992218661313415,
      "loss": 1.2206,
      "step": 2990
    },
    {
      "epoch": 0.12155345313101436,
      "grad_norm": 0.46392837166786194,
      "learning_rate": 0.0001992165706355369,
      "loss": 1.2283,
      "step": 3000
    },
    {
      "epoch": 0.12195863130811775,
      "grad_norm": 0.5477924346923828,
      "learning_rate": 0.00019921125725271628,
      "loss": 1.202,
      "step": 3010
    },
    {
      "epoch": 0.12236380948522113,
      "grad_norm": 0.48720109462738037,
      "learning_rate": 0.00019920592598383753,
      "loss": 1.1965,
      "step": 3020
    },
    {
      "epoch": 0.12276898766232451,
      "grad_norm": 0.5792083740234375,
      "learning_rate": 0.0001992005768298618,
      "loss": 1.1711,
      "step": 3030
    },
    {
      "epoch": 0.12317416583942789,
      "grad_norm": 0.501604437828064,
      "learning_rate": 0.00019919520979175343,
      "loss": 1.1807,
      "step": 3040
    },
    {
      "epoch": 0.12357934401653127,
      "grad_norm": 0.5761296153068542,
      "learning_rate": 0.00019918982487048003,
      "loss": 1.1986,
      "step": 3050
    },
    {
      "epoch": 0.12398452219363465,
      "grad_norm": 0.5346300005912781,
      "learning_rate": 0.00019918442206701245,
      "loss": 1.2096,
      "step": 3060
    },
    {
      "epoch": 0.12438970037073803,
      "grad_norm": 0.5527538061141968,
      "learning_rate": 0.0001991790013823246,
      "loss": 1.1947,
      "step": 3070
    },
    {
      "epoch": 0.12479487854784141,
      "grad_norm": 0.6186645030975342,
      "learning_rate": 0.00019917356281739384,
      "loss": 1.2446,
      "step": 3080
    },
    {
      "epoch": 0.1252000567249448,
      "grad_norm": 0.5622219443321228,
      "learning_rate": 0.00019916810637320064,
      "loss": 1.2026,
      "step": 3090
    },
    {
      "epoch": 0.12560523490204817,
      "grad_norm": 0.5638346076011658,
      "learning_rate": 0.00019916263205072866,
      "loss": 1.2096,
      "step": 3100
    },
    {
      "epoch": 0.12601041307915156,
      "grad_norm": 0.47904664278030396,
      "learning_rate": 0.00019915713985096484,
      "loss": 1.1616,
      "step": 3110
    },
    {
      "epoch": 0.12641559125625493,
      "grad_norm": 0.5343408584594727,
      "learning_rate": 0.00019915162977489934,
      "loss": 1.2293,
      "step": 3120
    },
    {
      "epoch": 0.12682076943335832,
      "grad_norm": 0.5563844442367554,
      "learning_rate": 0.00019914610182352548,
      "loss": 1.1741,
      "step": 3130
    },
    {
      "epoch": 0.1272259476104617,
      "grad_norm": 0.5526943802833557,
      "learning_rate": 0.00019914055599783997,
      "loss": 1.1844,
      "step": 3140
    },
    {
      "epoch": 0.12763112578756508,
      "grad_norm": 0.542137086391449,
      "learning_rate": 0.0001991349922988425,
      "loss": 1.1849,
      "step": 3150
    },
    {
      "epoch": 0.12803630396466847,
      "grad_norm": 0.5083839297294617,
      "learning_rate": 0.00019912941072753614,
      "loss": 1.1653,
      "step": 3160
    },
    {
      "epoch": 0.12844148214177184,
      "grad_norm": 0.5019254088401794,
      "learning_rate": 0.00019912381128492717,
      "loss": 1.1738,
      "step": 3170
    },
    {
      "epoch": 0.12884666031887523,
      "grad_norm": 0.49254459142684937,
      "learning_rate": 0.0001991181939720251,
      "loss": 1.216,
      "step": 3180
    },
    {
      "epoch": 0.1292518384959786,
      "grad_norm": 0.4886765480041504,
      "learning_rate": 0.0001991125587898426,
      "loss": 1.168,
      "step": 3190
    },
    {
      "epoch": 0.129657016673082,
      "grad_norm": 0.5217731595039368,
      "learning_rate": 0.00019910690573939557,
      "loss": 1.1973,
      "step": 3200
    },
    {
      "epoch": 0.13006219485018536,
      "grad_norm": 0.4953235685825348,
      "learning_rate": 0.00019910123482170318,
      "loss": 1.1767,
      "step": 3210
    },
    {
      "epoch": 0.13046737302728875,
      "grad_norm": 0.48881539702415466,
      "learning_rate": 0.0001990955460377878,
      "loss": 1.1952,
      "step": 3220
    },
    {
      "epoch": 0.13087255120439212,
      "grad_norm": 0.5046960711479187,
      "learning_rate": 0.00019908983938867508,
      "loss": 1.2226,
      "step": 3230
    },
    {
      "epoch": 0.1312777293814955,
      "grad_norm": 0.5387177467346191,
      "learning_rate": 0.0001990841148753937,
      "loss": 1.2005,
      "step": 3240
    },
    {
      "epoch": 0.1316829075585989,
      "grad_norm": 0.51825350522995,
      "learning_rate": 0.0001990783724989758,
      "loss": 1.1927,
      "step": 3250
    },
    {
      "epoch": 0.13208808573570227,
      "grad_norm": 0.4898778796195984,
      "learning_rate": 0.00019907261226045656,
      "loss": 1.1833,
      "step": 3260
    },
    {
      "epoch": 0.13249326391280566,
      "grad_norm": 0.5447530746459961,
      "learning_rate": 0.00019906683416087448,
      "loss": 1.1814,
      "step": 3270
    },
    {
      "epoch": 0.13289844208990903,
      "grad_norm": 0.5295965075492859,
      "learning_rate": 0.0001990610382012713,
      "loss": 1.165,
      "step": 3280
    },
    {
      "epoch": 0.13330362026701242,
      "grad_norm": 0.5740903615951538,
      "learning_rate": 0.00019905522438269183,
      "loss": 1.1706,
      "step": 3290
    },
    {
      "epoch": 0.1337087984441158,
      "grad_norm": 0.5635108947753906,
      "learning_rate": 0.00019904939270618425,
      "loss": 1.2086,
      "step": 3300
    },
    {
      "epoch": 0.13411397662121918,
      "grad_norm": 0.5328755378723145,
      "learning_rate": 0.0001990435431727999,
      "loss": 1.193,
      "step": 3310
    },
    {
      "epoch": 0.13451915479832255,
      "grad_norm": 0.5842874050140381,
      "learning_rate": 0.0001990376757835934,
      "loss": 1.1801,
      "step": 3320
    },
    {
      "epoch": 0.13492433297542594,
      "grad_norm": 0.5608965754508972,
      "learning_rate": 0.00019903179053962246,
      "loss": 1.2276,
      "step": 3330
    },
    {
      "epoch": 0.13532951115252934,
      "grad_norm": 0.5050598382949829,
      "learning_rate": 0.00019902588744194813,
      "loss": 1.1728,
      "step": 3340
    },
    {
      "epoch": 0.1357346893296327,
      "grad_norm": 0.5049337148666382,
      "learning_rate": 0.00019901996649163465,
      "loss": 1.2078,
      "step": 3350
    },
    {
      "epoch": 0.1361398675067361,
      "grad_norm": 0.5450803637504578,
      "learning_rate": 0.0001990140276897494,
      "loss": 1.1833,
      "step": 3360
    },
    {
      "epoch": 0.13654504568383946,
      "grad_norm": 0.510260283946991,
      "learning_rate": 0.00019900807103736314,
      "loss": 1.2208,
      "step": 3370
    },
    {
      "epoch": 0.13695022386094285,
      "grad_norm": 0.5351301431655884,
      "learning_rate": 0.00019900209653554967,
      "loss": 1.1791,
      "step": 3380
    },
    {
      "epoch": 0.13735540203804622,
      "grad_norm": 0.627776563167572,
      "learning_rate": 0.0001989961041853861,
      "loss": 1.1686,
      "step": 3390
    },
    {
      "epoch": 0.13776058021514961,
      "grad_norm": 0.4731081426143646,
      "learning_rate": 0.00019899009398795277,
      "loss": 1.1777,
      "step": 3400
    },
    {
      "epoch": 0.13816575839225298,
      "grad_norm": 0.4708987772464752,
      "learning_rate": 0.0001989840659443332,
      "loss": 1.1525,
      "step": 3410
    },
    {
      "epoch": 0.13857093656935637,
      "grad_norm": 0.5507491827011108,
      "learning_rate": 0.00019897802005561417,
      "loss": 1.1863,
      "step": 3420
    },
    {
      "epoch": 0.13897611474645977,
      "grad_norm": 0.5090895295143127,
      "learning_rate": 0.0001989719563228856,
      "loss": 1.15,
      "step": 3430
    },
    {
      "epoch": 0.13938129292356313,
      "grad_norm": 0.4595957398414612,
      "learning_rate": 0.00019896587474724068,
      "loss": 1.1725,
      "step": 3440
    },
    {
      "epoch": 0.13978647110066653,
      "grad_norm": 0.48999813199043274,
      "learning_rate": 0.00019895977532977585,
      "loss": 1.1863,
      "step": 3450
    },
    {
      "epoch": 0.1401916492777699,
      "grad_norm": 0.48516976833343506,
      "learning_rate": 0.00019895365807159071,
      "loss": 1.1529,
      "step": 3460
    },
    {
      "epoch": 0.14059682745487329,
      "grad_norm": 0.553671658039093,
      "learning_rate": 0.0001989475229737881,
      "loss": 1.2199,
      "step": 3470
    },
    {
      "epoch": 0.14100200563197665,
      "grad_norm": 0.5794891715049744,
      "learning_rate": 0.00019894137003747403,
      "loss": 1.2266,
      "step": 3480
    },
    {
      "epoch": 0.14140718380908004,
      "grad_norm": 0.5778114199638367,
      "learning_rate": 0.00019893519926375783,
      "loss": 1.1518,
      "step": 3490
    },
    {
      "epoch": 0.1418123619861834,
      "grad_norm": 0.5673059821128845,
      "learning_rate": 0.00019892901065375195,
      "loss": 1.2012,
      "step": 3500
    },
    {
      "epoch": 0.1422175401632868,
      "grad_norm": 0.5270010828971863,
      "learning_rate": 0.0001989228042085721,
      "loss": 1.1476,
      "step": 3510
    },
    {
      "epoch": 0.1426227183403902,
      "grad_norm": 0.6476484537124634,
      "learning_rate": 0.00019891657992933715,
      "loss": 1.2184,
      "step": 3520
    },
    {
      "epoch": 0.14302789651749356,
      "grad_norm": 0.5372319221496582,
      "learning_rate": 0.00019891033781716932,
      "loss": 1.2161,
      "step": 3530
    },
    {
      "epoch": 0.14343307469459696,
      "grad_norm": 0.5563995242118835,
      "learning_rate": 0.00019890407787319386,
      "loss": 1.203,
      "step": 3540
    },
    {
      "epoch": 0.14383825287170032,
      "grad_norm": 0.45160818099975586,
      "learning_rate": 0.0001988978000985394,
      "loss": 1.1828,
      "step": 3550
    },
    {
      "epoch": 0.14424343104880372,
      "grad_norm": 0.5513964295387268,
      "learning_rate": 0.00019889150449433764,
      "loss": 1.1342,
      "step": 3560
    },
    {
      "epoch": 0.14464860922590708,
      "grad_norm": 0.5496894717216492,
      "learning_rate": 0.00019888519106172364,
      "loss": 1.1578,
      "step": 3570
    },
    {
      "epoch": 0.14505378740301048,
      "grad_norm": 0.5630394220352173,
      "learning_rate": 0.00019887885980183554,
      "loss": 1.1629,
      "step": 3580
    },
    {
      "epoch": 0.14545896558011384,
      "grad_norm": 0.5818898677825928,
      "learning_rate": 0.0001988725107158148,
      "loss": 1.2154,
      "step": 3590
    },
    {
      "epoch": 0.14586414375721724,
      "grad_norm": 0.6013431549072266,
      "learning_rate": 0.00019886614380480605,
      "loss": 1.1867,
      "step": 3600
    },
    {
      "epoch": 0.14626932193432063,
      "grad_norm": 0.493478387594223,
      "learning_rate": 0.0001988597590699571,
      "loss": 1.1838,
      "step": 3610
    },
    {
      "epoch": 0.146674500111424,
      "grad_norm": 0.5780825018882751,
      "learning_rate": 0.00019885335651241903,
      "loss": 1.1733,
      "step": 3620
    },
    {
      "epoch": 0.1470796782885274,
      "grad_norm": 0.5627794861793518,
      "learning_rate": 0.0001988469361333461,
      "loss": 1.2137,
      "step": 3630
    },
    {
      "epoch": 0.14748485646563075,
      "grad_norm": 0.5765925049781799,
      "learning_rate": 0.00019884049793389582,
      "loss": 1.2249,
      "step": 3640
    },
    {
      "epoch": 0.14789003464273415,
      "grad_norm": 0.5342803597450256,
      "learning_rate": 0.00019883404191522883,
      "loss": 1.167,
      "step": 3650
    },
    {
      "epoch": 0.1482952128198375,
      "grad_norm": 0.5779811143875122,
      "learning_rate": 0.0001988275680785091,
      "loss": 1.168,
      "step": 3660
    },
    {
      "epoch": 0.1487003909969409,
      "grad_norm": 0.5396864414215088,
      "learning_rate": 0.00019882107642490366,
      "loss": 1.1672,
      "step": 3670
    },
    {
      "epoch": 0.14910556917404427,
      "grad_norm": 0.5380362868309021,
      "learning_rate": 0.00019881456695558297,
      "loss": 1.2306,
      "step": 3680
    },
    {
      "epoch": 0.14951074735114767,
      "grad_norm": 0.5618273019790649,
      "learning_rate": 0.00019880803967172047,
      "loss": 1.1599,
      "step": 3690
    },
    {
      "epoch": 0.14991592552825106,
      "grad_norm": 0.5917264223098755,
      "learning_rate": 0.00019880149457449293,
      "loss": 1.1848,
      "step": 3700
    },
    {
      "epoch": 0.15032110370535443,
      "grad_norm": 0.5425720810890198,
      "learning_rate": 0.00019879493166508036,
      "loss": 1.1748,
      "step": 3710
    },
    {
      "epoch": 0.15072628188245782,
      "grad_norm": 0.5223249197006226,
      "learning_rate": 0.00019878835094466591,
      "loss": 1.2102,
      "step": 3720
    },
    {
      "epoch": 0.15113146005956118,
      "grad_norm": 0.5962805151939392,
      "learning_rate": 0.00019878175241443597,
      "loss": 1.1747,
      "step": 3730
    },
    {
      "epoch": 0.15153663823666458,
      "grad_norm": 0.5542666912078857,
      "learning_rate": 0.00019877513607558016,
      "loss": 1.1995,
      "step": 3740
    },
    {
      "epoch": 0.15194181641376794,
      "grad_norm": 0.5542714595794678,
      "learning_rate": 0.00019876850192929125,
      "loss": 1.202,
      "step": 3750
    },
    {
      "epoch": 0.15234699459087134,
      "grad_norm": 0.4992958903312683,
      "learning_rate": 0.0001987618499767653,
      "loss": 1.2205,
      "step": 3760
    },
    {
      "epoch": 0.1527521727679747,
      "grad_norm": 0.549767255783081,
      "learning_rate": 0.0001987551802192015,
      "loss": 1.1664,
      "step": 3770
    },
    {
      "epoch": 0.1531573509450781,
      "grad_norm": 0.5012533068656921,
      "learning_rate": 0.00019874849265780236,
      "loss": 1.185,
      "step": 3780
    },
    {
      "epoch": 0.1535625291221815,
      "grad_norm": 0.5907428860664368,
      "learning_rate": 0.00019874178729377344,
      "loss": 1.2135,
      "step": 3790
    },
    {
      "epoch": 0.15396770729928486,
      "grad_norm": 0.5544156432151794,
      "learning_rate": 0.00019873506412832368,
      "loss": 1.1636,
      "step": 3800
    },
    {
      "epoch": 0.15437288547638825,
      "grad_norm": 0.6347226500511169,
      "learning_rate": 0.0001987283231626651,
      "loss": 1.1874,
      "step": 3810
    },
    {
      "epoch": 0.15477806365349162,
      "grad_norm": 0.5499520301818848,
      "learning_rate": 0.00019872156439801298,
      "loss": 1.1985,
      "step": 3820
    },
    {
      "epoch": 0.155183241830595,
      "grad_norm": 0.5601209998130798,
      "learning_rate": 0.00019871478783558587,
      "loss": 1.1897,
      "step": 3830
    },
    {
      "epoch": 0.15558842000769837,
      "grad_norm": 0.514732837677002,
      "learning_rate": 0.00019870799347660535,
      "loss": 1.2036,
      "step": 3840
    },
    {
      "epoch": 0.15599359818480177,
      "grad_norm": 0.490898072719574,
      "learning_rate": 0.00019870118132229645,
      "loss": 1.1586,
      "step": 3850
    },
    {
      "epoch": 0.15639877636190513,
      "grad_norm": 0.5515938401222229,
      "learning_rate": 0.0001986943513738872,
      "loss": 1.1559,
      "step": 3860
    },
    {
      "epoch": 0.15680395453900853,
      "grad_norm": 0.5001558661460876,
      "learning_rate": 0.00019868750363260894,
      "loss": 1.1978,
      "step": 3870
    },
    {
      "epoch": 0.15720913271611192,
      "grad_norm": 0.5464389324188232,
      "learning_rate": 0.00019868063809969623,
      "loss": 1.2053,
      "step": 3880
    },
    {
      "epoch": 0.1576143108932153,
      "grad_norm": 0.5357148051261902,
      "learning_rate": 0.00019867375477638675,
      "loss": 1.2036,
      "step": 3890
    },
    {
      "epoch": 0.15801948907031868,
      "grad_norm": 0.5269990563392639,
      "learning_rate": 0.0001986668536639215,
      "loss": 1.1921,
      "step": 3900
    },
    {
      "epoch": 0.15842466724742205,
      "grad_norm": 0.5944133400917053,
      "learning_rate": 0.00019865993476354457,
      "loss": 1.1755,
      "step": 3910
    },
    {
      "epoch": 0.15882984542452544,
      "grad_norm": 0.4713591933250427,
      "learning_rate": 0.00019865299807650337,
      "loss": 1.1478,
      "step": 3920
    },
    {
      "epoch": 0.1592350236016288,
      "grad_norm": 0.5976880788803101,
      "learning_rate": 0.00019864604360404843,
      "loss": 1.1618,
      "step": 3930
    },
    {
      "epoch": 0.1596402017787322,
      "grad_norm": 0.5041770935058594,
      "learning_rate": 0.00019863907134743354,
      "loss": 1.1869,
      "step": 3940
    },
    {
      "epoch": 0.16004537995583556,
      "grad_norm": 0.5543539524078369,
      "learning_rate": 0.00019863208130791566,
      "loss": 1.1978,
      "step": 3950
    },
    {
      "epoch": 0.16045055813293896,
      "grad_norm": 0.5323906540870667,
      "learning_rate": 0.00019862507348675497,
      "loss": 1.1899,
      "step": 3960
    },
    {
      "epoch": 0.16085573631004235,
      "grad_norm": 0.5329884886741638,
      "learning_rate": 0.00019861804788521493,
      "loss": 1.1602,
      "step": 3970
    },
    {
      "epoch": 0.16126091448714572,
      "grad_norm": 0.5395342111587524,
      "learning_rate": 0.000198611004504562,
      "loss": 1.1908,
      "step": 3980
    },
    {
      "epoch": 0.1616660926642491,
      "grad_norm": 0.49201568961143494,
      "learning_rate": 0.00019860394334606608,
      "loss": 1.1906,
      "step": 3990
    },
    {
      "epoch": 0.16207127084135248,
      "grad_norm": 0.5451162457466125,
      "learning_rate": 0.00019859686441100016,
      "loss": 1.1444,
      "step": 4000
    },
    {
      "epoch": 0.16247644901845587,
      "grad_norm": 0.5425843596458435,
      "learning_rate": 0.00019858976770064038,
      "loss": 1.2067,
      "step": 4010
    },
    {
      "epoch": 0.16288162719555924,
      "grad_norm": 0.5842928886413574,
      "learning_rate": 0.00019858265321626628,
      "loss": 1.173,
      "step": 4020
    },
    {
      "epoch": 0.16328680537266263,
      "grad_norm": 0.5583239197731018,
      "learning_rate": 0.00019857552095916035,
      "loss": 1.1898,
      "step": 4030
    },
    {
      "epoch": 0.163691983549766,
      "grad_norm": 0.49922531843185425,
      "learning_rate": 0.00019856837093060848,
      "loss": 1.1742,
      "step": 4040
    },
    {
      "epoch": 0.1640971617268694,
      "grad_norm": 0.4921461045742035,
      "learning_rate": 0.00019856120313189969,
      "loss": 1.1984,
      "step": 4050
    },
    {
      "epoch": 0.16450233990397278,
      "grad_norm": 0.5322550535202026,
      "learning_rate": 0.0001985540175643262,
      "loss": 1.1964,
      "step": 4060
    },
    {
      "epoch": 0.16490751808107615,
      "grad_norm": 0.5548688173294067,
      "learning_rate": 0.00019854681422918342,
      "loss": 1.2015,
      "step": 4070
    },
    {
      "epoch": 0.16531269625817954,
      "grad_norm": 0.5768460631370544,
      "learning_rate": 0.00019853959312777,
      "loss": 1.1913,
      "step": 4080
    },
    {
      "epoch": 0.1657178744352829,
      "grad_norm": 0.5354451537132263,
      "learning_rate": 0.00019853235426138782,
      "loss": 1.1934,
      "step": 4090
    },
    {
      "epoch": 0.1661230526123863,
      "grad_norm": 0.5146519541740417,
      "learning_rate": 0.00019852509763134188,
      "loss": 1.1821,
      "step": 4100
    },
    {
      "epoch": 0.16652823078948967,
      "grad_norm": 0.5761153101921082,
      "learning_rate": 0.00019851782323894042,
      "loss": 1.1781,
      "step": 4110
    },
    {
      "epoch": 0.16693340896659306,
      "grad_norm": 0.5566679835319519,
      "learning_rate": 0.00019851053108549489,
      "loss": 1.1823,
      "step": 4120
    },
    {
      "epoch": 0.16733858714369643,
      "grad_norm": 0.51987224817276,
      "learning_rate": 0.00019850322117231995,
      "loss": 1.2013,
      "step": 4130
    },
    {
      "epoch": 0.16774376532079982,
      "grad_norm": 0.525178074836731,
      "learning_rate": 0.00019849589350073345,
      "loss": 1.1392,
      "step": 4140
    },
    {
      "epoch": 0.1681489434979032,
      "grad_norm": 0.5563574433326721,
      "learning_rate": 0.00019848854807205646,
      "loss": 1.2082,
      "step": 4150
    },
    {
      "epoch": 0.16855412167500658,
      "grad_norm": 0.5701225399971008,
      "learning_rate": 0.0001984811848876132,
      "loss": 1.1697,
      "step": 4160
    },
    {
      "epoch": 0.16895929985210997,
      "grad_norm": 0.5190890431404114,
      "learning_rate": 0.0001984738039487311,
      "loss": 1.1893,
      "step": 4170
    },
    {
      "epoch": 0.16936447802921334,
      "grad_norm": 0.5622174739837646,
      "learning_rate": 0.00019846640525674082,
      "loss": 1.2004,
      "step": 4180
    },
    {
      "epoch": 0.16976965620631673,
      "grad_norm": 0.5650730133056641,
      "learning_rate": 0.0001984589888129763,
      "loss": 1.2006,
      "step": 4190
    },
    {
      "epoch": 0.1701748343834201,
      "grad_norm": 0.5711023211479187,
      "learning_rate": 0.00019845155461877453,
      "loss": 1.205,
      "step": 4200
    },
    {
      "epoch": 0.1705800125605235,
      "grad_norm": 0.5364766120910645,
      "learning_rate": 0.00019844410267547575,
      "loss": 1.1616,
      "step": 4210
    },
    {
      "epoch": 0.17098519073762686,
      "grad_norm": 0.47628334164619446,
      "learning_rate": 0.00019843663298442344,
      "loss": 1.1347,
      "step": 4220
    },
    {
      "epoch": 0.17139036891473025,
      "grad_norm": 0.5675886273384094,
      "learning_rate": 0.00019842914554696429,
      "loss": 1.1907,
      "step": 4230
    },
    {
      "epoch": 0.17179554709183364,
      "grad_norm": 0.5264836549758911,
      "learning_rate": 0.00019842164036444808,
      "loss": 1.1789,
      "step": 4240
    },
    {
      "epoch": 0.172200725268937,
      "grad_norm": 0.5260933637619019,
      "learning_rate": 0.0001984141174382279,
      "loss": 1.2078,
      "step": 4250
    },
    {
      "epoch": 0.1726059034460404,
      "grad_norm": 0.5060850381851196,
      "learning_rate": 0.00019840657676966,
      "loss": 1.1728,
      "step": 4260
    },
    {
      "epoch": 0.17301108162314377,
      "grad_norm": 0.582562267780304,
      "learning_rate": 0.00019839901836010387,
      "loss": 1.1981,
      "step": 4270
    },
    {
      "epoch": 0.17341625980024716,
      "grad_norm": 0.5389764308929443,
      "learning_rate": 0.0001983914422109221,
      "loss": 1.1736,
      "step": 4280
    },
    {
      "epoch": 0.17382143797735053,
      "grad_norm": 0.502723753452301,
      "learning_rate": 0.00019838384832348056,
      "loss": 1.1843,
      "step": 4290
    },
    {
      "epoch": 0.17422661615445392,
      "grad_norm": 0.5601800680160522,
      "learning_rate": 0.00019837623669914826,
      "loss": 1.2068,
      "step": 4300
    },
    {
      "epoch": 0.1746317943315573,
      "grad_norm": 0.6284868121147156,
      "learning_rate": 0.00019836860733929754,
      "loss": 1.1825,
      "step": 4310
    },
    {
      "epoch": 0.17503697250866068,
      "grad_norm": 0.5746595859527588,
      "learning_rate": 0.00019836096024530373,
      "loss": 1.1711,
      "step": 4320
    },
    {
      "epoch": 0.17544215068576408,
      "grad_norm": 0.5385555624961853,
      "learning_rate": 0.00019835329541854555,
      "loss": 1.2005,
      "step": 4330
    },
    {
      "epoch": 0.17584732886286744,
      "grad_norm": 0.5753733515739441,
      "learning_rate": 0.00019834561286040479,
      "loss": 1.1353,
      "step": 4340
    },
    {
      "epoch": 0.17625250703997083,
      "grad_norm": 0.5159021615982056,
      "learning_rate": 0.0001983379125722665,
      "loss": 1.2313,
      "step": 4350
    },
    {
      "epoch": 0.1766576852170742,
      "grad_norm": 0.5378077626228333,
      "learning_rate": 0.0001983301945555189,
      "loss": 1.1578,
      "step": 4360
    },
    {
      "epoch": 0.1770628633941776,
      "grad_norm": 0.5632703304290771,
      "learning_rate": 0.00019832245881155338,
      "loss": 1.1805,
      "step": 4370
    },
    {
      "epoch": 0.17746804157128096,
      "grad_norm": 0.6352266669273376,
      "learning_rate": 0.00019831470534176462,
      "loss": 1.2057,
      "step": 4380
    },
    {
      "epoch": 0.17787321974838435,
      "grad_norm": 0.5569392442703247,
      "learning_rate": 0.0001983069341475504,
      "loss": 1.1908,
      "step": 4390
    },
    {
      "epoch": 0.17827839792548772,
      "grad_norm": 0.5662495493888855,
      "learning_rate": 0.00019829914523031175,
      "loss": 1.164,
      "step": 4400
    },
    {
      "epoch": 0.1786835761025911,
      "grad_norm": 0.5639083981513977,
      "learning_rate": 0.00019829133859145285,
      "loss": 1.181,
      "step": 4410
    },
    {
      "epoch": 0.1790887542796945,
      "grad_norm": 0.5491625070571899,
      "learning_rate": 0.00019828351423238112,
      "loss": 1.2067,
      "step": 4420
    },
    {
      "epoch": 0.17949393245679787,
      "grad_norm": 0.5765697360038757,
      "learning_rate": 0.00019827567215450716,
      "loss": 1.2037,
      "step": 4430
    },
    {
      "epoch": 0.17989911063390127,
      "grad_norm": 0.5551952123641968,
      "learning_rate": 0.00019826781235924475,
      "loss": 1.167,
      "step": 4440
    },
    {
      "epoch": 0.18030428881100463,
      "grad_norm": 0.520746648311615,
      "learning_rate": 0.00019825993484801088,
      "loss": 1.1541,
      "step": 4450
    },
    {
      "epoch": 0.18070946698810803,
      "grad_norm": 0.4712527096271515,
      "learning_rate": 0.00019825203962222572,
      "loss": 1.1641,
      "step": 4460
    },
    {
      "epoch": 0.1811146451652114,
      "grad_norm": 0.4983481466770172,
      "learning_rate": 0.00019824412668331264,
      "loss": 1.1701,
      "step": 4470
    },
    {
      "epoch": 0.18151982334231478,
      "grad_norm": 0.5271497964859009,
      "learning_rate": 0.0001982361960326982,
      "loss": 1.2025,
      "step": 4480
    },
    {
      "epoch": 0.18192500151941818,
      "grad_norm": 0.7264400124549866,
      "learning_rate": 0.00019822824767181215,
      "loss": 1.183,
      "step": 4490
    },
    {
      "epoch": 0.18233017969652154,
      "grad_norm": 0.577533483505249,
      "learning_rate": 0.0001982202816020875,
      "loss": 1.1346,
      "step": 4500
    },
    {
      "epoch": 0.18273535787362494,
      "grad_norm": 0.5625945925712585,
      "learning_rate": 0.00019821229782496032,
      "loss": 1.2042,
      "step": 4510
    },
    {
      "epoch": 0.1831405360507283,
      "grad_norm": 0.5709630846977234,
      "learning_rate": 0.00019820429634187,
      "loss": 1.1603,
      "step": 4520
    },
    {
      "epoch": 0.1835457142278317,
      "grad_norm": 0.5115880966186523,
      "learning_rate": 0.00019819627715425903,
      "loss": 1.1666,
      "step": 4530
    },
    {
      "epoch": 0.18395089240493506,
      "grad_norm": 0.5092071890830994,
      "learning_rate": 0.00019818824026357316,
      "loss": 1.1588,
      "step": 4540
    },
    {
      "epoch": 0.18435607058203846,
      "grad_norm": 0.5321983098983765,
      "learning_rate": 0.00019818018567126127,
      "loss": 1.158,
      "step": 4550
    },
    {
      "epoch": 0.18476124875914182,
      "grad_norm": 0.5401438474655151,
      "learning_rate": 0.00019817211337877546,
      "loss": 1.1867,
      "step": 4560
    },
    {
      "epoch": 0.18516642693624522,
      "grad_norm": 0.5726510286331177,
      "learning_rate": 0.00019816402338757106,
      "loss": 1.1596,
      "step": 4570
    },
    {
      "epoch": 0.1855716051133486,
      "grad_norm": 0.630236029624939,
      "learning_rate": 0.00019815591569910654,
      "loss": 1.1947,
      "step": 4580
    },
    {
      "epoch": 0.18597678329045197,
      "grad_norm": 0.5077782273292542,
      "learning_rate": 0.00019814779031484357,
      "loss": 1.1848,
      "step": 4590
    },
    {
      "epoch": 0.18638196146755537,
      "grad_norm": 0.572134792804718,
      "learning_rate": 0.000198139647236247,
      "loss": 1.1709,
      "step": 4600
    },
    {
      "epoch": 0.18678713964465873,
      "grad_norm": 0.5199133157730103,
      "learning_rate": 0.00019813148646478492,
      "loss": 1.2086,
      "step": 4610
    },
    {
      "epoch": 0.18719231782176213,
      "grad_norm": 0.5801503658294678,
      "learning_rate": 0.00019812330800192854,
      "loss": 1.1773,
      "step": 4620
    },
    {
      "epoch": 0.1875974959988655,
      "grad_norm": 0.5252848863601685,
      "learning_rate": 0.0001981151118491523,
      "loss": 1.1653,
      "step": 4630
    },
    {
      "epoch": 0.1880026741759689,
      "grad_norm": 0.5934649109840393,
      "learning_rate": 0.00019810689800793386,
      "loss": 1.1198,
      "step": 4640
    },
    {
      "epoch": 0.18840785235307225,
      "grad_norm": 0.5457133650779724,
      "learning_rate": 0.00019809866647975397,
      "loss": 1.1633,
      "step": 4650
    },
    {
      "epoch": 0.18881303053017565,
      "grad_norm": 0.5340641736984253,
      "learning_rate": 0.00019809041726609667,
      "loss": 1.1993,
      "step": 4660
    },
    {
      "epoch": 0.18921820870727904,
      "grad_norm": 0.5302497148513794,
      "learning_rate": 0.00019808215036844917,
      "loss": 1.1539,
      "step": 4670
    },
    {
      "epoch": 0.1896233868843824,
      "grad_norm": 0.5069331526756287,
      "learning_rate": 0.0001980738657883018,
      "loss": 1.2013,
      "step": 4680
    },
    {
      "epoch": 0.1900285650614858,
      "grad_norm": 0.5215924978256226,
      "learning_rate": 0.00019806556352714816,
      "loss": 1.1419,
      "step": 4690
    },
    {
      "epoch": 0.19043374323858916,
      "grad_norm": 0.5689036846160889,
      "learning_rate": 0.00019805724358648496,
      "loss": 1.1842,
      "step": 4700
    },
    {
      "epoch": 0.19083892141569256,
      "grad_norm": 0.542327344417572,
      "learning_rate": 0.0001980489059678122,
      "loss": 1.2111,
      "step": 4710
    },
    {
      "epoch": 0.19124409959279592,
      "grad_norm": 0.5050385594367981,
      "learning_rate": 0.00019804055067263298,
      "loss": 1.1805,
      "step": 4720
    },
    {
      "epoch": 0.19164927776989932,
      "grad_norm": 0.5669646263122559,
      "learning_rate": 0.00019803217770245359,
      "loss": 1.1646,
      "step": 4730
    },
    {
      "epoch": 0.19205445594700268,
      "grad_norm": 0.5728495121002197,
      "learning_rate": 0.00019802378705878354,
      "loss": 1.2078,
      "step": 4740
    },
    {
      "epoch": 0.19245963412410608,
      "grad_norm": 0.5116891860961914,
      "learning_rate": 0.00019801537874313555,
      "loss": 1.1731,
      "step": 4750
    },
    {
      "epoch": 0.19286481230120947,
      "grad_norm": 0.5457078814506531,
      "learning_rate": 0.00019800695275702548,
      "loss": 1.1502,
      "step": 4760
    },
    {
      "epoch": 0.19326999047831284,
      "grad_norm": 0.5586498379707336,
      "learning_rate": 0.00019799850910197234,
      "loss": 1.1532,
      "step": 4770
    },
    {
      "epoch": 0.19367516865541623,
      "grad_norm": 0.5968454480171204,
      "learning_rate": 0.00019799004777949845,
      "loss": 1.1797,
      "step": 4780
    },
    {
      "epoch": 0.1940803468325196,
      "grad_norm": 0.5715598464012146,
      "learning_rate": 0.0001979815687911292,
      "loss": 1.187,
      "step": 4790
    },
    {
      "epoch": 0.194485525009623,
      "grad_norm": 0.5167731046676636,
      "learning_rate": 0.00019797307213839318,
      "loss": 1.1775,
      "step": 4800
    },
    {
      "epoch": 0.19489070318672636,
      "grad_norm": 0.5027974843978882,
      "learning_rate": 0.0001979645578228222,
      "loss": 1.1865,
      "step": 4810
    },
    {
      "epoch": 0.19529588136382975,
      "grad_norm": 0.5357294082641602,
      "learning_rate": 0.0001979560258459513,
      "loss": 1.1411,
      "step": 4820
    },
    {
      "epoch": 0.19570105954093311,
      "grad_norm": 0.5046802163124084,
      "learning_rate": 0.00019794747620931858,
      "loss": 1.1697,
      "step": 4830
    },
    {
      "epoch": 0.1961062377180365,
      "grad_norm": 0.5280042886734009,
      "learning_rate": 0.0001979389089144654,
      "loss": 1.1628,
      "step": 4840
    },
    {
      "epoch": 0.1965114158951399,
      "grad_norm": 0.5428780317306519,
      "learning_rate": 0.0001979303239629363,
      "loss": 1.1891,
      "step": 4850
    },
    {
      "epoch": 0.19691659407224327,
      "grad_norm": 0.5249727964401245,
      "learning_rate": 0.00019792172135627904,
      "loss": 1.1313,
      "step": 4860
    },
    {
      "epoch": 0.19732177224934666,
      "grad_norm": 0.6743662357330322,
      "learning_rate": 0.00019791310109604446,
      "loss": 1.1714,
      "step": 4870
    },
    {
      "epoch": 0.19772695042645003,
      "grad_norm": 0.5400875210762024,
      "learning_rate": 0.00019790446318378665,
      "loss": 1.1722,
      "step": 4880
    },
    {
      "epoch": 0.19813212860355342,
      "grad_norm": 0.5147051811218262,
      "learning_rate": 0.0001978958076210629,
      "loss": 1.2032,
      "step": 4890
    },
    {
      "epoch": 0.19853730678065679,
      "grad_norm": 0.5483666658401489,
      "learning_rate": 0.00019788713440943364,
      "loss": 1.169,
      "step": 4900
    },
    {
      "epoch": 0.19894248495776018,
      "grad_norm": 0.5123507380485535,
      "learning_rate": 0.00019787844355046255,
      "loss": 1.1297,
      "step": 4910
    },
    {
      "epoch": 0.19934766313486355,
      "grad_norm": 0.5753879547119141,
      "learning_rate": 0.00019786973504571636,
      "loss": 1.2224,
      "step": 4920
    },
    {
      "epoch": 0.19975284131196694,
      "grad_norm": 0.5713999271392822,
      "learning_rate": 0.00019786100889676507,
      "loss": 1.2126,
      "step": 4930
    },
    {
      "epoch": 0.20015801948907033,
      "grad_norm": 0.5685895085334778,
      "learning_rate": 0.0001978522651051819,
      "loss": 1.2014,
      "step": 4940
    },
    {
      "epoch": 0.2005631976661737,
      "grad_norm": 0.601669430732727,
      "learning_rate": 0.00019784350367254322,
      "loss": 1.1385,
      "step": 4950
    },
    {
      "epoch": 0.2009683758432771,
      "grad_norm": 0.5316070318222046,
      "learning_rate": 0.0001978347246004285,
      "loss": 1.1835,
      "step": 4960
    },
    {
      "epoch": 0.20137355402038046,
      "grad_norm": 0.5628255009651184,
      "learning_rate": 0.00019782592789042048,
      "loss": 1.1718,
      "step": 4970
    },
    {
      "epoch": 0.20177873219748385,
      "grad_norm": 0.5227285027503967,
      "learning_rate": 0.000197817113544105,
      "loss": 1.1523,
      "step": 4980
    },
    {
      "epoch": 0.20218391037458722,
      "grad_norm": 0.5679491758346558,
      "learning_rate": 0.00019780828156307125,
      "loss": 1.1584,
      "step": 4990
    },
    {
      "epoch": 0.2025890885516906,
      "grad_norm": 0.5761301517486572,
      "learning_rate": 0.0001977994319489114,
      "loss": 1.1642,
      "step": 5000
    },
    {
      "epoch": 0.20299426672879398,
      "grad_norm": 0.503194272518158,
      "learning_rate": 0.0001977905647032209,
      "loss": 1.1803,
      "step": 5010
    },
    {
      "epoch": 0.20339944490589737,
      "grad_norm": 0.5673863887786865,
      "learning_rate": 0.00019778167982759833,
      "loss": 1.1732,
      "step": 5020
    },
    {
      "epoch": 0.20380462308300076,
      "grad_norm": 0.6259693503379822,
      "learning_rate": 0.00019777277732364553,
      "loss": 1.2213,
      "step": 5030
    },
    {
      "epoch": 0.20420980126010413,
      "grad_norm": 0.5718217492103577,
      "learning_rate": 0.00019776385719296746,
      "loss": 1.1343,
      "step": 5040
    },
    {
      "epoch": 0.20461497943720752,
      "grad_norm": 0.45593589544296265,
      "learning_rate": 0.00019775491943717222,
      "loss": 1.1488,
      "step": 5050
    },
    {
      "epoch": 0.2050201576143109,
      "grad_norm": 0.5223670601844788,
      "learning_rate": 0.00019774596405787117,
      "loss": 1.1907,
      "step": 5060
    },
    {
      "epoch": 0.20542533579141428,
      "grad_norm": 0.5180227160453796,
      "learning_rate": 0.00019773699105667879,
      "loss": 1.1853,
      "step": 5070
    },
    {
      "epoch": 0.20583051396851765,
      "grad_norm": 0.5505089163780212,
      "learning_rate": 0.00019772800043521273,
      "loss": 1.1995,
      "step": 5080
    },
    {
      "epoch": 0.20623569214562104,
      "grad_norm": 0.539402425289154,
      "learning_rate": 0.0001977189921950939,
      "loss": 1.1859,
      "step": 5090
    },
    {
      "epoch": 0.2066408703227244,
      "grad_norm": 0.541958749294281,
      "learning_rate": 0.0001977099663379463,
      "loss": 1.1689,
      "step": 5100
    },
    {
      "epoch": 0.2070460484998278,
      "grad_norm": 0.546512246131897,
      "learning_rate": 0.00019770092286539715,
      "loss": 1.2073,
      "step": 5110
    },
    {
      "epoch": 0.2074512266769312,
      "grad_norm": 0.5232321619987488,
      "learning_rate": 0.00019769186177907679,
      "loss": 1.1265,
      "step": 5120
    },
    {
      "epoch": 0.20785640485403456,
      "grad_norm": 0.5407322645187378,
      "learning_rate": 0.00019768278308061882,
      "loss": 1.184,
      "step": 5130
    },
    {
      "epoch": 0.20826158303113795,
      "grad_norm": 0.5231706500053406,
      "learning_rate": 0.00019767368677165996,
      "loss": 1.1566,
      "step": 5140
    },
    {
      "epoch": 0.20866676120824132,
      "grad_norm": 0.5490206480026245,
      "learning_rate": 0.00019766457285384008,
      "loss": 1.1788,
      "step": 5150
    },
    {
      "epoch": 0.2090719393853447,
      "grad_norm": 0.5038927793502808,
      "learning_rate": 0.0001976554413288023,
      "loss": 1.1582,
      "step": 5160
    },
    {
      "epoch": 0.20947711756244808,
      "grad_norm": 0.6065676212310791,
      "learning_rate": 0.00019764629219819289,
      "loss": 1.1707,
      "step": 5170
    },
    {
      "epoch": 0.20988229573955147,
      "grad_norm": 0.49953189492225647,
      "learning_rate": 0.00019763712546366123,
      "loss": 1.1788,
      "step": 5180
    },
    {
      "epoch": 0.21028747391665484,
      "grad_norm": 0.530707836151123,
      "learning_rate": 0.00019762794112685995,
      "loss": 1.1798,
      "step": 5190
    },
    {
      "epoch": 0.21069265209375823,
      "grad_norm": 0.6046406626701355,
      "learning_rate": 0.00019761873918944483,
      "loss": 1.1934,
      "step": 5200
    },
    {
      "epoch": 0.21109783027086162,
      "grad_norm": 0.5179711580276489,
      "learning_rate": 0.00019760951965307476,
      "loss": 1.1429,
      "step": 5210
    },
    {
      "epoch": 0.211503008447965,
      "grad_norm": 0.47917333245277405,
      "learning_rate": 0.00019760028251941195,
      "loss": 1.1815,
      "step": 5220
    },
    {
      "epoch": 0.21190818662506838,
      "grad_norm": 0.5496241450309753,
      "learning_rate": 0.00019759102779012166,
      "loss": 1.1585,
      "step": 5230
    },
    {
      "epoch": 0.21231336480217175,
      "grad_norm": 0.5546241998672485,
      "learning_rate": 0.00019758175546687236,
      "loss": 1.1863,
      "step": 5240
    },
    {
      "epoch": 0.21271854297927514,
      "grad_norm": 0.5125406384468079,
      "learning_rate": 0.00019757246555133565,
      "loss": 1.1807,
      "step": 5250
    },
    {
      "epoch": 0.2131237211563785,
      "grad_norm": 0.5718168020248413,
      "learning_rate": 0.0001975631580451864,
      "loss": 1.1682,
      "step": 5260
    },
    {
      "epoch": 0.2135288993334819,
      "grad_norm": 0.5660309791564941,
      "learning_rate": 0.00019755383295010253,
      "loss": 1.1967,
      "step": 5270
    },
    {
      "epoch": 0.21393407751058527,
      "grad_norm": 0.7835785150527954,
      "learning_rate": 0.00019754449026776523,
      "loss": 1.1648,
      "step": 5280
    },
    {
      "epoch": 0.21433925568768866,
      "grad_norm": 0.5077794790267944,
      "learning_rate": 0.00019753512999985882,
      "loss": 1.1488,
      "step": 5290
    },
    {
      "epoch": 0.21474443386479206,
      "grad_norm": 0.5517593622207642,
      "learning_rate": 0.00019752575214807076,
      "loss": 1.167,
      "step": 5300
    },
    {
      "epoch": 0.21514961204189542,
      "grad_norm": 0.524542510509491,
      "learning_rate": 0.00019751635671409176,
      "loss": 1.1629,
      "step": 5310
    },
    {
      "epoch": 0.21555479021899882,
      "grad_norm": 0.581014096736908,
      "learning_rate": 0.00019750694369961562,
      "loss": 1.1657,
      "step": 5320
    },
    {
      "epoch": 0.21595996839610218,
      "grad_norm": 0.5856320858001709,
      "learning_rate": 0.00019749751310633938,
      "loss": 1.1667,
      "step": 5330
    },
    {
      "epoch": 0.21636514657320557,
      "grad_norm": 0.5554258823394775,
      "learning_rate": 0.00019748806493596316,
      "loss": 1.1939,
      "step": 5340
    },
    {
      "epoch": 0.21677032475030894,
      "grad_norm": 0.5078227519989014,
      "learning_rate": 0.0001974785991901903,
      "loss": 1.1582,
      "step": 5350
    },
    {
      "epoch": 0.21717550292741233,
      "grad_norm": 0.4366932511329651,
      "learning_rate": 0.0001974691158707274,
      "loss": 1.1663,
      "step": 5360
    },
    {
      "epoch": 0.2175806811045157,
      "grad_norm": 0.5286881923675537,
      "learning_rate": 0.00019745961497928406,
      "loss": 1.1483,
      "step": 5370
    },
    {
      "epoch": 0.2179858592816191,
      "grad_norm": 0.5618920922279358,
      "learning_rate": 0.0001974500965175731,
      "loss": 1.1838,
      "step": 5380
    },
    {
      "epoch": 0.2183910374587225,
      "grad_norm": 0.5668398141860962,
      "learning_rate": 0.0001974405604873106,
      "loss": 1.1658,
      "step": 5390
    },
    {
      "epoch": 0.21879621563582585,
      "grad_norm": 0.63944411277771,
      "learning_rate": 0.0001974310068902157,
      "loss": 1.1123,
      "step": 5400
    },
    {
      "epoch": 0.21920139381292925,
      "grad_norm": 0.5339101552963257,
      "learning_rate": 0.00019742143572801073,
      "loss": 1.187,
      "step": 5410
    },
    {
      "epoch": 0.2196065719900326,
      "grad_norm": 0.5317809581756592,
      "learning_rate": 0.00019741184700242126,
      "loss": 1.2023,
      "step": 5420
    },
    {
      "epoch": 0.220011750167136,
      "grad_norm": 0.5171332955360413,
      "learning_rate": 0.00019740224071517593,
      "loss": 1.18,
      "step": 5430
    },
    {
      "epoch": 0.22041692834423937,
      "grad_norm": 0.48060742020606995,
      "learning_rate": 0.0001973926168680066,
      "loss": 1.1741,
      "step": 5440
    },
    {
      "epoch": 0.22082210652134276,
      "grad_norm": 0.5342585444450378,
      "learning_rate": 0.00019738297546264825,
      "loss": 1.188,
      "step": 5450
    },
    {
      "epoch": 0.22122728469844613,
      "grad_norm": 0.5569594502449036,
      "learning_rate": 0.0001973733165008391,
      "loss": 1.1864,
      "step": 5460
    },
    {
      "epoch": 0.22163246287554952,
      "grad_norm": 0.5001917481422424,
      "learning_rate": 0.0001973636399843205,
      "loss": 1.181,
      "step": 5470
    },
    {
      "epoch": 0.22203764105265292,
      "grad_norm": 0.5047023892402649,
      "learning_rate": 0.0001973539459148369,
      "loss": 1.1433,
      "step": 5480
    },
    {
      "epoch": 0.22244281922975628,
      "grad_norm": 0.5660410523414612,
      "learning_rate": 0.00019734423429413603,
      "loss": 1.1419,
      "step": 5490
    },
    {
      "epoch": 0.22284799740685968,
      "grad_norm": 0.7009663581848145,
      "learning_rate": 0.00019733450512396869,
      "loss": 1.1748,
      "step": 5500
    },
    {
      "epoch": 0.22325317558396304,
      "grad_norm": 0.5346315503120422,
      "learning_rate": 0.00019732475840608888,
      "loss": 1.1925,
      "step": 5510
    },
    {
      "epoch": 0.22365835376106644,
      "grad_norm": 0.465589702129364,
      "learning_rate": 0.00019731499414225382,
      "loss": 1.1588,
      "step": 5520
    },
    {
      "epoch": 0.2240635319381698,
      "grad_norm": 0.4984413981437683,
      "learning_rate": 0.00019730521233422374,
      "loss": 1.167,
      "step": 5530
    },
    {
      "epoch": 0.2244687101152732,
      "grad_norm": 0.5355839133262634,
      "learning_rate": 0.00019729541298376218,
      "loss": 1.1982,
      "step": 5540
    },
    {
      "epoch": 0.22487388829237656,
      "grad_norm": 0.5662782192230225,
      "learning_rate": 0.00019728559609263582,
      "loss": 1.1703,
      "step": 5550
    },
    {
      "epoch": 0.22527906646947995,
      "grad_norm": 0.6540381908416748,
      "learning_rate": 0.00019727576166261442,
      "loss": 1.1824,
      "step": 5560
    },
    {
      "epoch": 0.22568424464658335,
      "grad_norm": 0.6353569030761719,
      "learning_rate": 0.000197265909695471,
      "loss": 1.1618,
      "step": 5570
    },
    {
      "epoch": 0.22608942282368671,
      "grad_norm": 0.5339533686637878,
      "learning_rate": 0.00019725604019298163,
      "loss": 1.1779,
      "step": 5580
    },
    {
      "epoch": 0.2264946010007901,
      "grad_norm": 0.5707919597625732,
      "learning_rate": 0.00019724615315692572,
      "loss": 1.1594,
      "step": 5590
    },
    {
      "epoch": 0.22689977917789347,
      "grad_norm": 0.5856225490570068,
      "learning_rate": 0.00019723624858908562,
      "loss": 1.1745,
      "step": 5600
    },
    {
      "epoch": 0.22730495735499687,
      "grad_norm": 0.5078884959220886,
      "learning_rate": 0.00019722632649124701,
      "loss": 1.1893,
      "step": 5610
    },
    {
      "epoch": 0.22771013553210023,
      "grad_norm": 0.5582584142684937,
      "learning_rate": 0.0001972163868651987,
      "loss": 1.1711,
      "step": 5620
    },
    {
      "epoch": 0.22811531370920363,
      "grad_norm": 0.52281254529953,
      "learning_rate": 0.00019720642971273256,
      "loss": 1.1584,
      "step": 5630
    },
    {
      "epoch": 0.228520491886307,
      "grad_norm": 0.5087818503379822,
      "learning_rate": 0.0001971964550356437,
      "loss": 1.194,
      "step": 5640
    },
    {
      "epoch": 0.22892567006341039,
      "grad_norm": 0.5426724553108215,
      "learning_rate": 0.0001971864628357304,
      "loss": 1.2139,
      "step": 5650
    },
    {
      "epoch": 0.22933084824051378,
      "grad_norm": 0.5169321894645691,
      "learning_rate": 0.0001971764531147941,
      "loss": 1.1915,
      "step": 5660
    },
    {
      "epoch": 0.22973602641761715,
      "grad_norm": 0.5730605721473694,
      "learning_rate": 0.00019716642587463938,
      "loss": 1.1559,
      "step": 5670
    },
    {
      "epoch": 0.23014120459472054,
      "grad_norm": 0.5308975577354431,
      "learning_rate": 0.00019715638111707394,
      "loss": 1.1621,
      "step": 5680
    },
    {
      "epoch": 0.2305463827718239,
      "grad_norm": 0.5564334392547607,
      "learning_rate": 0.00019714631884390867,
      "loss": 1.2048,
      "step": 5690
    },
    {
      "epoch": 0.2309515609489273,
      "grad_norm": 0.5901036858558655,
      "learning_rate": 0.00019713623905695768,
      "loss": 1.198,
      "step": 5700
    },
    {
      "epoch": 0.23135673912603066,
      "grad_norm": 0.5357282161712646,
      "learning_rate": 0.0001971261417580381,
      "loss": 1.1543,
      "step": 5710
    },
    {
      "epoch": 0.23176191730313406,
      "grad_norm": 0.5604302287101746,
      "learning_rate": 0.00019711602694897037,
      "loss": 1.1837,
      "step": 5720
    },
    {
      "epoch": 0.23216709548023742,
      "grad_norm": 0.562340497970581,
      "learning_rate": 0.00019710589463157798,
      "loss": 1.2233,
      "step": 5730
    },
    {
      "epoch": 0.23257227365734082,
      "grad_norm": 0.5671872496604919,
      "learning_rate": 0.0001970957448076876,
      "loss": 1.2105,
      "step": 5740
    },
    {
      "epoch": 0.2329774518344442,
      "grad_norm": 0.6275280117988586,
      "learning_rate": 0.00019708557747912905,
      "loss": 1.1666,
      "step": 5750
    },
    {
      "epoch": 0.23338263001154758,
      "grad_norm": 0.49065765738487244,
      "learning_rate": 0.00019707539264773538,
      "loss": 1.1439,
      "step": 5760
    },
    {
      "epoch": 0.23378780818865097,
      "grad_norm": 0.6018632650375366,
      "learning_rate": 0.0001970651903153427,
      "loss": 1.1764,
      "step": 5770
    },
    {
      "epoch": 0.23419298636575434,
      "grad_norm": 0.5387911796569824,
      "learning_rate": 0.00019705497048379034,
      "loss": 1.1749,
      "step": 5780
    },
    {
      "epoch": 0.23459816454285773,
      "grad_norm": 0.5163089632987976,
      "learning_rate": 0.0001970447331549207,
      "loss": 1.1792,
      "step": 5790
    },
    {
      "epoch": 0.2350033427199611,
      "grad_norm": 0.5371166467666626,
      "learning_rate": 0.00019703447833057948,
      "loss": 1.2057,
      "step": 5800
    },
    {
      "epoch": 0.2354085208970645,
      "grad_norm": 0.5619894862174988,
      "learning_rate": 0.00019702420601261535,
      "loss": 1.1441,
      "step": 5810
    },
    {
      "epoch": 0.23581369907416785,
      "grad_norm": 0.5829156041145325,
      "learning_rate": 0.0001970139162028803,
      "loss": 1.1658,
      "step": 5820
    },
    {
      "epoch": 0.23621887725127125,
      "grad_norm": 0.5374458432197571,
      "learning_rate": 0.00019700360890322936,
      "loss": 1.1766,
      "step": 5830
    },
    {
      "epoch": 0.23662405542837464,
      "grad_norm": 0.5131230354309082,
      "learning_rate": 0.00019699328411552079,
      "loss": 1.1759,
      "step": 5840
    },
    {
      "epoch": 0.237029233605478,
      "grad_norm": 0.5547551512718201,
      "learning_rate": 0.0001969829418416159,
      "loss": 1.1648,
      "step": 5850
    },
    {
      "epoch": 0.2374344117825814,
      "grad_norm": 0.5594378113746643,
      "learning_rate": 0.00019697258208337934,
      "loss": 1.1545,
      "step": 5860
    },
    {
      "epoch": 0.23783958995968477,
      "grad_norm": 0.6086752414703369,
      "learning_rate": 0.0001969622048426787,
      "loss": 1.1765,
      "step": 5870
    },
    {
      "epoch": 0.23824476813678816,
      "grad_norm": 0.49191781878471375,
      "learning_rate": 0.00019695181012138483,
      "loss": 1.1758,
      "step": 5880
    },
    {
      "epoch": 0.23864994631389153,
      "grad_norm": 0.5813024044036865,
      "learning_rate": 0.00019694139792137175,
      "loss": 1.1777,
      "step": 5890
    },
    {
      "epoch": 0.23905512449099492,
      "grad_norm": 0.539185106754303,
      "learning_rate": 0.00019693096824451657,
      "loss": 1.1602,
      "step": 5900
    },
    {
      "epoch": 0.23946030266809828,
      "grad_norm": 0.5935100317001343,
      "learning_rate": 0.0001969205210926996,
      "loss": 1.1571,
      "step": 5910
    },
    {
      "epoch": 0.23986548084520168,
      "grad_norm": 0.5834608674049377,
      "learning_rate": 0.00019691005646780424,
      "loss": 1.1737,
      "step": 5920
    },
    {
      "epoch": 0.24027065902230507,
      "grad_norm": 0.5882130265235901,
      "learning_rate": 0.0001968995743717171,
      "loss": 1.1569,
      "step": 5930
    },
    {
      "epoch": 0.24067583719940844,
      "grad_norm": 0.5095469355583191,
      "learning_rate": 0.00019688907480632794,
      "loss": 1.1658,
      "step": 5940
    },
    {
      "epoch": 0.24108101537651183,
      "grad_norm": 0.580394446849823,
      "learning_rate": 0.00019687855777352967,
      "loss": 1.1819,
      "step": 5950
    },
    {
      "epoch": 0.2414861935536152,
      "grad_norm": 0.5318992733955383,
      "learning_rate": 0.00019686802327521823,
      "loss": 1.183,
      "step": 5960
    },
    {
      "epoch": 0.2418913717307186,
      "grad_norm": 0.5144644379615784,
      "learning_rate": 0.00019685747131329288,
      "loss": 1.1146,
      "step": 5970
    },
    {
      "epoch": 0.24229654990782196,
      "grad_norm": 0.547622561454773,
      "learning_rate": 0.00019684690188965595,
      "loss": 1.1576,
      "step": 5980
    },
    {
      "epoch": 0.24270172808492535,
      "grad_norm": 0.5114852786064148,
      "learning_rate": 0.0001968363150062129,
      "loss": 1.1918,
      "step": 5990
    },
    {
      "epoch": 0.24310690626202872,
      "grad_norm": 0.5906286239624023,
      "learning_rate": 0.0001968257106648724,
      "loss": 1.1398,
      "step": 6000
    },
    {
      "epoch": 0.2435120844391321,
      "grad_norm": 0.6370722055435181,
      "learning_rate": 0.0001968150888675462,
      "loss": 1.1958,
      "step": 6010
    },
    {
      "epoch": 0.2439172626162355,
      "grad_norm": 0.561944305896759,
      "learning_rate": 0.00019680444961614918,
      "loss": 1.1586,
      "step": 6020
    },
    {
      "epoch": 0.24432244079333887,
      "grad_norm": 0.5237104296684265,
      "learning_rate": 0.0001967937929125995,
      "loss": 1.1226,
      "step": 6030
    },
    {
      "epoch": 0.24472761897044226,
      "grad_norm": 0.6933553218841553,
      "learning_rate": 0.0001967831187588183,
      "loss": 1.1908,
      "step": 6040
    },
    {
      "epoch": 0.24513279714754563,
      "grad_norm": 0.5669260621070862,
      "learning_rate": 0.00019677242715673002,
      "loss": 1.1896,
      "step": 6050
    },
    {
      "epoch": 0.24553797532464902,
      "grad_norm": 0.5470685362815857,
      "learning_rate": 0.0001967617181082621,
      "loss": 1.1937,
      "step": 6060
    },
    {
      "epoch": 0.2459431535017524,
      "grad_norm": 0.590371310710907,
      "learning_rate": 0.00019675099161534521,
      "loss": 1.1913,
      "step": 6070
    },
    {
      "epoch": 0.24634833167885578,
      "grad_norm": 0.7251625061035156,
      "learning_rate": 0.0001967402476799132,
      "loss": 1.1965,
      "step": 6080
    },
    {
      "epoch": 0.24675350985595915,
      "grad_norm": 0.528200626373291,
      "learning_rate": 0.00019672948630390294,
      "loss": 1.1484,
      "step": 6090
    },
    {
      "epoch": 0.24715868803306254,
      "grad_norm": 0.517109751701355,
      "learning_rate": 0.0001967187074892546,
      "loss": 1.1796,
      "step": 6100
    },
    {
      "epoch": 0.24756386621016593,
      "grad_norm": 0.5118247270584106,
      "learning_rate": 0.0001967079112379113,
      "loss": 1.1698,
      "step": 6110
    },
    {
      "epoch": 0.2479690443872693,
      "grad_norm": 0.595554769039154,
      "learning_rate": 0.00019669709755181953,
      "loss": 1.1481,
      "step": 6120
    },
    {
      "epoch": 0.2483742225643727,
      "grad_norm": 0.5335313677787781,
      "learning_rate": 0.00019668626643292876,
      "loss": 1.1732,
      "step": 6130
    },
    {
      "epoch": 0.24877940074147606,
      "grad_norm": 0.6014825701713562,
      "learning_rate": 0.00019667541788319162,
      "loss": 1.1106,
      "step": 6140
    },
    {
      "epoch": 0.24918457891857945,
      "grad_norm": 0.5055657029151917,
      "learning_rate": 0.00019666455190456397,
      "loss": 1.1877,
      "step": 6150
    },
    {
      "epoch": 0.24958975709568282,
      "grad_norm": 0.5903727412223816,
      "learning_rate": 0.00019665366849900472,
      "loss": 1.1922,
      "step": 6160
    },
    {
      "epoch": 0.2499949352727862,
      "grad_norm": 0.525941014289856,
      "learning_rate": 0.00019664276766847598,
      "loss": 1.1531,
      "step": 6170
    },
    {
      "epoch": 0.2504001134498896,
      "grad_norm": 0.5801976323127747,
      "learning_rate": 0.00019663184941494296,
      "loss": 1.1847,
      "step": 6180
    },
    {
      "epoch": 0.25080529162699294,
      "grad_norm": 0.5361466407775879,
      "learning_rate": 0.00019662091374037405,
      "loss": 1.1683,
      "step": 6190
    },
    {
      "epoch": 0.25121046980409634,
      "grad_norm": 0.48347002267837524,
      "learning_rate": 0.00019660996064674075,
      "loss": 1.183,
      "step": 6200
    },
    {
      "epoch": 0.25161564798119973,
      "grad_norm": 0.5880921483039856,
      "learning_rate": 0.00019659899013601772,
      "loss": 1.223,
      "step": 6210
    },
    {
      "epoch": 0.2520208261583031,
      "grad_norm": 0.5168225169181824,
      "learning_rate": 0.00019658800221018273,
      "loss": 1.1822,
      "step": 6220
    },
    {
      "epoch": 0.2524260043354065,
      "grad_norm": 0.5005984902381897,
      "learning_rate": 0.00019657699687121674,
      "loss": 1.1648,
      "step": 6230
    },
    {
      "epoch": 0.25283118251250986,
      "grad_norm": 0.5291897654533386,
      "learning_rate": 0.0001965659741211038,
      "loss": 1.1729,
      "step": 6240
    },
    {
      "epoch": 0.25323636068961325,
      "grad_norm": 0.527432382106781,
      "learning_rate": 0.00019655493396183112,
      "loss": 1.1758,
      "step": 6250
    },
    {
      "epoch": 0.25364153886671664,
      "grad_norm": 0.5146752595901489,
      "learning_rate": 0.00019654387639538907,
      "loss": 1.1829,
      "step": 6260
    },
    {
      "epoch": 0.25404671704382004,
      "grad_norm": 0.5501819252967834,
      "learning_rate": 0.00019653280142377112,
      "loss": 1.1735,
      "step": 6270
    },
    {
      "epoch": 0.2544518952209234,
      "grad_norm": 0.4999786913394928,
      "learning_rate": 0.00019652170904897387,
      "loss": 1.2064,
      "step": 6280
    },
    {
      "epoch": 0.25485707339802677,
      "grad_norm": 0.5666659474372864,
      "learning_rate": 0.0001965105992729971,
      "loss": 1.1722,
      "step": 6290
    },
    {
      "epoch": 0.25526225157513016,
      "grad_norm": 0.5590999126434326,
      "learning_rate": 0.00019649947209784375,
      "loss": 1.1529,
      "step": 6300
    },
    {
      "epoch": 0.25566742975223355,
      "grad_norm": 0.5721120238304138,
      "learning_rate": 0.00019648832752551981,
      "loss": 1.1569,
      "step": 6310
    },
    {
      "epoch": 0.25607260792933695,
      "grad_norm": 0.5910996794700623,
      "learning_rate": 0.00019647716555803445,
      "loss": 1.1631,
      "step": 6320
    },
    {
      "epoch": 0.2564777861064403,
      "grad_norm": 0.5812075138092041,
      "learning_rate": 0.0001964659861974,
      "loss": 1.1545,
      "step": 6330
    },
    {
      "epoch": 0.2568829642835437,
      "grad_norm": 0.5094364285469055,
      "learning_rate": 0.0001964547894456319,
      "loss": 1.1488,
      "step": 6340
    },
    {
      "epoch": 0.2572881424606471,
      "grad_norm": 0.522078812122345,
      "learning_rate": 0.00019644357530474872,
      "loss": 1.1736,
      "step": 6350
    },
    {
      "epoch": 0.25769332063775047,
      "grad_norm": 0.5267487168312073,
      "learning_rate": 0.00019643234377677218,
      "loss": 1.1648,
      "step": 6360
    },
    {
      "epoch": 0.2580984988148538,
      "grad_norm": 0.5658043026924133,
      "learning_rate": 0.0001964210948637271,
      "loss": 1.1794,
      "step": 6370
    },
    {
      "epoch": 0.2585036769919572,
      "grad_norm": 0.5856928825378418,
      "learning_rate": 0.0001964098285676415,
      "loss": 1.1763,
      "step": 6380
    },
    {
      "epoch": 0.2589088551690606,
      "grad_norm": 0.5380805730819702,
      "learning_rate": 0.0001963985448905465,
      "loss": 1.1871,
      "step": 6390
    },
    {
      "epoch": 0.259314033346164,
      "grad_norm": 0.6512140035629272,
      "learning_rate": 0.00019638724383447636,
      "loss": 1.1544,
      "step": 6400
    },
    {
      "epoch": 0.2597192115232674,
      "grad_norm": 0.5136595964431763,
      "learning_rate": 0.00019637592540146836,
      "loss": 1.1453,
      "step": 6410
    },
    {
      "epoch": 0.2601243897003707,
      "grad_norm": 0.5320270657539368,
      "learning_rate": 0.00019636458959356316,
      "loss": 1.142,
      "step": 6420
    },
    {
      "epoch": 0.2605295678774741,
      "grad_norm": 0.6904148459434509,
      "learning_rate": 0.0001963532364128043,
      "loss": 1.1954,
      "step": 6430
    },
    {
      "epoch": 0.2609347460545775,
      "grad_norm": 0.5379882454872131,
      "learning_rate": 0.00019634186586123863,
      "loss": 1.1713,
      "step": 6440
    },
    {
      "epoch": 0.2613399242316809,
      "grad_norm": 0.8108576536178589,
      "learning_rate": 0.00019633047794091602,
      "loss": 1.1723,
      "step": 6450
    },
    {
      "epoch": 0.26174510240878424,
      "grad_norm": 0.565364420413971,
      "learning_rate": 0.00019631907265388953,
      "loss": 1.1832,
      "step": 6460
    },
    {
      "epoch": 0.26215028058588763,
      "grad_norm": 0.6023277640342712,
      "learning_rate": 0.00019630765000221533,
      "loss": 1.1665,
      "step": 6470
    },
    {
      "epoch": 0.262555458762991,
      "grad_norm": 0.5886109471321106,
      "learning_rate": 0.00019629620998795275,
      "loss": 1.147,
      "step": 6480
    },
    {
      "epoch": 0.2629606369400944,
      "grad_norm": 0.5822970867156982,
      "learning_rate": 0.00019628475261316417,
      "loss": 1.1689,
      "step": 6490
    },
    {
      "epoch": 0.2633658151171978,
      "grad_norm": 0.48983490467071533,
      "learning_rate": 0.0001962732778799152,
      "loss": 1.1849,
      "step": 6500
    },
    {
      "epoch": 0.26377099329430115,
      "grad_norm": 0.5303608775138855,
      "learning_rate": 0.00019626178579027452,
      "loss": 1.1993,
      "step": 6510
    },
    {
      "epoch": 0.26417617147140454,
      "grad_norm": 0.5375282168388367,
      "learning_rate": 0.00019625027634631394,
      "loss": 1.1732,
      "step": 6520
    },
    {
      "epoch": 0.26458134964850794,
      "grad_norm": 0.5944948792457581,
      "learning_rate": 0.00019623874955010842,
      "loss": 1.1957,
      "step": 6530
    },
    {
      "epoch": 0.26498652782561133,
      "grad_norm": 0.5414245128631592,
      "learning_rate": 0.00019622720540373608,
      "loss": 1.1229,
      "step": 6540
    },
    {
      "epoch": 0.26539170600271467,
      "grad_norm": 0.5423265695571899,
      "learning_rate": 0.00019621564390927807,
      "loss": 1.1593,
      "step": 6550
    },
    {
      "epoch": 0.26579688417981806,
      "grad_norm": 0.5683432221412659,
      "learning_rate": 0.00019620406506881875,
      "loss": 1.1857,
      "step": 6560
    },
    {
      "epoch": 0.26620206235692145,
      "grad_norm": 0.5435795187950134,
      "learning_rate": 0.0001961924688844456,
      "loss": 1.1515,
      "step": 6570
    },
    {
      "epoch": 0.26660724053402485,
      "grad_norm": 0.5660802721977234,
      "learning_rate": 0.00019618085535824914,
      "loss": 1.1722,
      "step": 6580
    },
    {
      "epoch": 0.26701241871112824,
      "grad_norm": 0.6081596612930298,
      "learning_rate": 0.0001961692244923232,
      "loss": 1.2189,
      "step": 6590
    },
    {
      "epoch": 0.2674175968882316,
      "grad_norm": 0.5070329904556274,
      "learning_rate": 0.00019615757628876454,
      "loss": 1.1489,
      "step": 6600
    },
    {
      "epoch": 0.267822775065335,
      "grad_norm": 0.5850003957748413,
      "learning_rate": 0.00019614591074967313,
      "loss": 1.1601,
      "step": 6610
    },
    {
      "epoch": 0.26822795324243837,
      "grad_norm": 0.5109196305274963,
      "learning_rate": 0.0001961342278771521,
      "loss": 1.1817,
      "step": 6620
    },
    {
      "epoch": 0.26863313141954176,
      "grad_norm": 0.548793613910675,
      "learning_rate": 0.00019612252767330763,
      "loss": 1.1806,
      "step": 6630
    },
    {
      "epoch": 0.2690383095966451,
      "grad_norm": 0.5543352365493774,
      "learning_rate": 0.00019611081014024907,
      "loss": 1.1902,
      "step": 6640
    },
    {
      "epoch": 0.2694434877737485,
      "grad_norm": 0.5291150212287903,
      "learning_rate": 0.00019609907528008895,
      "loss": 1.1802,
      "step": 6650
    },
    {
      "epoch": 0.2698486659508519,
      "grad_norm": 0.5004898309707642,
      "learning_rate": 0.00019608732309494276,
      "loss": 1.1652,
      "step": 6660
    },
    {
      "epoch": 0.2702538441279553,
      "grad_norm": 0.5950507521629333,
      "learning_rate": 0.0001960755535869293,
      "loss": 1.1697,
      "step": 6670
    },
    {
      "epoch": 0.27065902230505867,
      "grad_norm": 0.525779664516449,
      "learning_rate": 0.00019606376675817034,
      "loss": 1.1449,
      "step": 6680
    },
    {
      "epoch": 0.271064200482162,
      "grad_norm": 0.5076351165771484,
      "learning_rate": 0.0001960519626107909,
      "loss": 1.1417,
      "step": 6690
    },
    {
      "epoch": 0.2714693786592654,
      "grad_norm": 0.5348721742630005,
      "learning_rate": 0.000196040141146919,
      "loss": 1.1694,
      "step": 6700
    },
    {
      "epoch": 0.2718745568363688,
      "grad_norm": 0.5014850497245789,
      "learning_rate": 0.0001960283023686859,
      "loss": 1.1733,
      "step": 6710
    },
    {
      "epoch": 0.2722797350134722,
      "grad_norm": 0.5808080434799194,
      "learning_rate": 0.00019601644627822586,
      "loss": 1.1564,
      "step": 6720
    },
    {
      "epoch": 0.27268491319057553,
      "grad_norm": 0.5368320941925049,
      "learning_rate": 0.00019600457287767637,
      "loss": 1.1636,
      "step": 6730
    },
    {
      "epoch": 0.2730900913676789,
      "grad_norm": 0.5455905199050903,
      "learning_rate": 0.000195992682169178,
      "loss": 1.1605,
      "step": 6740
    },
    {
      "epoch": 0.2734952695447823,
      "grad_norm": 0.5441325902938843,
      "learning_rate": 0.00019598077415487443,
      "loss": 1.1472,
      "step": 6750
    },
    {
      "epoch": 0.2739004477218857,
      "grad_norm": 0.5865528583526611,
      "learning_rate": 0.0001959688488369125,
      "loss": 1.1774,
      "step": 6760
    },
    {
      "epoch": 0.2743056258989891,
      "grad_norm": 0.5334582328796387,
      "learning_rate": 0.00019595690621744208,
      "loss": 1.1821,
      "step": 6770
    },
    {
      "epoch": 0.27471080407609244,
      "grad_norm": 0.5589249730110168,
      "learning_rate": 0.00019594494629861621,
      "loss": 1.183,
      "step": 6780
    },
    {
      "epoch": 0.27511598225319583,
      "grad_norm": 0.5807943940162659,
      "learning_rate": 0.00019593296908259114,
      "loss": 1.1783,
      "step": 6790
    },
    {
      "epoch": 0.27552116043029923,
      "grad_norm": 0.5728068947792053,
      "learning_rate": 0.00019592097457152608,
      "loss": 1.2051,
      "step": 6800
    },
    {
      "epoch": 0.2759263386074026,
      "grad_norm": 0.5463709235191345,
      "learning_rate": 0.0001959089627675834,
      "loss": 1.1727,
      "step": 6810
    },
    {
      "epoch": 0.27633151678450596,
      "grad_norm": 0.5451804995536804,
      "learning_rate": 0.0001958969336729287,
      "loss": 1.2047,
      "step": 6820
    },
    {
      "epoch": 0.27673669496160935,
      "grad_norm": 0.5473561882972717,
      "learning_rate": 0.00019588488728973058,
      "loss": 1.1835,
      "step": 6830
    },
    {
      "epoch": 0.27714187313871275,
      "grad_norm": 0.5013165473937988,
      "learning_rate": 0.00019587282362016083,
      "loss": 1.1794,
      "step": 6840
    },
    {
      "epoch": 0.27754705131581614,
      "grad_norm": 0.5587889552116394,
      "learning_rate": 0.00019586074266639421,
      "loss": 1.1833,
      "step": 6850
    },
    {
      "epoch": 0.27795222949291953,
      "grad_norm": 0.48778119683265686,
      "learning_rate": 0.00019584864443060885,
      "loss": 1.1592,
      "step": 6860
    },
    {
      "epoch": 0.27835740767002287,
      "grad_norm": 0.5364330410957336,
      "learning_rate": 0.00019583652891498577,
      "loss": 1.1706,
      "step": 6870
    },
    {
      "epoch": 0.27876258584712627,
      "grad_norm": 0.5198081135749817,
      "learning_rate": 0.00019582439612170915,
      "loss": 1.1355,
      "step": 6880
    },
    {
      "epoch": 0.27916776402422966,
      "grad_norm": 0.5750464797019958,
      "learning_rate": 0.0001958122460529664,
      "loss": 1.1709,
      "step": 6890
    },
    {
      "epoch": 0.27957294220133305,
      "grad_norm": 0.5297413468360901,
      "learning_rate": 0.00019580007871094793,
      "loss": 1.1857,
      "step": 6900
    },
    {
      "epoch": 0.2799781203784364,
      "grad_norm": 0.5634208917617798,
      "learning_rate": 0.00019578789409784727,
      "loss": 1.214,
      "step": 6910
    },
    {
      "epoch": 0.2803832985555398,
      "grad_norm": 0.5760539174079895,
      "learning_rate": 0.00019577569221586116,
      "loss": 1.1954,
      "step": 6920
    },
    {
      "epoch": 0.2807884767326432,
      "grad_norm": 0.5531413555145264,
      "learning_rate": 0.00019576347306718935,
      "loss": 1.1488,
      "step": 6930
    },
    {
      "epoch": 0.28119365490974657,
      "grad_norm": 0.4932734966278076,
      "learning_rate": 0.00019575123665403472,
      "loss": 1.1855,
      "step": 6940
    },
    {
      "epoch": 0.28159883308684996,
      "grad_norm": 0.5545091032981873,
      "learning_rate": 0.00019573898297860332,
      "loss": 1.1607,
      "step": 6950
    },
    {
      "epoch": 0.2820040112639533,
      "grad_norm": 0.5413622260093689,
      "learning_rate": 0.00019572671204310424,
      "loss": 1.1489,
      "step": 6960
    },
    {
      "epoch": 0.2824091894410567,
      "grad_norm": 0.5722427368164062,
      "learning_rate": 0.00019571442384974975,
      "loss": 1.1452,
      "step": 6970
    },
    {
      "epoch": 0.2828143676181601,
      "grad_norm": 0.5612212419509888,
      "learning_rate": 0.00019570211840075517,
      "loss": 1.1921,
      "step": 6980
    },
    {
      "epoch": 0.2832195457952635,
      "grad_norm": 0.5440688133239746,
      "learning_rate": 0.00019568979569833895,
      "loss": 1.1612,
      "step": 6990
    },
    {
      "epoch": 0.2836247239723668,
      "grad_norm": 0.6112028360366821,
      "learning_rate": 0.0001956774557447227,
      "loss": 1.1881,
      "step": 7000
    },
    {
      "epoch": 0.2840299021494702,
      "grad_norm": 0.5216151475906372,
      "learning_rate": 0.00019566509854213103,
      "loss": 1.1962,
      "step": 7010
    },
    {
      "epoch": 0.2844350803265736,
      "grad_norm": 0.5488293170928955,
      "learning_rate": 0.00019565272409279182,
      "loss": 1.1819,
      "step": 7020
    },
    {
      "epoch": 0.284840258503677,
      "grad_norm": 0.5156698226928711,
      "learning_rate": 0.0001956403323989359,
      "loss": 1.1915,
      "step": 7030
    },
    {
      "epoch": 0.2852454366807804,
      "grad_norm": 0.5855851769447327,
      "learning_rate": 0.00019562792346279728,
      "loss": 1.2187,
      "step": 7040
    },
    {
      "epoch": 0.28565061485788373,
      "grad_norm": 0.546122670173645,
      "learning_rate": 0.0001956154972866131,
      "loss": 1.1985,
      "step": 7050
    },
    {
      "epoch": 0.2860557930349871,
      "grad_norm": 0.540699303150177,
      "learning_rate": 0.00019560305387262357,
      "loss": 1.1731,
      "step": 7060
    },
    {
      "epoch": 0.2864609712120905,
      "grad_norm": 0.5138152241706848,
      "learning_rate": 0.000195590593223072,
      "loss": 1.126,
      "step": 7070
    },
    {
      "epoch": 0.2868661493891939,
      "grad_norm": 0.6853144764900208,
      "learning_rate": 0.00019557811534020488,
      "loss": 1.1928,
      "step": 7080
    },
    {
      "epoch": 0.28727132756629725,
      "grad_norm": 0.4901999533176422,
      "learning_rate": 0.0001955656202262717,
      "loss": 1.1578,
      "step": 7090
    },
    {
      "epoch": 0.28767650574340065,
      "grad_norm": 0.5817059278488159,
      "learning_rate": 0.00019555310788352514,
      "loss": 1.1713,
      "step": 7100
    },
    {
      "epoch": 0.28808168392050404,
      "grad_norm": 0.550896167755127,
      "learning_rate": 0.00019554057831422096,
      "loss": 1.1534,
      "step": 7110
    },
    {
      "epoch": 0.28848686209760743,
      "grad_norm": 0.5452569723129272,
      "learning_rate": 0.000195528031520618,
      "loss": 1.1758,
      "step": 7120
    },
    {
      "epoch": 0.2888920402747108,
      "grad_norm": 0.5463007092475891,
      "learning_rate": 0.00019551546750497828,
      "loss": 1.173,
      "step": 7130
    },
    {
      "epoch": 0.28929721845181416,
      "grad_norm": 0.5753864049911499,
      "learning_rate": 0.00019550288626956678,
      "loss": 1.1699,
      "step": 7140
    },
    {
      "epoch": 0.28970239662891756,
      "grad_norm": 0.5803713202476501,
      "learning_rate": 0.00019549028781665176,
      "loss": 1.1745,
      "step": 7150
    },
    {
      "epoch": 0.29010757480602095,
      "grad_norm": 0.5798937678337097,
      "learning_rate": 0.00019547767214850444,
      "loss": 1.1433,
      "step": 7160
    },
    {
      "epoch": 0.29051275298312434,
      "grad_norm": 0.5648534893989563,
      "learning_rate": 0.0001954650392673993,
      "loss": 1.169,
      "step": 7170
    },
    {
      "epoch": 0.2909179311602277,
      "grad_norm": 0.5847086310386658,
      "learning_rate": 0.0001954523891756137,
      "loss": 1.1731,
      "step": 7180
    },
    {
      "epoch": 0.2913231093373311,
      "grad_norm": 0.5477168560028076,
      "learning_rate": 0.00019543972187542833,
      "loss": 1.1574,
      "step": 7190
    },
    {
      "epoch": 0.29172828751443447,
      "grad_norm": 0.5235229134559631,
      "learning_rate": 0.00019542703736912686,
      "loss": 1.1766,
      "step": 7200
    },
    {
      "epoch": 0.29213346569153786,
      "grad_norm": 0.540350079536438,
      "learning_rate": 0.00019541433565899605,
      "loss": 1.1625,
      "step": 7210
    },
    {
      "epoch": 0.29253864386864126,
      "grad_norm": 0.5294696688652039,
      "learning_rate": 0.00019540161674732584,
      "loss": 1.1588,
      "step": 7220
    },
    {
      "epoch": 0.2929438220457446,
      "grad_norm": 0.5891201496124268,
      "learning_rate": 0.0001953888806364092,
      "loss": 1.1834,
      "step": 7230
    },
    {
      "epoch": 0.293349000222848,
      "grad_norm": 0.5499357581138611,
      "learning_rate": 0.00019537612732854225,
      "loss": 1.1733,
      "step": 7240
    },
    {
      "epoch": 0.2937541783999514,
      "grad_norm": 0.5249863862991333,
      "learning_rate": 0.00019536335682602415,
      "loss": 1.1726,
      "step": 7250
    },
    {
      "epoch": 0.2941593565770548,
      "grad_norm": 0.5570205450057983,
      "learning_rate": 0.00019535056913115725,
      "loss": 1.1675,
      "step": 7260
    },
    {
      "epoch": 0.2945645347541581,
      "grad_norm": 0.8068438172340393,
      "learning_rate": 0.00019533776424624692,
      "loss": 1.1811,
      "step": 7270
    },
    {
      "epoch": 0.2949697129312615,
      "grad_norm": 0.5697993040084839,
      "learning_rate": 0.00019532494217360164,
      "loss": 1.1816,
      "step": 7280
    },
    {
      "epoch": 0.2953748911083649,
      "grad_norm": 0.5398805737495422,
      "learning_rate": 0.00019531210291553303,
      "loss": 1.1292,
      "step": 7290
    },
    {
      "epoch": 0.2957800692854683,
      "grad_norm": 0.490166038274765,
      "learning_rate": 0.0001952992464743558,
      "loss": 1.1657,
      "step": 7300
    },
    {
      "epoch": 0.2961852474625717,
      "grad_norm": 0.6137372255325317,
      "learning_rate": 0.00019528637285238772,
      "loss": 1.1759,
      "step": 7310
    },
    {
      "epoch": 0.296590425639675,
      "grad_norm": 0.5205552577972412,
      "learning_rate": 0.00019527348205194967,
      "loss": 1.1552,
      "step": 7320
    },
    {
      "epoch": 0.2969956038167784,
      "grad_norm": 0.5713649392127991,
      "learning_rate": 0.00019526057407536564,
      "loss": 1.154,
      "step": 7330
    },
    {
      "epoch": 0.2974007819938818,
      "grad_norm": 0.5888938903808594,
      "learning_rate": 0.00019524764892496274,
      "loss": 1.174,
      "step": 7340
    },
    {
      "epoch": 0.2978059601709852,
      "grad_norm": 0.5628374814987183,
      "learning_rate": 0.00019523470660307113,
      "loss": 1.1822,
      "step": 7350
    },
    {
      "epoch": 0.29821113834808854,
      "grad_norm": 0.5364239811897278,
      "learning_rate": 0.00019522174711202408,
      "loss": 1.1717,
      "step": 7360
    },
    {
      "epoch": 0.29861631652519194,
      "grad_norm": 0.5376259684562683,
      "learning_rate": 0.00019520877045415796,
      "loss": 1.1299,
      "step": 7370
    },
    {
      "epoch": 0.29902149470229533,
      "grad_norm": 0.5214993953704834,
      "learning_rate": 0.00019519577663181222,
      "loss": 1.1506,
      "step": 7380
    },
    {
      "epoch": 0.2994266728793987,
      "grad_norm": 0.572052001953125,
      "learning_rate": 0.00019518276564732946,
      "loss": 1.1356,
      "step": 7390
    },
    {
      "epoch": 0.2998318510565021,
      "grad_norm": 0.6231511235237122,
      "learning_rate": 0.00019516973750305532,
      "loss": 1.146,
      "step": 7400
    },
    {
      "epoch": 0.30023702923360546,
      "grad_norm": 0.5484368205070496,
      "learning_rate": 0.00019515669220133855,
      "loss": 1.1954,
      "step": 7410
    },
    {
      "epoch": 0.30064220741070885,
      "grad_norm": 0.6126089096069336,
      "learning_rate": 0.00019514362974453096,
      "loss": 1.1683,
      "step": 7420
    },
    {
      "epoch": 0.30104738558781224,
      "grad_norm": 0.5776017904281616,
      "learning_rate": 0.0001951305501349875,
      "loss": 1.1679,
      "step": 7430
    },
    {
      "epoch": 0.30145256376491564,
      "grad_norm": 0.6506449580192566,
      "learning_rate": 0.0001951174533750662,
      "loss": 1.1584,
      "step": 7440
    },
    {
      "epoch": 0.301857741942019,
      "grad_norm": 0.5610262155532837,
      "learning_rate": 0.0001951043394671282,
      "loss": 1.1831,
      "step": 7450
    },
    {
      "epoch": 0.30226292011912237,
      "grad_norm": 0.5503144860267639,
      "learning_rate": 0.00019509120841353764,
      "loss": 1.2032,
      "step": 7460
    },
    {
      "epoch": 0.30266809829622576,
      "grad_norm": 0.5817610621452332,
      "learning_rate": 0.00019507806021666188,
      "loss": 1.1739,
      "step": 7470
    },
    {
      "epoch": 0.30307327647332916,
      "grad_norm": 0.5607344508171082,
      "learning_rate": 0.00019506489487887128,
      "loss": 1.2177,
      "step": 7480
    },
    {
      "epoch": 0.30347845465043255,
      "grad_norm": 0.5275853872299194,
      "learning_rate": 0.00019505171240253937,
      "loss": 1.1804,
      "step": 7490
    },
    {
      "epoch": 0.3038836328275359,
      "grad_norm": 0.5283041596412659,
      "learning_rate": 0.00019503851279004266,
      "loss": 1.1359,
      "step": 7500
    },
    {
      "epoch": 0.3042888110046393,
      "grad_norm": 0.4863729476928711,
      "learning_rate": 0.00019502529604376086,
      "loss": 1.1556,
      "step": 7510
    },
    {
      "epoch": 0.3046939891817427,
      "grad_norm": 0.5005844831466675,
      "learning_rate": 0.00019501206216607667,
      "loss": 1.1628,
      "step": 7520
    },
    {
      "epoch": 0.30509916735884607,
      "grad_norm": 0.5271387100219727,
      "learning_rate": 0.00019499881115937596,
      "loss": 1.1644,
      "step": 7530
    },
    {
      "epoch": 0.3055043455359494,
      "grad_norm": 0.5330926775932312,
      "learning_rate": 0.00019498554302604766,
      "loss": 1.1573,
      "step": 7540
    },
    {
      "epoch": 0.3059095237130528,
      "grad_norm": 0.5666901469230652,
      "learning_rate": 0.00019497225776848374,
      "loss": 1.1897,
      "step": 7550
    },
    {
      "epoch": 0.3063147018901562,
      "grad_norm": 0.5858877301216125,
      "learning_rate": 0.00019495895538907933,
      "loss": 1.1597,
      "step": 7560
    },
    {
      "epoch": 0.3067198800672596,
      "grad_norm": 0.531516432762146,
      "learning_rate": 0.00019494563589023264,
      "loss": 1.1676,
      "step": 7570
    },
    {
      "epoch": 0.307125058244363,
      "grad_norm": 0.5720564126968384,
      "learning_rate": 0.0001949322992743449,
      "loss": 1.1347,
      "step": 7580
    },
    {
      "epoch": 0.3075302364214663,
      "grad_norm": 0.5844447016716003,
      "learning_rate": 0.00019491894554382052,
      "loss": 1.1896,
      "step": 7590
    },
    {
      "epoch": 0.3079354145985697,
      "grad_norm": 0.5412753224372864,
      "learning_rate": 0.00019490557470106686,
      "loss": 1.1506,
      "step": 7600
    },
    {
      "epoch": 0.3083405927756731,
      "grad_norm": 0.5415418148040771,
      "learning_rate": 0.00019489218674849455,
      "loss": 1.1667,
      "step": 7610
    },
    {
      "epoch": 0.3087457709527765,
      "grad_norm": 0.5018081068992615,
      "learning_rate": 0.00019487878168851714,
      "loss": 1.2007,
      "step": 7620
    },
    {
      "epoch": 0.30915094912987984,
      "grad_norm": 0.5357879400253296,
      "learning_rate": 0.00019486535952355135,
      "loss": 1.156,
      "step": 7630
    },
    {
      "epoch": 0.30955612730698323,
      "grad_norm": 0.5055729746818542,
      "learning_rate": 0.00019485192025601693,
      "loss": 1.1327,
      "step": 7640
    },
    {
      "epoch": 0.3099613054840866,
      "grad_norm": 0.6071487069129944,
      "learning_rate": 0.0001948384638883368,
      "loss": 1.1755,
      "step": 7650
    },
    {
      "epoch": 0.31036648366119,
      "grad_norm": 0.5427196621894836,
      "learning_rate": 0.00019482499042293688,
      "loss": 1.1564,
      "step": 7660
    },
    {
      "epoch": 0.3107716618382934,
      "grad_norm": 0.5550917387008667,
      "learning_rate": 0.0001948114998622462,
      "loss": 1.182,
      "step": 7670
    },
    {
      "epoch": 0.31117684001539675,
      "grad_norm": 0.5451796650886536,
      "learning_rate": 0.00019479799220869682,
      "loss": 1.1476,
      "step": 7680
    },
    {
      "epoch": 0.31158201819250014,
      "grad_norm": 0.5894560217857361,
      "learning_rate": 0.00019478446746472409,
      "loss": 1.1286,
      "step": 7690
    },
    {
      "epoch": 0.31198719636960354,
      "grad_norm": 0.5963010787963867,
      "learning_rate": 0.0001947709256327661,
      "loss": 1.1525,
      "step": 7700
    },
    {
      "epoch": 0.31239237454670693,
      "grad_norm": 0.5389785766601562,
      "learning_rate": 0.00019475736671526432,
      "loss": 1.1676,
      "step": 7710
    },
    {
      "epoch": 0.31279755272381027,
      "grad_norm": 0.5249213576316833,
      "learning_rate": 0.00019474379071466317,
      "loss": 1.1565,
      "step": 7720
    },
    {
      "epoch": 0.31320273090091366,
      "grad_norm": 0.5237864851951599,
      "learning_rate": 0.00019473019763341013,
      "loss": 1.132,
      "step": 7730
    },
    {
      "epoch": 0.31360790907801706,
      "grad_norm": 0.5444650053977966,
      "learning_rate": 0.0001947165874739558,
      "loss": 1.1491,
      "step": 7740
    },
    {
      "epoch": 0.31401308725512045,
      "grad_norm": 0.5129560828208923,
      "learning_rate": 0.00019470296023875387,
      "loss": 1.143,
      "step": 7750
    },
    {
      "epoch": 0.31441826543222384,
      "grad_norm": 0.566936194896698,
      "learning_rate": 0.00019468931593026115,
      "loss": 1.1866,
      "step": 7760
    },
    {
      "epoch": 0.3148234436093272,
      "grad_norm": 0.5268216729164124,
      "learning_rate": 0.00019467565455093735,
      "loss": 1.1634,
      "step": 7770
    },
    {
      "epoch": 0.3152286217864306,
      "grad_norm": 0.48936551809310913,
      "learning_rate": 0.00019466197610324548,
      "loss": 1.1402,
      "step": 7780
    },
    {
      "epoch": 0.31563379996353397,
      "grad_norm": 0.5692466497421265,
      "learning_rate": 0.00019464828058965149,
      "loss": 1.1776,
      "step": 7790
    },
    {
      "epoch": 0.31603897814063736,
      "grad_norm": 0.5340446829795837,
      "learning_rate": 0.00019463456801262444,
      "loss": 1.1409,
      "step": 7800
    },
    {
      "epoch": 0.3164441563177407,
      "grad_norm": 0.5517352223396301,
      "learning_rate": 0.00019462083837463643,
      "loss": 1.1382,
      "step": 7810
    },
    {
      "epoch": 0.3168493344948441,
      "grad_norm": 0.5469731092453003,
      "learning_rate": 0.00019460709167816274,
      "loss": 1.1715,
      "step": 7820
    },
    {
      "epoch": 0.3172545126719475,
      "grad_norm": 0.5440459251403809,
      "learning_rate": 0.0001945933279256816,
      "loss": 1.1761,
      "step": 7830
    },
    {
      "epoch": 0.3176596908490509,
      "grad_norm": 0.5324951410293579,
      "learning_rate": 0.00019457954711967447,
      "loss": 1.205,
      "step": 7840
    },
    {
      "epoch": 0.3180648690261543,
      "grad_norm": 0.5794885158538818,
      "learning_rate": 0.0001945657492626257,
      "loss": 1.1905,
      "step": 7850
    },
    {
      "epoch": 0.3184700472032576,
      "grad_norm": 0.5542827844619751,
      "learning_rate": 0.0001945519343570228,
      "loss": 1.1487,
      "step": 7860
    },
    {
      "epoch": 0.318875225380361,
      "grad_norm": 0.6022090315818787,
      "learning_rate": 0.0001945381024053564,
      "loss": 1.1453,
      "step": 7870
    },
    {
      "epoch": 0.3192804035574644,
      "grad_norm": 0.5916401147842407,
      "learning_rate": 0.00019452425341012015,
      "loss": 1.1355,
      "step": 7880
    },
    {
      "epoch": 0.3196855817345678,
      "grad_norm": 0.5113353729248047,
      "learning_rate": 0.00019451038737381077,
      "loss": 1.1407,
      "step": 7890
    },
    {
      "epoch": 0.32009075991167113,
      "grad_norm": 0.5158296227455139,
      "learning_rate": 0.00019449650429892809,
      "loss": 1.1908,
      "step": 7900
    },
    {
      "epoch": 0.3204959380887745,
      "grad_norm": 0.6274235248565674,
      "learning_rate": 0.00019448260418797493,
      "loss": 1.1882,
      "step": 7910
    },
    {
      "epoch": 0.3209011162658779,
      "grad_norm": 0.5001250505447388,
      "learning_rate": 0.00019446868704345728,
      "loss": 1.175,
      "step": 7920
    },
    {
      "epoch": 0.3213062944429813,
      "grad_norm": 0.4985031187534332,
      "learning_rate": 0.00019445475286788418,
      "loss": 1.174,
      "step": 7930
    },
    {
      "epoch": 0.3217114726200847,
      "grad_norm": 0.610883355140686,
      "learning_rate": 0.0001944408016637677,
      "loss": 1.143,
      "step": 7940
    },
    {
      "epoch": 0.32211665079718804,
      "grad_norm": 0.5004714727401733,
      "learning_rate": 0.0001944268334336229,
      "loss": 1.1612,
      "step": 7950
    },
    {
      "epoch": 0.32252182897429144,
      "grad_norm": 0.5297211408615112,
      "learning_rate": 0.0001944128481799682,
      "loss": 1.1551,
      "step": 7960
    },
    {
      "epoch": 0.32292700715139483,
      "grad_norm": 0.567433774471283,
      "learning_rate": 0.00019439884590532477,
      "loss": 1.1637,
      "step": 7970
    },
    {
      "epoch": 0.3233321853284982,
      "grad_norm": 0.48388245701789856,
      "learning_rate": 0.00019438482661221696,
      "loss": 1.1568,
      "step": 7980
    },
    {
      "epoch": 0.32373736350560156,
      "grad_norm": 0.5460348129272461,
      "learning_rate": 0.00019437079030317227,
      "loss": 1.1628,
      "step": 7990
    },
    {
      "epoch": 0.32414254168270495,
      "grad_norm": 0.4971807897090912,
      "learning_rate": 0.00019435673698072115,
      "loss": 1.1531,
      "step": 8000
    },
    {
      "epoch": 0.32454771985980835,
      "grad_norm": 0.5979961156845093,
      "learning_rate": 0.00019434266664739721,
      "loss": 1.1382,
      "step": 8010
    },
    {
      "epoch": 0.32495289803691174,
      "grad_norm": 0.5116289854049683,
      "learning_rate": 0.0001943285793057371,
      "loss": 1.1791,
      "step": 8020
    },
    {
      "epoch": 0.32535807621401514,
      "grad_norm": 0.7331942319869995,
      "learning_rate": 0.00019431447495828045,
      "loss": 1.1368,
      "step": 8030
    },
    {
      "epoch": 0.3257632543911185,
      "grad_norm": 0.5842032432556152,
      "learning_rate": 0.0001943003536075701,
      "loss": 1.1551,
      "step": 8040
    },
    {
      "epoch": 0.32616843256822187,
      "grad_norm": 0.5512654781341553,
      "learning_rate": 0.00019428621525615185,
      "loss": 1.1967,
      "step": 8050
    },
    {
      "epoch": 0.32657361074532526,
      "grad_norm": 0.5211259126663208,
      "learning_rate": 0.00019427205990657457,
      "loss": 1.1308,
      "step": 8060
    },
    {
      "epoch": 0.32697878892242865,
      "grad_norm": 0.543393611907959,
      "learning_rate": 0.00019425788756139028,
      "loss": 1.2008,
      "step": 8070
    },
    {
      "epoch": 0.327383967099532,
      "grad_norm": 0.5631904602050781,
      "learning_rate": 0.00019424369822315397,
      "loss": 1.153,
      "step": 8080
    },
    {
      "epoch": 0.3277891452766354,
      "grad_norm": 0.5244068503379822,
      "learning_rate": 0.00019422949189442374,
      "loss": 1.1869,
      "step": 8090
    },
    {
      "epoch": 0.3281943234537388,
      "grad_norm": 0.5577735304832458,
      "learning_rate": 0.00019421526857776072,
      "loss": 1.1633,
      "step": 8100
    },
    {
      "epoch": 0.3285995016308422,
      "grad_norm": 0.5775492191314697,
      "learning_rate": 0.0001942010282757292,
      "loss": 1.195,
      "step": 8110
    },
    {
      "epoch": 0.32900467980794557,
      "grad_norm": 0.5658437609672546,
      "learning_rate": 0.00019418677099089637,
      "loss": 1.1609,
      "step": 8120
    },
    {
      "epoch": 0.3294098579850489,
      "grad_norm": 0.6724469661712646,
      "learning_rate": 0.0001941724967258326,
      "loss": 1.1398,
      "step": 8130
    },
    {
      "epoch": 0.3298150361621523,
      "grad_norm": 0.5468661785125732,
      "learning_rate": 0.00019415820548311131,
      "loss": 1.1828,
      "step": 8140
    },
    {
      "epoch": 0.3302202143392557,
      "grad_norm": 0.5473029613494873,
      "learning_rate": 0.00019414389726530896,
      "loss": 1.1649,
      "step": 8150
    },
    {
      "epoch": 0.3306253925163591,
      "grad_norm": 0.518471896648407,
      "learning_rate": 0.00019412957207500504,
      "loss": 1.1538,
      "step": 8160
    },
    {
      "epoch": 0.3310305706934624,
      "grad_norm": 0.5464076995849609,
      "learning_rate": 0.00019411522991478214,
      "loss": 1.1852,
      "step": 8170
    },
    {
      "epoch": 0.3314357488705658,
      "grad_norm": 0.554458737373352,
      "learning_rate": 0.00019410087078722596,
      "loss": 1.1717,
      "step": 8180
    },
    {
      "epoch": 0.3318409270476692,
      "grad_norm": 0.5081684589385986,
      "learning_rate": 0.0001940864946949251,
      "loss": 1.1477,
      "step": 8190
    },
    {
      "epoch": 0.3322461052247726,
      "grad_norm": 0.5801004767417908,
      "learning_rate": 0.0001940721016404714,
      "loss": 1.1968,
      "step": 8200
    },
    {
      "epoch": 0.332651283401876,
      "grad_norm": 0.5117192268371582,
      "learning_rate": 0.00019405769162645966,
      "loss": 1.16,
      "step": 8210
    },
    {
      "epoch": 0.33305646157897933,
      "grad_norm": 0.5471206307411194,
      "learning_rate": 0.0001940432646554877,
      "loss": 1.1926,
      "step": 8220
    },
    {
      "epoch": 0.33346163975608273,
      "grad_norm": 0.7296261787414551,
      "learning_rate": 0.00019402882073015653,
      "loss": 1.1346,
      "step": 8230
    },
    {
      "epoch": 0.3338668179331861,
      "grad_norm": 0.5716671347618103,
      "learning_rate": 0.00019401435985307012,
      "loss": 1.1521,
      "step": 8240
    },
    {
      "epoch": 0.3342719961102895,
      "grad_norm": 0.6165962219238281,
      "learning_rate": 0.00019399988202683542,
      "loss": 1.169,
      "step": 8250
    },
    {
      "epoch": 0.33467717428739285,
      "grad_norm": 0.572355329990387,
      "learning_rate": 0.00019398538725406267,
      "loss": 1.135,
      "step": 8260
    },
    {
      "epoch": 0.33508235246449625,
      "grad_norm": 0.5236608386039734,
      "learning_rate": 0.00019397087553736495,
      "loss": 1.1425,
      "step": 8270
    },
    {
      "epoch": 0.33548753064159964,
      "grad_norm": 0.5052462220191956,
      "learning_rate": 0.00019395634687935847,
      "loss": 1.1605,
      "step": 8280
    },
    {
      "epoch": 0.33589270881870303,
      "grad_norm": 0.564010500907898,
      "learning_rate": 0.00019394180128266247,
      "loss": 1.1354,
      "step": 8290
    },
    {
      "epoch": 0.3362978869958064,
      "grad_norm": 0.5275155305862427,
      "learning_rate": 0.00019392723874989934,
      "loss": 1.1745,
      "step": 8300
    },
    {
      "epoch": 0.33670306517290977,
      "grad_norm": 0.5634216666221619,
      "learning_rate": 0.0001939126592836944,
      "loss": 1.145,
      "step": 8310
    },
    {
      "epoch": 0.33710824335001316,
      "grad_norm": 0.4916865825653076,
      "learning_rate": 0.00019389806288667606,
      "loss": 1.1357,
      "step": 8320
    },
    {
      "epoch": 0.33751342152711655,
      "grad_norm": 0.5655689239501953,
      "learning_rate": 0.00019388344956147583,
      "loss": 1.1484,
      "step": 8330
    },
    {
      "epoch": 0.33791859970421995,
      "grad_norm": 0.5690198540687561,
      "learning_rate": 0.00019386881931072824,
      "loss": 1.1764,
      "step": 8340
    },
    {
      "epoch": 0.3383237778813233,
      "grad_norm": 0.6636972427368164,
      "learning_rate": 0.00019385417213707085,
      "loss": 1.1795,
      "step": 8350
    },
    {
      "epoch": 0.3387289560584267,
      "grad_norm": 0.53648442029953,
      "learning_rate": 0.00019383950804314427,
      "loss": 1.1728,
      "step": 8360
    },
    {
      "epoch": 0.33913413423553007,
      "grad_norm": 0.4924868643283844,
      "learning_rate": 0.00019382482703159224,
      "loss": 1.1708,
      "step": 8370
    },
    {
      "epoch": 0.33953931241263346,
      "grad_norm": 0.5547946691513062,
      "learning_rate": 0.00019381012910506146,
      "loss": 1.1481,
      "step": 8380
    },
    {
      "epoch": 0.33994449058973686,
      "grad_norm": 0.5303977727890015,
      "learning_rate": 0.00019379541426620163,
      "loss": 1.1599,
      "step": 8390
    },
    {
      "epoch": 0.3403496687668402,
      "grad_norm": 0.5251225233078003,
      "learning_rate": 0.00019378068251766573,
      "loss": 1.1882,
      "step": 8400
    },
    {
      "epoch": 0.3407548469439436,
      "grad_norm": 0.5438474416732788,
      "learning_rate": 0.00019376593386210954,
      "loss": 1.1609,
      "step": 8410
    },
    {
      "epoch": 0.341160025121047,
      "grad_norm": 0.6512997150421143,
      "learning_rate": 0.000193751168302192,
      "loss": 1.1544,
      "step": 8420
    },
    {
      "epoch": 0.3415652032981504,
      "grad_norm": 0.4715906083583832,
      "learning_rate": 0.00019373638584057506,
      "loss": 1.1728,
      "step": 8430
    },
    {
      "epoch": 0.3419703814752537,
      "grad_norm": 0.592898964881897,
      "learning_rate": 0.0001937215864799238,
      "loss": 1.1508,
      "step": 8440
    },
    {
      "epoch": 0.3423755596523571,
      "grad_norm": 0.5592702627182007,
      "learning_rate": 0.00019370677022290624,
      "loss": 1.1532,
      "step": 8450
    },
    {
      "epoch": 0.3427807378294605,
      "grad_norm": 0.5228490233421326,
      "learning_rate": 0.0001936919370721935,
      "loss": 1.152,
      "step": 8460
    },
    {
      "epoch": 0.3431859160065639,
      "grad_norm": 0.5241566896438599,
      "learning_rate": 0.00019367708703045978,
      "loss": 1.1366,
      "step": 8470
    },
    {
      "epoch": 0.3435910941836673,
      "grad_norm": 0.5437764525413513,
      "learning_rate": 0.00019366222010038222,
      "loss": 1.1387,
      "step": 8480
    },
    {
      "epoch": 0.3439962723607706,
      "grad_norm": 0.542452335357666,
      "learning_rate": 0.0001936473362846411,
      "loss": 1.1613,
      "step": 8490
    },
    {
      "epoch": 0.344401450537874,
      "grad_norm": 0.5705374479293823,
      "learning_rate": 0.00019363243558591972,
      "loss": 1.1214,
      "step": 8500
    },
    {
      "epoch": 0.3448066287149774,
      "grad_norm": 0.5786519050598145,
      "learning_rate": 0.00019361751800690437,
      "loss": 1.1232,
      "step": 8510
    },
    {
      "epoch": 0.3452118068920808,
      "grad_norm": 0.5558506846427917,
      "learning_rate": 0.0001936025835502845,
      "loss": 1.1156,
      "step": 8520
    },
    {
      "epoch": 0.34561698506918415,
      "grad_norm": 0.5102654695510864,
      "learning_rate": 0.00019358763221875248,
      "loss": 1.1546,
      "step": 8530
    },
    {
      "epoch": 0.34602216324628754,
      "grad_norm": 0.5398111343383789,
      "learning_rate": 0.00019357266401500377,
      "loss": 1.1723,
      "step": 8540
    },
    {
      "epoch": 0.34642734142339093,
      "grad_norm": 0.558383047580719,
      "learning_rate": 0.00019355767894173694,
      "loss": 1.1528,
      "step": 8550
    },
    {
      "epoch": 0.3468325196004943,
      "grad_norm": 0.5148330330848694,
      "learning_rate": 0.00019354267700165347,
      "loss": 1.1624,
      "step": 8560
    },
    {
      "epoch": 0.3472376977775977,
      "grad_norm": 0.6650025248527527,
      "learning_rate": 0.00019352765819745797,
      "loss": 1.1571,
      "step": 8570
    },
    {
      "epoch": 0.34764287595470106,
      "grad_norm": 0.5181428790092468,
      "learning_rate": 0.00019351262253185808,
      "loss": 1.1636,
      "step": 8580
    },
    {
      "epoch": 0.34804805413180445,
      "grad_norm": 0.6499121189117432,
      "learning_rate": 0.00019349757000756444,
      "loss": 1.136,
      "step": 8590
    },
    {
      "epoch": 0.34845323230890785,
      "grad_norm": 0.6723763346672058,
      "learning_rate": 0.00019348250062729076,
      "loss": 1.1491,
      "step": 8600
    },
    {
      "epoch": 0.34885841048601124,
      "grad_norm": 0.5413874387741089,
      "learning_rate": 0.00019346741439375383,
      "loss": 1.1944,
      "step": 8610
    },
    {
      "epoch": 0.3492635886631146,
      "grad_norm": 0.6056035161018372,
      "learning_rate": 0.0001934523113096734,
      "loss": 1.1615,
      "step": 8620
    },
    {
      "epoch": 0.34966876684021797,
      "grad_norm": 0.5539503693580627,
      "learning_rate": 0.00019343719137777227,
      "loss": 1.1414,
      "step": 8630
    },
    {
      "epoch": 0.35007394501732136,
      "grad_norm": 0.5582520961761475,
      "learning_rate": 0.00019342205460077633,
      "loss": 1.1856,
      "step": 8640
    },
    {
      "epoch": 0.35047912319442476,
      "grad_norm": 0.5007537007331848,
      "learning_rate": 0.00019340690098141453,
      "loss": 1.1367,
      "step": 8650
    },
    {
      "epoch": 0.35088430137152815,
      "grad_norm": 0.6193983554840088,
      "learning_rate": 0.0001933917305224187,
      "loss": 1.1417,
      "step": 8660
    },
    {
      "epoch": 0.3512894795486315,
      "grad_norm": 0.5897209644317627,
      "learning_rate": 0.00019337654322652386,
      "loss": 1.1415,
      "step": 8670
    },
    {
      "epoch": 0.3516946577257349,
      "grad_norm": 0.5420259237289429,
      "learning_rate": 0.00019336133909646802,
      "loss": 1.2013,
      "step": 8680
    },
    {
      "epoch": 0.3520998359028383,
      "grad_norm": 0.5215559601783752,
      "learning_rate": 0.00019334611813499216,
      "loss": 1.1512,
      "step": 8690
    },
    {
      "epoch": 0.35250501407994167,
      "grad_norm": 0.5385690927505493,
      "learning_rate": 0.00019333088034484048,
      "loss": 1.1273,
      "step": 8700
    },
    {
      "epoch": 0.352910192257045,
      "grad_norm": 0.5425608158111572,
      "learning_rate": 0.00019331562572875996,
      "loss": 1.1621,
      "step": 8710
    },
    {
      "epoch": 0.3533153704341484,
      "grad_norm": 0.51172935962677,
      "learning_rate": 0.00019330035428950085,
      "loss": 1.1704,
      "step": 8720
    },
    {
      "epoch": 0.3537205486112518,
      "grad_norm": 0.5679565668106079,
      "learning_rate": 0.0001932850660298162,
      "loss": 1.1816,
      "step": 8730
    },
    {
      "epoch": 0.3541257267883552,
      "grad_norm": 0.5418999195098877,
      "learning_rate": 0.0001932697609524623,
      "loss": 1.1871,
      "step": 8740
    },
    {
      "epoch": 0.3545309049654586,
      "grad_norm": 0.49425557255744934,
      "learning_rate": 0.00019325443906019837,
      "loss": 1.1239,
      "step": 8750
    },
    {
      "epoch": 0.3549360831425619,
      "grad_norm": 0.6118369102478027,
      "learning_rate": 0.0001932391003557867,
      "loss": 1.1757,
      "step": 8760
    },
    {
      "epoch": 0.3553412613196653,
      "grad_norm": 0.5392184853553772,
      "learning_rate": 0.00019322374484199257,
      "loss": 1.1252,
      "step": 8770
    },
    {
      "epoch": 0.3557464394967687,
      "grad_norm": 0.537943422794342,
      "learning_rate": 0.0001932083725215843,
      "loss": 1.151,
      "step": 8780
    },
    {
      "epoch": 0.3561516176738721,
      "grad_norm": 0.6165295243263245,
      "learning_rate": 0.00019319298339733325,
      "loss": 1.1267,
      "step": 8790
    },
    {
      "epoch": 0.35655679585097544,
      "grad_norm": 0.5713896155357361,
      "learning_rate": 0.00019317757747201384,
      "loss": 1.1578,
      "step": 8800
    },
    {
      "epoch": 0.35696197402807883,
      "grad_norm": 0.5356034636497498,
      "learning_rate": 0.0001931621547484035,
      "loss": 1.1733,
      "step": 8810
    },
    {
      "epoch": 0.3573671522051822,
      "grad_norm": 0.5164676308631897,
      "learning_rate": 0.00019314671522928262,
      "loss": 1.2046,
      "step": 8820
    },
    {
      "epoch": 0.3577723303822856,
      "grad_norm": 0.6506962776184082,
      "learning_rate": 0.0001931312589174347,
      "loss": 1.1713,
      "step": 8830
    },
    {
      "epoch": 0.358177508559389,
      "grad_norm": 0.532515287399292,
      "learning_rate": 0.00019311578581564627,
      "loss": 1.1462,
      "step": 8840
    },
    {
      "epoch": 0.35858268673649235,
      "grad_norm": 0.6302481889724731,
      "learning_rate": 0.0001931002959267068,
      "loss": 1.1554,
      "step": 8850
    },
    {
      "epoch": 0.35898786491359574,
      "grad_norm": 0.4934979975223541,
      "learning_rate": 0.00019308478925340894,
      "loss": 1.1504,
      "step": 8860
    },
    {
      "epoch": 0.35939304309069914,
      "grad_norm": 0.5373842716217041,
      "learning_rate": 0.00019306926579854821,
      "loss": 1.1421,
      "step": 8870
    },
    {
      "epoch": 0.35979822126780253,
      "grad_norm": 0.552449643611908,
      "learning_rate": 0.00019305372556492322,
      "loss": 1.1731,
      "step": 8880
    },
    {
      "epoch": 0.3602033994449059,
      "grad_norm": 0.5033772587776184,
      "learning_rate": 0.0001930381685553356,
      "loss": 1.1969,
      "step": 8890
    },
    {
      "epoch": 0.36060857762200926,
      "grad_norm": 0.49306532740592957,
      "learning_rate": 0.00019302259477259,
      "loss": 1.1693,
      "step": 8900
    },
    {
      "epoch": 0.36101375579911266,
      "grad_norm": 0.5244967937469482,
      "learning_rate": 0.00019300700421949417,
      "loss": 1.1469,
      "step": 8910
    },
    {
      "epoch": 0.36141893397621605,
      "grad_norm": 0.53569495677948,
      "learning_rate": 0.00019299139689885872,
      "loss": 1.1424,
      "step": 8920
    },
    {
      "epoch": 0.36182411215331944,
      "grad_norm": 0.5408770442008972,
      "learning_rate": 0.00019297577281349746,
      "loss": 1.145,
      "step": 8930
    },
    {
      "epoch": 0.3622292903304228,
      "grad_norm": 0.5181036591529846,
      "learning_rate": 0.00019296013196622706,
      "loss": 1.1879,
      "step": 8940
    },
    {
      "epoch": 0.3626344685075262,
      "grad_norm": 0.5371408462524414,
      "learning_rate": 0.00019294447435986735,
      "loss": 1.1176,
      "step": 8950
    },
    {
      "epoch": 0.36303964668462957,
      "grad_norm": 0.5194632411003113,
      "learning_rate": 0.00019292879999724112,
      "loss": 1.1681,
      "step": 8960
    },
    {
      "epoch": 0.36344482486173296,
      "grad_norm": 0.6200748682022095,
      "learning_rate": 0.00019291310888117417,
      "loss": 1.1808,
      "step": 8970
    },
    {
      "epoch": 0.36385000303883636,
      "grad_norm": 0.4716326594352722,
      "learning_rate": 0.0001928974010144953,
      "loss": 1.1183,
      "step": 8980
    },
    {
      "epoch": 0.3642551812159397,
      "grad_norm": 0.5383079648017883,
      "learning_rate": 0.00019288167640003643,
      "loss": 1.1645,
      "step": 8990
    },
    {
      "epoch": 0.3646603593930431,
      "grad_norm": 0.5155847668647766,
      "learning_rate": 0.00019286593504063237,
      "loss": 1.1277,
      "step": 9000
    },
    {
      "epoch": 0.3650655375701465,
      "grad_norm": 0.4837091267108917,
      "learning_rate": 0.00019285017693912107,
      "loss": 1.1601,
      "step": 9010
    },
    {
      "epoch": 0.3654707157472499,
      "grad_norm": 0.5722483396530151,
      "learning_rate": 0.00019283440209834344,
      "loss": 1.1795,
      "step": 9020
    },
    {
      "epoch": 0.3658758939243532,
      "grad_norm": 0.5485572814941406,
      "learning_rate": 0.00019281861052114333,
      "loss": 1.1516,
      "step": 9030
    },
    {
      "epoch": 0.3662810721014566,
      "grad_norm": 0.5527223348617554,
      "learning_rate": 0.00019280280221036777,
      "loss": 1.1719,
      "step": 9040
    },
    {
      "epoch": 0.36668625027856,
      "grad_norm": 0.47823378443717957,
      "learning_rate": 0.00019278697716886668,
      "loss": 1.161,
      "step": 9050
    },
    {
      "epoch": 0.3670914284556634,
      "grad_norm": 0.5318389534950256,
      "learning_rate": 0.00019277113539949308,
      "loss": 1.1403,
      "step": 9060
    },
    {
      "epoch": 0.3674966066327668,
      "grad_norm": 0.5542261600494385,
      "learning_rate": 0.0001927552769051029,
      "loss": 1.1233,
      "step": 9070
    },
    {
      "epoch": 0.3679017848098701,
      "grad_norm": 0.5315076112747192,
      "learning_rate": 0.00019273940168855518,
      "loss": 1.1738,
      "step": 9080
    },
    {
      "epoch": 0.3683069629869735,
      "grad_norm": 0.5321982502937317,
      "learning_rate": 0.000192723509752712,
      "loss": 1.1684,
      "step": 9090
    },
    {
      "epoch": 0.3687121411640769,
      "grad_norm": 0.5200474858283997,
      "learning_rate": 0.00019270760110043827,
      "loss": 1.1649,
      "step": 9100
    },
    {
      "epoch": 0.3691173193411803,
      "grad_norm": 0.591718852519989,
      "learning_rate": 0.0001926916757346022,
      "loss": 1.1617,
      "step": 9110
    },
    {
      "epoch": 0.36952249751828364,
      "grad_norm": 0.5367197394371033,
      "learning_rate": 0.00019267573365807473,
      "loss": 1.1731,
      "step": 9120
    },
    {
      "epoch": 0.36992767569538704,
      "grad_norm": 0.6061881184577942,
      "learning_rate": 0.00019265977487372999,
      "loss": 1.1603,
      "step": 9130
    },
    {
      "epoch": 0.37033285387249043,
      "grad_norm": 0.5740823149681091,
      "learning_rate": 0.00019264379938444508,
      "loss": 1.1406,
      "step": 9140
    },
    {
      "epoch": 0.3707380320495938,
      "grad_norm": 0.65552818775177,
      "learning_rate": 0.00019262780719310008,
      "loss": 1.184,
      "step": 9150
    },
    {
      "epoch": 0.3711432102266972,
      "grad_norm": 0.5972305536270142,
      "learning_rate": 0.0001926117983025781,
      "loss": 1.1797,
      "step": 9160
    },
    {
      "epoch": 0.37154838840380056,
      "grad_norm": 0.5455591082572937,
      "learning_rate": 0.00019259577271576528,
      "loss": 1.1392,
      "step": 9170
    },
    {
      "epoch": 0.37195356658090395,
      "grad_norm": 0.5609455108642578,
      "learning_rate": 0.00019257973043555074,
      "loss": 1.1689,
      "step": 9180
    },
    {
      "epoch": 0.37235874475800734,
      "grad_norm": 0.5423162579536438,
      "learning_rate": 0.00019256367146482664,
      "loss": 1.1484,
      "step": 9190
    },
    {
      "epoch": 0.37276392293511074,
      "grad_norm": 0.5789716839790344,
      "learning_rate": 0.0001925475958064881,
      "loss": 1.1588,
      "step": 9200
    },
    {
      "epoch": 0.3731691011122141,
      "grad_norm": 0.52518630027771,
      "learning_rate": 0.00019253150346343331,
      "loss": 1.0981,
      "step": 9210
    },
    {
      "epoch": 0.37357427928931747,
      "grad_norm": 0.5490677356719971,
      "learning_rate": 0.00019251539443856344,
      "loss": 1.1941,
      "step": 9220
    },
    {
      "epoch": 0.37397945746642086,
      "grad_norm": 0.5571455955505371,
      "learning_rate": 0.00019249926873478266,
      "loss": 1.1399,
      "step": 9230
    },
    {
      "epoch": 0.37438463564352426,
      "grad_norm": 0.5040227770805359,
      "learning_rate": 0.00019248312635499813,
      "loss": 1.1681,
      "step": 9240
    },
    {
      "epoch": 0.37478981382062765,
      "grad_norm": 0.6279113292694092,
      "learning_rate": 0.00019246696730212007,
      "loss": 1.1617,
      "step": 9250
    },
    {
      "epoch": 0.375194991997731,
      "grad_norm": 0.51105135679245,
      "learning_rate": 0.0001924507915790617,
      "loss": 1.1343,
      "step": 9260
    },
    {
      "epoch": 0.3756001701748344,
      "grad_norm": 0.5037890076637268,
      "learning_rate": 0.0001924345991887391,
      "loss": 1.1506,
      "step": 9270
    },
    {
      "epoch": 0.3760053483519378,
      "grad_norm": 0.511907696723938,
      "learning_rate": 0.00019241839013407167,
      "loss": 1.1543,
      "step": 9280
    },
    {
      "epoch": 0.37641052652904117,
      "grad_norm": 0.5326095819473267,
      "learning_rate": 0.00019240216441798142,
      "loss": 1.1545,
      "step": 9290
    },
    {
      "epoch": 0.3768157047061445,
      "grad_norm": 0.4838440716266632,
      "learning_rate": 0.0001923859220433937,
      "loss": 1.1492,
      "step": 9300
    },
    {
      "epoch": 0.3772208828832479,
      "grad_norm": 0.6308650970458984,
      "learning_rate": 0.00019236966301323664,
      "loss": 1.1617,
      "step": 9310
    },
    {
      "epoch": 0.3776260610603513,
      "grad_norm": 0.5061782598495483,
      "learning_rate": 0.00019235338733044154,
      "loss": 1.1597,
      "step": 9320
    },
    {
      "epoch": 0.3780312392374547,
      "grad_norm": 0.5869961380958557,
      "learning_rate": 0.00019233709499794256,
      "loss": 1.1578,
      "step": 9330
    },
    {
      "epoch": 0.3784364174145581,
      "grad_norm": 0.5178478956222534,
      "learning_rate": 0.00019232078601867696,
      "loss": 1.1416,
      "step": 9340
    },
    {
      "epoch": 0.3788415955916614,
      "grad_norm": 0.5345783233642578,
      "learning_rate": 0.00019230446039558493,
      "loss": 1.1398,
      "step": 9350
    },
    {
      "epoch": 0.3792467737687648,
      "grad_norm": 0.5786774754524231,
      "learning_rate": 0.0001922881181316097,
      "loss": 1.1412,
      "step": 9360
    },
    {
      "epoch": 0.3796519519458682,
      "grad_norm": 0.4837273955345154,
      "learning_rate": 0.00019227175922969755,
      "loss": 1.156,
      "step": 9370
    },
    {
      "epoch": 0.3800571301229716,
      "grad_norm": 0.5547227263450623,
      "learning_rate": 0.00019225538369279765,
      "loss": 1.1579,
      "step": 9380
    },
    {
      "epoch": 0.38046230830007494,
      "grad_norm": 0.5275659561157227,
      "learning_rate": 0.0001922389915238622,
      "loss": 1.1917,
      "step": 9390
    },
    {
      "epoch": 0.38086748647717833,
      "grad_norm": 0.55735182762146,
      "learning_rate": 0.0001922225827258465,
      "loss": 1.1503,
      "step": 9400
    },
    {
      "epoch": 0.3812726646542817,
      "grad_norm": 0.5448446273803711,
      "learning_rate": 0.00019220615730170874,
      "loss": 1.2023,
      "step": 9410
    },
    {
      "epoch": 0.3816778428313851,
      "grad_norm": 0.5968173742294312,
      "learning_rate": 0.0001921897152544101,
      "loss": 1.1342,
      "step": 9420
    },
    {
      "epoch": 0.3820830210084885,
      "grad_norm": 0.5029596090316772,
      "learning_rate": 0.00019217325658691482,
      "loss": 1.1615,
      "step": 9430
    },
    {
      "epoch": 0.38248819918559185,
      "grad_norm": 0.5464463829994202,
      "learning_rate": 0.0001921567813021901,
      "loss": 1.1687,
      "step": 9440
    },
    {
      "epoch": 0.38289337736269524,
      "grad_norm": 0.6132645010948181,
      "learning_rate": 0.0001921402894032062,
      "loss": 1.1905,
      "step": 9450
    },
    {
      "epoch": 0.38329855553979864,
      "grad_norm": 0.5720375776290894,
      "learning_rate": 0.00019212378089293624,
      "loss": 1.1539,
      "step": 9460
    },
    {
      "epoch": 0.38370373371690203,
      "grad_norm": 0.6570445895195007,
      "learning_rate": 0.00019210725577435643,
      "loss": 1.1292,
      "step": 9470
    },
    {
      "epoch": 0.38410891189400537,
      "grad_norm": 0.593341588973999,
      "learning_rate": 0.00019209071405044604,
      "loss": 1.1539,
      "step": 9480
    },
    {
      "epoch": 0.38451409007110876,
      "grad_norm": 0.5940881371498108,
      "learning_rate": 0.00019207415572418713,
      "loss": 1.1535,
      "step": 9490
    },
    {
      "epoch": 0.38491926824821215,
      "grad_norm": 0.5453243255615234,
      "learning_rate": 0.00019205758079856498,
      "loss": 1.1972,
      "step": 9500
    },
    {
      "epoch": 0.38532444642531555,
      "grad_norm": 0.5481905341148376,
      "learning_rate": 0.00019204098927656777,
      "loss": 1.1577,
      "step": 9510
    },
    {
      "epoch": 0.38572962460241894,
      "grad_norm": 0.5540195107460022,
      "learning_rate": 0.00019202438116118657,
      "loss": 1.1423,
      "step": 9520
    },
    {
      "epoch": 0.3861348027795223,
      "grad_norm": 0.5641026496887207,
      "learning_rate": 0.0001920077564554156,
      "loss": 1.137,
      "step": 9530
    },
    {
      "epoch": 0.3865399809566257,
      "grad_norm": 0.5985888838768005,
      "learning_rate": 0.000191991115162252,
      "loss": 1.1171,
      "step": 9540
    },
    {
      "epoch": 0.38694515913372907,
      "grad_norm": 0.5523642897605896,
      "learning_rate": 0.00019197445728469584,
      "loss": 1.1539,
      "step": 9550
    },
    {
      "epoch": 0.38735033731083246,
      "grad_norm": 0.5972816944122314,
      "learning_rate": 0.00019195778282575033,
      "loss": 1.1307,
      "step": 9560
    },
    {
      "epoch": 0.3877555154879358,
      "grad_norm": 0.5561144351959229,
      "learning_rate": 0.00019194109178842153,
      "loss": 1.1211,
      "step": 9570
    },
    {
      "epoch": 0.3881606936650392,
      "grad_norm": 0.575575053691864,
      "learning_rate": 0.00019192438417571855,
      "loss": 1.1812,
      "step": 9580
    },
    {
      "epoch": 0.3885658718421426,
      "grad_norm": 0.5557425022125244,
      "learning_rate": 0.00019190765999065355,
      "loss": 1.1549,
      "step": 9590
    },
    {
      "epoch": 0.388971050019246,
      "grad_norm": 0.5671688914299011,
      "learning_rate": 0.00019189091923624152,
      "loss": 1.1598,
      "step": 9600
    },
    {
      "epoch": 0.38937622819634937,
      "grad_norm": 0.589186429977417,
      "learning_rate": 0.00019187416191550054,
      "loss": 1.1283,
      "step": 9610
    },
    {
      "epoch": 0.3897814063734527,
      "grad_norm": 0.5875573754310608,
      "learning_rate": 0.00019185738803145167,
      "loss": 1.1619,
      "step": 9620
    },
    {
      "epoch": 0.3901865845505561,
      "grad_norm": 0.4808312654495239,
      "learning_rate": 0.000191840597587119,
      "loss": 1.1683,
      "step": 9630
    },
    {
      "epoch": 0.3905917627276595,
      "grad_norm": 0.572636604309082,
      "learning_rate": 0.00019182379058552948,
      "loss": 1.147,
      "step": 9640
    },
    {
      "epoch": 0.3909969409047629,
      "grad_norm": 0.540720522403717,
      "learning_rate": 0.00019180696702971318,
      "loss": 1.1616,
      "step": 9650
    },
    {
      "epoch": 0.39140211908186623,
      "grad_norm": 0.5249291658401489,
      "learning_rate": 0.000191790126922703,
      "loss": 1.1666,
      "step": 9660
    },
    {
      "epoch": 0.3918072972589696,
      "grad_norm": 0.5439531803131104,
      "learning_rate": 0.00019177327026753503,
      "loss": 1.1494,
      "step": 9670
    },
    {
      "epoch": 0.392212475436073,
      "grad_norm": 0.53830486536026,
      "learning_rate": 0.0001917563970672482,
      "loss": 1.1603,
      "step": 9680
    },
    {
      "epoch": 0.3926176536131764,
      "grad_norm": 0.535106897354126,
      "learning_rate": 0.00019173950732488443,
      "loss": 1.1941,
      "step": 9690
    },
    {
      "epoch": 0.3930228317902798,
      "grad_norm": 0.4972535967826843,
      "learning_rate": 0.00019172260104348863,
      "loss": 1.1469,
      "step": 9700
    },
    {
      "epoch": 0.39342800996738314,
      "grad_norm": 0.5931172370910645,
      "learning_rate": 0.00019170567822610873,
      "loss": 1.1615,
      "step": 9710
    },
    {
      "epoch": 0.39383318814448653,
      "grad_norm": 0.5457938313484192,
      "learning_rate": 0.00019168873887579562,
      "loss": 1.1467,
      "step": 9720
    },
    {
      "epoch": 0.39423836632158993,
      "grad_norm": 0.5183608531951904,
      "learning_rate": 0.00019167178299560317,
      "loss": 1.1398,
      "step": 9730
    },
    {
      "epoch": 0.3946435444986933,
      "grad_norm": 0.6167070865631104,
      "learning_rate": 0.00019165481058858823,
      "loss": 1.1547,
      "step": 9740
    },
    {
      "epoch": 0.39504872267579666,
      "grad_norm": 0.6432676315307617,
      "learning_rate": 0.00019163782165781062,
      "loss": 1.1704,
      "step": 9750
    },
    {
      "epoch": 0.39545390085290005,
      "grad_norm": 0.5353190898895264,
      "learning_rate": 0.00019162081620633314,
      "loss": 1.1194,
      "step": 9760
    },
    {
      "epoch": 0.39585907903000345,
      "grad_norm": 0.6180189251899719,
      "learning_rate": 0.0001916037942372216,
      "loss": 1.132,
      "step": 9770
    },
    {
      "epoch": 0.39626425720710684,
      "grad_norm": 0.5208213925361633,
      "learning_rate": 0.00019158675575354478,
      "loss": 1.1609,
      "step": 9780
    },
    {
      "epoch": 0.39666943538421023,
      "grad_norm": 0.607912003993988,
      "learning_rate": 0.00019156970075837436,
      "loss": 1.2046,
      "step": 9790
    },
    {
      "epoch": 0.39707461356131357,
      "grad_norm": 0.7231870889663696,
      "learning_rate": 0.0001915526292547851,
      "loss": 1.1498,
      "step": 9800
    },
    {
      "epoch": 0.39747979173841697,
      "grad_norm": 0.5095289349555969,
      "learning_rate": 0.0001915355412458547,
      "loss": 1.1341,
      "step": 9810
    },
    {
      "epoch": 0.39788496991552036,
      "grad_norm": 0.5758703947067261,
      "learning_rate": 0.0001915184367346638,
      "loss": 1.1914,
      "step": 9820
    },
    {
      "epoch": 0.39829014809262375,
      "grad_norm": 0.5736504197120667,
      "learning_rate": 0.00019150131572429605,
      "loss": 1.1404,
      "step": 9830
    },
    {
      "epoch": 0.3986953262697271,
      "grad_norm": 0.6464651823043823,
      "learning_rate": 0.0001914841782178381,
      "loss": 1.1768,
      "step": 9840
    },
    {
      "epoch": 0.3991005044468305,
      "grad_norm": 0.5574869513511658,
      "learning_rate": 0.0001914670242183795,
      "loss": 1.1451,
      "step": 9850
    },
    {
      "epoch": 0.3995056826239339,
      "grad_norm": 0.5855175256729126,
      "learning_rate": 0.00019144985372901286,
      "loss": 1.1772,
      "step": 9860
    },
    {
      "epoch": 0.39991086080103727,
      "grad_norm": 0.5778918266296387,
      "learning_rate": 0.00019143266675283372,
      "loss": 1.1038,
      "step": 9870
    },
    {
      "epoch": 0.40031603897814066,
      "grad_norm": 0.5478941202163696,
      "learning_rate": 0.00019141546329294053,
      "loss": 1.171,
      "step": 9880
    },
    {
      "epoch": 0.400721217155244,
      "grad_norm": 0.5234413146972656,
      "learning_rate": 0.00019139824335243484,
      "loss": 1.1735,
      "step": 9890
    },
    {
      "epoch": 0.4011263953323474,
      "grad_norm": 0.5855927467346191,
      "learning_rate": 0.00019138100693442107,
      "loss": 1.1495,
      "step": 9900
    },
    {
      "epoch": 0.4015315735094508,
      "grad_norm": 0.5097750425338745,
      "learning_rate": 0.00019136375404200665,
      "loss": 1.1318,
      "step": 9910
    },
    {
      "epoch": 0.4019367516865542,
      "grad_norm": 0.5597834587097168,
      "learning_rate": 0.00019134648467830198,
      "loss": 1.1546,
      "step": 9920
    },
    {
      "epoch": 0.4023419298636575,
      "grad_norm": 0.5415542721748352,
      "learning_rate": 0.00019132919884642043,
      "loss": 1.1643,
      "step": 9930
    },
    {
      "epoch": 0.4027471080407609,
      "grad_norm": 0.5859198570251465,
      "learning_rate": 0.00019131189654947832,
      "loss": 1.1609,
      "step": 9940
    },
    {
      "epoch": 0.4031522862178643,
      "grad_norm": 0.5399028658866882,
      "learning_rate": 0.000191294577790595,
      "loss": 1.1587,
      "step": 9950
    },
    {
      "epoch": 0.4035574643949677,
      "grad_norm": 0.5352907776832581,
      "learning_rate": 0.00019127724257289266,
      "loss": 1.169,
      "step": 9960
    },
    {
      "epoch": 0.4039626425720711,
      "grad_norm": 0.5185984373092651,
      "learning_rate": 0.0001912598908994966,
      "loss": 1.1665,
      "step": 9970
    },
    {
      "epoch": 0.40436782074917443,
      "grad_norm": 0.6144821643829346,
      "learning_rate": 0.000191242522773535,
      "loss": 1.1532,
      "step": 9980
    },
    {
      "epoch": 0.4047729989262778,
      "grad_norm": 0.5322130918502808,
      "learning_rate": 0.00019122513819813902,
      "loss": 1.119,
      "step": 9990
    },
    {
      "epoch": 0.4051781771033812,
      "grad_norm": 0.6218458414077759,
      "learning_rate": 0.00019120773717644281,
      "loss": 1.1644,
      "step": 10000
    },
    {
      "epoch": 0.4055833552804846,
      "grad_norm": 0.5252158045768738,
      "learning_rate": 0.00019119031971158345,
      "loss": 1.1625,
      "step": 10010
    },
    {
      "epoch": 0.40598853345758795,
      "grad_norm": 0.5184443593025208,
      "learning_rate": 0.00019117288580670105,
      "loss": 1.1572,
      "step": 10020
    },
    {
      "epoch": 0.40639371163469135,
      "grad_norm": 0.6250405311584473,
      "learning_rate": 0.0001911554354649386,
      "loss": 1.145,
      "step": 10030
    },
    {
      "epoch": 0.40679888981179474,
      "grad_norm": 0.5869392156600952,
      "learning_rate": 0.00019113796868944213,
      "loss": 1.1391,
      "step": 10040
    },
    {
      "epoch": 0.40720406798889813,
      "grad_norm": 0.5914766788482666,
      "learning_rate": 0.00019112048548336054,
      "loss": 1.1983,
      "step": 10050
    },
    {
      "epoch": 0.4076092461660015,
      "grad_norm": 0.5182676911354065,
      "learning_rate": 0.00019110298584984578,
      "loss": 1.1727,
      "step": 10060
    },
    {
      "epoch": 0.40801442434310486,
      "grad_norm": 0.5320264101028442,
      "learning_rate": 0.00019108546979205272,
      "loss": 1.1963,
      "step": 10070
    },
    {
      "epoch": 0.40841960252020826,
      "grad_norm": 0.5638129711151123,
      "learning_rate": 0.00019106793731313925,
      "loss": 1.1302,
      "step": 10080
    },
    {
      "epoch": 0.40882478069731165,
      "grad_norm": 0.58304762840271,
      "learning_rate": 0.0001910503884162661,
      "loss": 1.1446,
      "step": 10090
    },
    {
      "epoch": 0.40922995887441505,
      "grad_norm": 0.5920695662498474,
      "learning_rate": 0.00019103282310459704,
      "loss": 1.1651,
      "step": 10100
    },
    {
      "epoch": 0.4096351370515184,
      "grad_norm": 0.5541982650756836,
      "learning_rate": 0.0001910152413812988,
      "loss": 1.1519,
      "step": 10110
    },
    {
      "epoch": 0.4100403152286218,
      "grad_norm": 0.515554666519165,
      "learning_rate": 0.00019099764324954105,
      "loss": 1.151,
      "step": 10120
    },
    {
      "epoch": 0.41044549340572517,
      "grad_norm": 0.5591425895690918,
      "learning_rate": 0.00019098002871249646,
      "loss": 1.1829,
      "step": 10130
    },
    {
      "epoch": 0.41085067158282856,
      "grad_norm": 0.525628387928009,
      "learning_rate": 0.00019096239777334058,
      "loss": 1.1433,
      "step": 10140
    },
    {
      "epoch": 0.41125584975993196,
      "grad_norm": 0.5111804604530334,
      "learning_rate": 0.00019094475043525196,
      "loss": 1.1273,
      "step": 10150
    },
    {
      "epoch": 0.4116610279370353,
      "grad_norm": 0.5184610486030579,
      "learning_rate": 0.00019092708670141216,
      "loss": 1.138,
      "step": 10160
    },
    {
      "epoch": 0.4120662061141387,
      "grad_norm": 0.5225592851638794,
      "learning_rate": 0.00019090940657500563,
      "loss": 1.154,
      "step": 10170
    },
    {
      "epoch": 0.4124713842912421,
      "grad_norm": 0.5461578965187073,
      "learning_rate": 0.00019089171005921975,
      "loss": 1.1653,
      "step": 10180
    },
    {
      "epoch": 0.4128765624683455,
      "grad_norm": 0.5846377611160278,
      "learning_rate": 0.0001908739971572449,
      "loss": 1.1765,
      "step": 10190
    },
    {
      "epoch": 0.4132817406454488,
      "grad_norm": 0.5721112489700317,
      "learning_rate": 0.00019085626787227443,
      "loss": 1.1485,
      "step": 10200
    },
    {
      "epoch": 0.4136869188225522,
      "grad_norm": 0.6612721085548401,
      "learning_rate": 0.0001908385222075046,
      "loss": 1.1731,
      "step": 10210
    },
    {
      "epoch": 0.4140920969996556,
      "grad_norm": 0.5190238952636719,
      "learning_rate": 0.00019082076016613465,
      "loss": 1.1398,
      "step": 10220
    },
    {
      "epoch": 0.414497275176759,
      "grad_norm": 0.5103272795677185,
      "learning_rate": 0.0001908029817513668,
      "loss": 1.1462,
      "step": 10230
    },
    {
      "epoch": 0.4149024533538624,
      "grad_norm": 0.5109339952468872,
      "learning_rate": 0.00019078518696640613,
      "loss": 1.1344,
      "step": 10240
    },
    {
      "epoch": 0.4153076315309657,
      "grad_norm": 0.5773093104362488,
      "learning_rate": 0.00019076737581446077,
      "loss": 1.1722,
      "step": 10250
    },
    {
      "epoch": 0.4157128097080691,
      "grad_norm": 0.5164252519607544,
      "learning_rate": 0.0001907495482987417,
      "loss": 1.1496,
      "step": 10260
    },
    {
      "epoch": 0.4161179878851725,
      "grad_norm": 0.5852121114730835,
      "learning_rate": 0.00019073170442246302,
      "loss": 1.1933,
      "step": 10270
    },
    {
      "epoch": 0.4165231660622759,
      "grad_norm": 0.5736199617385864,
      "learning_rate": 0.0001907138441888416,
      "loss": 1.1689,
      "step": 10280
    },
    {
      "epoch": 0.41692834423937924,
      "grad_norm": 0.511085033416748,
      "learning_rate": 0.00019069596760109728,
      "loss": 1.1461,
      "step": 10290
    },
    {
      "epoch": 0.41733352241648264,
      "grad_norm": 0.5795160531997681,
      "learning_rate": 0.000190678074662453,
      "loss": 1.1485,
      "step": 10300
    },
    {
      "epoch": 0.41773870059358603,
      "grad_norm": 0.5866442918777466,
      "learning_rate": 0.0001906601653761345,
      "loss": 1.1513,
      "step": 10310
    },
    {
      "epoch": 0.4181438787706894,
      "grad_norm": 0.5791808366775513,
      "learning_rate": 0.00019064223974537048,
      "loss": 1.1619,
      "step": 10320
    },
    {
      "epoch": 0.4185490569477928,
      "grad_norm": 0.601767361164093,
      "learning_rate": 0.00019062429777339264,
      "loss": 1.1688,
      "step": 10330
    },
    {
      "epoch": 0.41895423512489616,
      "grad_norm": 0.5682399272918701,
      "learning_rate": 0.0001906063394634356,
      "loss": 1.1718,
      "step": 10340
    },
    {
      "epoch": 0.41935941330199955,
      "grad_norm": 0.7047156691551208,
      "learning_rate": 0.00019058836481873699,
      "loss": 1.1457,
      "step": 10350
    },
    {
      "epoch": 0.41976459147910294,
      "grad_norm": 0.5332230925559998,
      "learning_rate": 0.00019057037384253724,
      "loss": 1.1212,
      "step": 10360
    },
    {
      "epoch": 0.42016976965620634,
      "grad_norm": 0.5777606964111328,
      "learning_rate": 0.0001905523665380798,
      "loss": 1.1607,
      "step": 10370
    },
    {
      "epoch": 0.4205749478333097,
      "grad_norm": 0.5991753935813904,
      "learning_rate": 0.0001905343429086112,
      "loss": 1.1362,
      "step": 10380
    },
    {
      "epoch": 0.42098012601041307,
      "grad_norm": 0.5734883546829224,
      "learning_rate": 0.00019051630295738065,
      "loss": 1.1511,
      "step": 10390
    },
    {
      "epoch": 0.42138530418751646,
      "grad_norm": 0.5789864659309387,
      "learning_rate": 0.00019049824668764052,
      "loss": 1.1696,
      "step": 10400
    },
    {
      "epoch": 0.42179048236461986,
      "grad_norm": 0.546612024307251,
      "learning_rate": 0.000190480174102646,
      "loss": 1.1513,
      "step": 10410
    },
    {
      "epoch": 0.42219566054172325,
      "grad_norm": 0.5242800116539001,
      "learning_rate": 0.00019046208520565524,
      "loss": 1.1674,
      "step": 10420
    },
    {
      "epoch": 0.4226008387188266,
      "grad_norm": 0.5544593334197998,
      "learning_rate": 0.00019044397999992945,
      "loss": 1.1612,
      "step": 10430
    },
    {
      "epoch": 0.42300601689593,
      "grad_norm": 0.5510793924331665,
      "learning_rate": 0.00019042585848873256,
      "loss": 1.1622,
      "step": 10440
    },
    {
      "epoch": 0.4234111950730334,
      "grad_norm": 0.5105305910110474,
      "learning_rate": 0.00019040772067533163,
      "loss": 1.1075,
      "step": 10450
    },
    {
      "epoch": 0.42381637325013677,
      "grad_norm": 0.5856278538703918,
      "learning_rate": 0.0001903895665629966,
      "loss": 1.134,
      "step": 10460
    },
    {
      "epoch": 0.4242215514272401,
      "grad_norm": 0.5839170217514038,
      "learning_rate": 0.0001903713961550003,
      "loss": 1.1693,
      "step": 10470
    },
    {
      "epoch": 0.4246267296043435,
      "grad_norm": 0.5757611393928528,
      "learning_rate": 0.0001903532094546186,
      "loss": 1.1571,
      "step": 10480
    },
    {
      "epoch": 0.4250319077814469,
      "grad_norm": 0.5651252865791321,
      "learning_rate": 0.00019033500646513013,
      "loss": 1.0948,
      "step": 10490
    },
    {
      "epoch": 0.4254370859585503,
      "grad_norm": 0.6132938861846924,
      "learning_rate": 0.0001903167871898167,
      "loss": 1.1504,
      "step": 10500
    },
    {
      "epoch": 0.4258422641356537,
      "grad_norm": 0.502085268497467,
      "learning_rate": 0.00019029855163196283,
      "loss": 1.1428,
      "step": 10510
    },
    {
      "epoch": 0.426247442312757,
      "grad_norm": 0.5258697271347046,
      "learning_rate": 0.00019028029979485615,
      "loss": 1.1665,
      "step": 10520
    },
    {
      "epoch": 0.4266526204898604,
      "grad_norm": 0.5693614482879639,
      "learning_rate": 0.0001902620316817871,
      "loss": 1.1808,
      "step": 10530
    },
    {
      "epoch": 0.4270577986669638,
      "grad_norm": 0.6027081608772278,
      "learning_rate": 0.0001902437472960491,
      "loss": 1.1444,
      "step": 10540
    },
    {
      "epoch": 0.4274629768440672,
      "grad_norm": 0.6010106801986694,
      "learning_rate": 0.00019022544664093854,
      "loss": 1.155,
      "step": 10550
    },
    {
      "epoch": 0.42786815502117054,
      "grad_norm": 0.5967320203781128,
      "learning_rate": 0.00019020712971975465,
      "loss": 1.167,
      "step": 10560
    },
    {
      "epoch": 0.42827333319827393,
      "grad_norm": 0.5097877383232117,
      "learning_rate": 0.00019018879653579974,
      "loss": 1.1692,
      "step": 10570
    },
    {
      "epoch": 0.4286785113753773,
      "grad_norm": 0.5340242981910706,
      "learning_rate": 0.00019017044709237888,
      "loss": 1.1447,
      "step": 10580
    },
    {
      "epoch": 0.4290836895524807,
      "grad_norm": 0.530229926109314,
      "learning_rate": 0.00019015208139280018,
      "loss": 1.1579,
      "step": 10590
    },
    {
      "epoch": 0.4294888677295841,
      "grad_norm": 0.4759095311164856,
      "learning_rate": 0.00019013369944037467,
      "loss": 1.1226,
      "step": 10600
    },
    {
      "epoch": 0.42989404590668745,
      "grad_norm": 0.5764183402061462,
      "learning_rate": 0.0001901153012384163,
      "loss": 1.188,
      "step": 10610
    },
    {
      "epoch": 0.43029922408379084,
      "grad_norm": 0.5032016038894653,
      "learning_rate": 0.0001900968867902419,
      "loss": 1.109,
      "step": 10620
    },
    {
      "epoch": 0.43070440226089424,
      "grad_norm": 0.5494162440299988,
      "learning_rate": 0.00019007845609917134,
      "loss": 1.1825,
      "step": 10630
    },
    {
      "epoch": 0.43110958043799763,
      "grad_norm": 0.5227355360984802,
      "learning_rate": 0.0001900600091685273,
      "loss": 1.191,
      "step": 10640
    },
    {
      "epoch": 0.43151475861510097,
      "grad_norm": 0.5847850441932678,
      "learning_rate": 0.00019004154600163546,
      "loss": 1.1666,
      "step": 10650
    },
    {
      "epoch": 0.43191993679220436,
      "grad_norm": 0.5747153759002686,
      "learning_rate": 0.00019002306660182442,
      "loss": 1.1658,
      "step": 10660
    },
    {
      "epoch": 0.43232511496930776,
      "grad_norm": 0.5277660489082336,
      "learning_rate": 0.00019000457097242566,
      "loss": 1.139,
      "step": 10670
    },
    {
      "epoch": 0.43273029314641115,
      "grad_norm": 0.5447021722793579,
      "learning_rate": 0.0001899860591167736,
      "loss": 1.1903,
      "step": 10680
    },
    {
      "epoch": 0.43313547132351454,
      "grad_norm": 0.5871759653091431,
      "learning_rate": 0.0001899675310382057,
      "loss": 1.1535,
      "step": 10690
    },
    {
      "epoch": 0.4335406495006179,
      "grad_norm": 0.605090320110321,
      "learning_rate": 0.00018994898674006217,
      "loss": 1.1517,
      "step": 10700
    },
    {
      "epoch": 0.4339458276777213,
      "grad_norm": 0.5748393535614014,
      "learning_rate": 0.00018993042622568625,
      "loss": 1.134,
      "step": 10710
    },
    {
      "epoch": 0.43435100585482467,
      "grad_norm": 0.5717476606369019,
      "learning_rate": 0.00018991184949842404,
      "loss": 1.1399,
      "step": 10720
    },
    {
      "epoch": 0.43475618403192806,
      "grad_norm": 0.547723114490509,
      "learning_rate": 0.00018989325656162463,
      "loss": 1.1646,
      "step": 10730
    },
    {
      "epoch": 0.4351613622090314,
      "grad_norm": 0.5072046518325806,
      "learning_rate": 0.00018987464741864,
      "loss": 1.0935,
      "step": 10740
    },
    {
      "epoch": 0.4355665403861348,
      "grad_norm": 0.5630298852920532,
      "learning_rate": 0.0001898560220728251,
      "loss": 1.1862,
      "step": 10750
    },
    {
      "epoch": 0.4359717185632382,
      "grad_norm": 0.5862226486206055,
      "learning_rate": 0.00018983738052753767,
      "loss": 1.1221,
      "step": 10760
    },
    {
      "epoch": 0.4363768967403416,
      "grad_norm": 0.5352805256843567,
      "learning_rate": 0.0001898187227861385,
      "loss": 1.1218,
      "step": 10770
    },
    {
      "epoch": 0.436782074917445,
      "grad_norm": 0.535635232925415,
      "learning_rate": 0.00018980004885199125,
      "loss": 1.1473,
      "step": 10780
    },
    {
      "epoch": 0.4371872530945483,
      "grad_norm": 0.56251460313797,
      "learning_rate": 0.00018978135872846252,
      "loss": 1.1782,
      "step": 10790
    },
    {
      "epoch": 0.4375924312716517,
      "grad_norm": 0.5977115035057068,
      "learning_rate": 0.00018976265241892176,
      "loss": 1.1595,
      "step": 10800
    },
    {
      "epoch": 0.4379976094487551,
      "grad_norm": 0.6224482655525208,
      "learning_rate": 0.0001897439299267415,
      "loss": 1.14,
      "step": 10810
    },
    {
      "epoch": 0.4384027876258585,
      "grad_norm": 0.7428423762321472,
      "learning_rate": 0.00018972519125529694,
      "loss": 1.1431,
      "step": 10820
    },
    {
      "epoch": 0.43880796580296183,
      "grad_norm": 0.5714390873908997,
      "learning_rate": 0.00018970643640796642,
      "loss": 1.1943,
      "step": 10830
    },
    {
      "epoch": 0.4392131439800652,
      "grad_norm": 0.534583568572998,
      "learning_rate": 0.00018968766538813108,
      "loss": 1.1752,
      "step": 10840
    },
    {
      "epoch": 0.4396183221571686,
      "grad_norm": 0.5747492909431458,
      "learning_rate": 0.00018966887819917504,
      "loss": 1.1488,
      "step": 10850
    },
    {
      "epoch": 0.440023500334272,
      "grad_norm": 0.5224570035934448,
      "learning_rate": 0.0001896500748444853,
      "loss": 1.1672,
      "step": 10860
    },
    {
      "epoch": 0.4404286785113754,
      "grad_norm": 0.5559102892875671,
      "learning_rate": 0.00018963125532745173,
      "loss": 1.137,
      "step": 10870
    },
    {
      "epoch": 0.44083385668847874,
      "grad_norm": 0.5638408660888672,
      "learning_rate": 0.00018961241965146722,
      "loss": 1.1735,
      "step": 10880
    },
    {
      "epoch": 0.44123903486558214,
      "grad_norm": 0.5326826572418213,
      "learning_rate": 0.00018959356781992744,
      "loss": 1.1123,
      "step": 10890
    },
    {
      "epoch": 0.44164421304268553,
      "grad_norm": 0.6280413866043091,
      "learning_rate": 0.00018957469983623112,
      "loss": 1.1779,
      "step": 10900
    },
    {
      "epoch": 0.4420493912197889,
      "grad_norm": 0.5057059526443481,
      "learning_rate": 0.0001895558157037798,
      "loss": 1.1838,
      "step": 10910
    },
    {
      "epoch": 0.44245456939689226,
      "grad_norm": 0.5783162713050842,
      "learning_rate": 0.00018953691542597796,
      "loss": 1.1881,
      "step": 10920
    },
    {
      "epoch": 0.44285974757399565,
      "grad_norm": 0.44804856181144714,
      "learning_rate": 0.00018951799900623298,
      "loss": 1.1753,
      "step": 10930
    },
    {
      "epoch": 0.44326492575109905,
      "grad_norm": 0.6011568307876587,
      "learning_rate": 0.0001894990664479552,
      "loss": 1.1631,
      "step": 10940
    },
    {
      "epoch": 0.44367010392820244,
      "grad_norm": 0.6212189793586731,
      "learning_rate": 0.00018948011775455774,
      "loss": 1.1654,
      "step": 10950
    },
    {
      "epoch": 0.44407528210530584,
      "grad_norm": 0.5510669350624084,
      "learning_rate": 0.0001894611529294568,
      "loss": 1.1364,
      "step": 10960
    },
    {
      "epoch": 0.4444804602824092,
      "grad_norm": 0.5580452084541321,
      "learning_rate": 0.0001894421719760714,
      "loss": 1.1596,
      "step": 10970
    },
    {
      "epoch": 0.44488563845951257,
      "grad_norm": 0.5483415722846985,
      "learning_rate": 0.00018942317489782347,
      "loss": 1.1623,
      "step": 10980
    },
    {
      "epoch": 0.44529081663661596,
      "grad_norm": 0.5082122087478638,
      "learning_rate": 0.0001894041616981378,
      "loss": 1.1269,
      "step": 10990
    },
    {
      "epoch": 0.44569599481371935,
      "grad_norm": 0.6066364645957947,
      "learning_rate": 0.0001893851323804422,
      "loss": 1.155,
      "step": 11000
    },
    {
      "epoch": 0.4461011729908227,
      "grad_norm": 0.5748014450073242,
      "learning_rate": 0.0001893660869481673,
      "loss": 1.1668,
      "step": 11010
    },
    {
      "epoch": 0.4465063511679261,
      "grad_norm": 0.5032930374145508,
      "learning_rate": 0.00018934702540474664,
      "loss": 1.1093,
      "step": 11020
    },
    {
      "epoch": 0.4469115293450295,
      "grad_norm": 0.6356216669082642,
      "learning_rate": 0.00018932794775361676,
      "loss": 1.1864,
      "step": 11030
    },
    {
      "epoch": 0.4473167075221329,
      "grad_norm": 0.5513843297958374,
      "learning_rate": 0.00018930885399821693,
      "loss": 1.1757,
      "step": 11040
    },
    {
      "epoch": 0.44772188569923627,
      "grad_norm": 0.543331503868103,
      "learning_rate": 0.00018928974414198947,
      "loss": 1.145,
      "step": 11050
    },
    {
      "epoch": 0.4481270638763396,
      "grad_norm": 0.5400907397270203,
      "learning_rate": 0.00018927061818837954,
      "loss": 1.1747,
      "step": 11060
    },
    {
      "epoch": 0.448532242053443,
      "grad_norm": 0.5370875597000122,
      "learning_rate": 0.00018925147614083524,
      "loss": 1.162,
      "step": 11070
    },
    {
      "epoch": 0.4489374202305464,
      "grad_norm": 0.5826300978660583,
      "learning_rate": 0.00018923231800280755,
      "loss": 1.1321,
      "step": 11080
    },
    {
      "epoch": 0.4493425984076498,
      "grad_norm": 0.580161988735199,
      "learning_rate": 0.0001892131437777503,
      "loss": 1.155,
      "step": 11090
    },
    {
      "epoch": 0.4497477765847531,
      "grad_norm": 0.5740492343902588,
      "learning_rate": 0.00018919395346912027,
      "loss": 1.1625,
      "step": 11100
    },
    {
      "epoch": 0.4501529547618565,
      "grad_norm": 0.5467064380645752,
      "learning_rate": 0.00018917474708037718,
      "loss": 1.1857,
      "step": 11110
    },
    {
      "epoch": 0.4505581329389599,
      "grad_norm": 0.6519179344177246,
      "learning_rate": 0.0001891555246149836,
      "loss": 1.1414,
      "step": 11120
    },
    {
      "epoch": 0.4509633111160633,
      "grad_norm": 0.5909104347229004,
      "learning_rate": 0.00018913628607640498,
      "loss": 1.174,
      "step": 11130
    },
    {
      "epoch": 0.4513684892931667,
      "grad_norm": 0.5740471482276917,
      "learning_rate": 0.00018911703146810976,
      "loss": 1.1421,
      "step": 11140
    },
    {
      "epoch": 0.45177366747027004,
      "grad_norm": 0.5573481917381287,
      "learning_rate": 0.00018909776079356912,
      "loss": 1.1487,
      "step": 11150
    },
    {
      "epoch": 0.45217884564737343,
      "grad_norm": 0.5617687702178955,
      "learning_rate": 0.00018907847405625724,
      "loss": 1.1855,
      "step": 11160
    },
    {
      "epoch": 0.4525840238244768,
      "grad_norm": 0.5033039450645447,
      "learning_rate": 0.00018905917125965123,
      "loss": 1.1485,
      "step": 11170
    },
    {
      "epoch": 0.4529892020015802,
      "grad_norm": 0.5481246709823608,
      "learning_rate": 0.00018903985240723104,
      "loss": 1.1278,
      "step": 11180
    },
    {
      "epoch": 0.45339438017868355,
      "grad_norm": 0.5969385504722595,
      "learning_rate": 0.00018902051750247946,
      "loss": 1.1417,
      "step": 11190
    },
    {
      "epoch": 0.45379955835578695,
      "grad_norm": 0.5970520377159119,
      "learning_rate": 0.00018900116654888233,
      "loss": 1.1385,
      "step": 11200
    },
    {
      "epoch": 0.45420473653289034,
      "grad_norm": 0.5434677600860596,
      "learning_rate": 0.0001889817995499282,
      "loss": 1.1581,
      "step": 11210
    },
    {
      "epoch": 0.45460991470999373,
      "grad_norm": 0.5101468563079834,
      "learning_rate": 0.00018896241650910865,
      "loss": 1.1507,
      "step": 11220
    },
    {
      "epoch": 0.45501509288709713,
      "grad_norm": 0.5459298491477966,
      "learning_rate": 0.00018894301742991814,
      "loss": 1.1606,
      "step": 11230
    },
    {
      "epoch": 0.45542027106420047,
      "grad_norm": 0.5552728176116943,
      "learning_rate": 0.0001889236023158539,
      "loss": 1.131,
      "step": 11240
    },
    {
      "epoch": 0.45582544924130386,
      "grad_norm": 0.5012040734291077,
      "learning_rate": 0.00018890417117041619,
      "loss": 1.1421,
      "step": 11250
    },
    {
      "epoch": 0.45623062741840725,
      "grad_norm": 0.6059371829032898,
      "learning_rate": 0.0001888847239971081,
      "loss": 1.1559,
      "step": 11260
    },
    {
      "epoch": 0.45663580559551065,
      "grad_norm": 0.5233181715011597,
      "learning_rate": 0.00018886526079943557,
      "loss": 1.158,
      "step": 11270
    },
    {
      "epoch": 0.457040983772614,
      "grad_norm": 0.555776059627533,
      "learning_rate": 0.00018884578158090756,
      "loss": 1.1675,
      "step": 11280
    },
    {
      "epoch": 0.4574461619497174,
      "grad_norm": 0.515190064907074,
      "learning_rate": 0.00018882628634503577,
      "loss": 1.1291,
      "step": 11290
    },
    {
      "epoch": 0.45785134012682077,
      "grad_norm": 0.594858705997467,
      "learning_rate": 0.0001888067750953349,
      "loss": 1.1521,
      "step": 11300
    },
    {
      "epoch": 0.45825651830392417,
      "grad_norm": 0.5134510397911072,
      "learning_rate": 0.0001887872478353224,
      "loss": 1.1901,
      "step": 11310
    },
    {
      "epoch": 0.45866169648102756,
      "grad_norm": 0.531812846660614,
      "learning_rate": 0.00018876770456851877,
      "loss": 1.1325,
      "step": 11320
    },
    {
      "epoch": 0.4590668746581309,
      "grad_norm": 0.5649051070213318,
      "learning_rate": 0.00018874814529844732,
      "loss": 1.1058,
      "step": 11330
    },
    {
      "epoch": 0.4594720528352343,
      "grad_norm": 0.5952717065811157,
      "learning_rate": 0.00018872857002863423,
      "loss": 1.1771,
      "step": 11340
    },
    {
      "epoch": 0.4598772310123377,
      "grad_norm": 0.5984222888946533,
      "learning_rate": 0.00018870897876260856,
      "loss": 1.1546,
      "step": 11350
    },
    {
      "epoch": 0.4602824091894411,
      "grad_norm": 0.5528778433799744,
      "learning_rate": 0.0001886893715039023,
      "loss": 1.1533,
      "step": 11360
    },
    {
      "epoch": 0.4606875873665444,
      "grad_norm": 0.5642794370651245,
      "learning_rate": 0.00018866974825605025,
      "loss": 1.1429,
      "step": 11370
    },
    {
      "epoch": 0.4610927655436478,
      "grad_norm": 0.5697764754295349,
      "learning_rate": 0.0001886501090225902,
      "loss": 1.1763,
      "step": 11380
    },
    {
      "epoch": 0.4614979437207512,
      "grad_norm": 0.6042892932891846,
      "learning_rate": 0.00018863045380706274,
      "loss": 1.1265,
      "step": 11390
    },
    {
      "epoch": 0.4619031218978546,
      "grad_norm": 0.5587393641471863,
      "learning_rate": 0.00018861078261301134,
      "loss": 1.1355,
      "step": 11400
    },
    {
      "epoch": 0.462308300074958,
      "grad_norm": 0.6207725405693054,
      "learning_rate": 0.0001885910954439824,
      "loss": 1.145,
      "step": 11410
    },
    {
      "epoch": 0.4627134782520613,
      "grad_norm": 0.5681396126747131,
      "learning_rate": 0.00018857139230352518,
      "loss": 1.1588,
      "step": 11420
    },
    {
      "epoch": 0.4631186564291647,
      "grad_norm": 0.6044288873672485,
      "learning_rate": 0.00018855167319519178,
      "loss": 1.1373,
      "step": 11430
    },
    {
      "epoch": 0.4635238346062681,
      "grad_norm": 0.6252747774124146,
      "learning_rate": 0.00018853193812253723,
      "loss": 1.147,
      "step": 11440
    },
    {
      "epoch": 0.4639290127833715,
      "grad_norm": 0.6267377138137817,
      "learning_rate": 0.00018851218708911942,
      "loss": 1.1543,
      "step": 11450
    },
    {
      "epoch": 0.46433419096047485,
      "grad_norm": 0.5742064118385315,
      "learning_rate": 0.0001884924200984991,
      "loss": 1.1752,
      "step": 11460
    },
    {
      "epoch": 0.46473936913757824,
      "grad_norm": 0.5940319299697876,
      "learning_rate": 0.00018847263715423993,
      "loss": 1.1632,
      "step": 11470
    },
    {
      "epoch": 0.46514454731468163,
      "grad_norm": 0.5192302465438843,
      "learning_rate": 0.00018845283825990843,
      "loss": 1.1408,
      "step": 11480
    },
    {
      "epoch": 0.465549725491785,
      "grad_norm": 0.5851195454597473,
      "learning_rate": 0.00018843302341907398,
      "loss": 1.1702,
      "step": 11490
    },
    {
      "epoch": 0.4659549036688884,
      "grad_norm": 0.5969977378845215,
      "learning_rate": 0.00018841319263530887,
      "loss": 1.2022,
      "step": 11500
    },
    {
      "epoch": 0.46636008184599176,
      "grad_norm": 0.5220630168914795,
      "learning_rate": 0.00018839334591218824,
      "loss": 1.1631,
      "step": 11510
    },
    {
      "epoch": 0.46676526002309515,
      "grad_norm": 0.5503666996955872,
      "learning_rate": 0.0001883734832532901,
      "loss": 1.1105,
      "step": 11520
    },
    {
      "epoch": 0.46717043820019855,
      "grad_norm": 0.6389178037643433,
      "learning_rate": 0.00018835360466219533,
      "loss": 1.2019,
      "step": 11530
    },
    {
      "epoch": 0.46757561637730194,
      "grad_norm": 0.6547725200653076,
      "learning_rate": 0.0001883337101424877,
      "loss": 1.1093,
      "step": 11540
    },
    {
      "epoch": 0.4679807945544053,
      "grad_norm": 0.5708798170089722,
      "learning_rate": 0.00018831379969775388,
      "loss": 1.1512,
      "step": 11550
    },
    {
      "epoch": 0.46838597273150867,
      "grad_norm": 0.5425140261650085,
      "learning_rate": 0.00018829387333158334,
      "loss": 1.1708,
      "step": 11560
    },
    {
      "epoch": 0.46879115090861206,
      "grad_norm": 0.6079493761062622,
      "learning_rate": 0.00018827393104756845,
      "loss": 1.1671,
      "step": 11570
    },
    {
      "epoch": 0.46919632908571546,
      "grad_norm": 0.5140082240104675,
      "learning_rate": 0.0001882539728493045,
      "loss": 1.1715,
      "step": 11580
    },
    {
      "epoch": 0.46960150726281885,
      "grad_norm": 0.5321645736694336,
      "learning_rate": 0.00018823399874038954,
      "loss": 1.1474,
      "step": 11590
    },
    {
      "epoch": 0.4700066854399222,
      "grad_norm": 0.562795102596283,
      "learning_rate": 0.00018821400872442458,
      "loss": 1.1717,
      "step": 11600
    },
    {
      "epoch": 0.4704118636170256,
      "grad_norm": 0.5429909825325012,
      "learning_rate": 0.0001881940028050135,
      "loss": 1.142,
      "step": 11610
    },
    {
      "epoch": 0.470817041794129,
      "grad_norm": 0.5278061032295227,
      "learning_rate": 0.000188173980985763,
      "loss": 1.1562,
      "step": 11620
    },
    {
      "epoch": 0.47122221997123237,
      "grad_norm": 0.48562783002853394,
      "learning_rate": 0.00018815394327028262,
      "loss": 1.1954,
      "step": 11630
    },
    {
      "epoch": 0.4716273981483357,
      "grad_norm": 0.5681062936782837,
      "learning_rate": 0.00018813388966218487,
      "loss": 1.1628,
      "step": 11640
    },
    {
      "epoch": 0.4720325763254391,
      "grad_norm": 0.6161751747131348,
      "learning_rate": 0.00018811382016508508,
      "loss": 1.1838,
      "step": 11650
    },
    {
      "epoch": 0.4724377545025425,
      "grad_norm": 0.500204861164093,
      "learning_rate": 0.00018809373478260137,
      "loss": 1.153,
      "step": 11660
    },
    {
      "epoch": 0.4728429326796459,
      "grad_norm": 0.5751171708106995,
      "learning_rate": 0.0001880736335183548,
      "loss": 1.13,
      "step": 11670
    },
    {
      "epoch": 0.4732481108567493,
      "grad_norm": 0.5370206832885742,
      "learning_rate": 0.0001880535163759693,
      "loss": 1.1058,
      "step": 11680
    },
    {
      "epoch": 0.4736532890338526,
      "grad_norm": 0.5846206545829773,
      "learning_rate": 0.00018803338335907164,
      "loss": 1.1469,
      "step": 11690
    },
    {
      "epoch": 0.474058467210956,
      "grad_norm": 0.5453063249588013,
      "learning_rate": 0.0001880132344712914,
      "loss": 1.1423,
      "step": 11700
    },
    {
      "epoch": 0.4744636453880594,
      "grad_norm": 0.5329487323760986,
      "learning_rate": 0.00018799306971626115,
      "loss": 1.152,
      "step": 11710
    },
    {
      "epoch": 0.4748688235651628,
      "grad_norm": 0.5277760028839111,
      "learning_rate": 0.00018797288909761618,
      "loss": 1.1767,
      "step": 11720
    },
    {
      "epoch": 0.47527400174226614,
      "grad_norm": 0.6098132133483887,
      "learning_rate": 0.00018795269261899473,
      "loss": 1.1226,
      "step": 11730
    },
    {
      "epoch": 0.47567917991936953,
      "grad_norm": 0.5511724352836609,
      "learning_rate": 0.00018793248028403788,
      "loss": 1.1372,
      "step": 11740
    },
    {
      "epoch": 0.4760843580964729,
      "grad_norm": 0.5849943161010742,
      "learning_rate": 0.00018791225209638957,
      "loss": 1.129,
      "step": 11750
    },
    {
      "epoch": 0.4764895362735763,
      "grad_norm": 0.6021835803985596,
      "learning_rate": 0.00018789200805969657,
      "loss": 1.1556,
      "step": 11760
    },
    {
      "epoch": 0.4768947144506797,
      "grad_norm": 0.4975357949733734,
      "learning_rate": 0.00018787174817760848,
      "loss": 1.1473,
      "step": 11770
    },
    {
      "epoch": 0.47729989262778305,
      "grad_norm": 0.5573726892471313,
      "learning_rate": 0.00018785147245377787,
      "loss": 1.1751,
      "step": 11780
    },
    {
      "epoch": 0.47770507080488644,
      "grad_norm": 0.5865565538406372,
      "learning_rate": 0.0001878311808918601,
      "loss": 1.1249,
      "step": 11790
    },
    {
      "epoch": 0.47811024898198984,
      "grad_norm": 0.5553157925605774,
      "learning_rate": 0.00018781087349551334,
      "loss": 1.1664,
      "step": 11800
    },
    {
      "epoch": 0.47851542715909323,
      "grad_norm": 0.6444417834281921,
      "learning_rate": 0.00018779055026839868,
      "loss": 1.163,
      "step": 11810
    },
    {
      "epoch": 0.47892060533619657,
      "grad_norm": 0.5064760446548462,
      "learning_rate": 0.00018777021121418007,
      "loss": 1.1548,
      "step": 11820
    },
    {
      "epoch": 0.47932578351329996,
      "grad_norm": 0.5896313786506653,
      "learning_rate": 0.00018774985633652425,
      "loss": 1.1386,
      "step": 11830
    },
    {
      "epoch": 0.47973096169040336,
      "grad_norm": 0.5431138277053833,
      "learning_rate": 0.00018772948563910088,
      "loss": 1.1459,
      "step": 11840
    },
    {
      "epoch": 0.48013613986750675,
      "grad_norm": 0.58415687084198,
      "learning_rate": 0.0001877090991255824,
      "loss": 1.1439,
      "step": 11850
    },
    {
      "epoch": 0.48054131804461014,
      "grad_norm": 0.5284492373466492,
      "learning_rate": 0.00018768869679964417,
      "loss": 1.1696,
      "step": 11860
    },
    {
      "epoch": 0.4809464962217135,
      "grad_norm": 0.6304959654808044,
      "learning_rate": 0.0001876682786649644,
      "loss": 1.1534,
      "step": 11870
    },
    {
      "epoch": 0.4813516743988169,
      "grad_norm": 0.55653977394104,
      "learning_rate": 0.00018764784472522403,
      "loss": 1.1301,
      "step": 11880
    },
    {
      "epoch": 0.48175685257592027,
      "grad_norm": 0.5565120577812195,
      "learning_rate": 0.00018762739498410705,
      "loss": 1.1625,
      "step": 11890
    },
    {
      "epoch": 0.48216203075302366,
      "grad_norm": 0.5158827304840088,
      "learning_rate": 0.00018760692944530015,
      "loss": 1.1396,
      "step": 11900
    },
    {
      "epoch": 0.482567208930127,
      "grad_norm": 0.5045274496078491,
      "learning_rate": 0.0001875864481124929,
      "loss": 1.1369,
      "step": 11910
    },
    {
      "epoch": 0.4829723871072304,
      "grad_norm": 0.5333916544914246,
      "learning_rate": 0.00018756595098937774,
      "loss": 1.1084,
      "step": 11920
    },
    {
      "epoch": 0.4833775652843338,
      "grad_norm": 0.5901340246200562,
      "learning_rate": 0.00018754543807964987,
      "loss": 1.1332,
      "step": 11930
    },
    {
      "epoch": 0.4837827434614372,
      "grad_norm": 0.5934543609619141,
      "learning_rate": 0.00018752490938700754,
      "loss": 1.1692,
      "step": 11940
    },
    {
      "epoch": 0.4841879216385406,
      "grad_norm": 0.5351195335388184,
      "learning_rate": 0.00018750436491515163,
      "loss": 1.1728,
      "step": 11950
    },
    {
      "epoch": 0.4845930998156439,
      "grad_norm": 0.5597553253173828,
      "learning_rate": 0.000187483804667786,
      "loss": 1.1659,
      "step": 11960
    },
    {
      "epoch": 0.4849982779927473,
      "grad_norm": 0.5412261486053467,
      "learning_rate": 0.00018746322864861728,
      "loss": 1.1879,
      "step": 11970
    },
    {
      "epoch": 0.4854034561698507,
      "grad_norm": 0.7172391414642334,
      "learning_rate": 0.00018744263686135492,
      "loss": 1.1293,
      "step": 11980
    },
    {
      "epoch": 0.4858086343469541,
      "grad_norm": 0.6168282628059387,
      "learning_rate": 0.0001874220293097113,
      "loss": 1.1504,
      "step": 11990
    },
    {
      "epoch": 0.48621381252405743,
      "grad_norm": 0.5493648052215576,
      "learning_rate": 0.00018740140599740162,
      "loss": 1.109,
      "step": 12000
    },
    {
      "epoch": 0.4866189907011608,
      "grad_norm": 0.5738518238067627,
      "learning_rate": 0.00018738076692814391,
      "loss": 1.1427,
      "step": 12010
    },
    {
      "epoch": 0.4870241688782642,
      "grad_norm": 0.5785367488861084,
      "learning_rate": 0.00018736011210565898,
      "loss": 1.1267,
      "step": 12020
    },
    {
      "epoch": 0.4874293470553676,
      "grad_norm": 0.5781382918357849,
      "learning_rate": 0.0001873394415336706,
      "loss": 1.2067,
      "step": 12030
    },
    {
      "epoch": 0.487834525232471,
      "grad_norm": 0.651322603225708,
      "learning_rate": 0.0001873187552159052,
      "loss": 1.1354,
      "step": 12040
    },
    {
      "epoch": 0.48823970340957434,
      "grad_norm": 0.5654751062393188,
      "learning_rate": 0.0001872980531560923,
      "loss": 1.1488,
      "step": 12050
    },
    {
      "epoch": 0.48864488158667774,
      "grad_norm": 0.5544155836105347,
      "learning_rate": 0.00018727733535796404,
      "loss": 1.1247,
      "step": 12060
    },
    {
      "epoch": 0.48905005976378113,
      "grad_norm": 0.5495392680168152,
      "learning_rate": 0.00018725660182525552,
      "loss": 1.1662,
      "step": 12070
    },
    {
      "epoch": 0.4894552379408845,
      "grad_norm": 0.5374956727027893,
      "learning_rate": 0.00018723585256170462,
      "loss": 1.1726,
      "step": 12080
    },
    {
      "epoch": 0.48986041611798786,
      "grad_norm": 0.5336000323295593,
      "learning_rate": 0.00018721508757105202,
      "loss": 1.1507,
      "step": 12090
    },
    {
      "epoch": 0.49026559429509126,
      "grad_norm": 0.5221179723739624,
      "learning_rate": 0.00018719430685704136,
      "loss": 1.1518,
      "step": 12100
    },
    {
      "epoch": 0.49067077247219465,
      "grad_norm": 0.568057656288147,
      "learning_rate": 0.000187173510423419,
      "loss": 1.1905,
      "step": 12110
    },
    {
      "epoch": 0.49107595064929804,
      "grad_norm": 0.6243038773536682,
      "learning_rate": 0.00018715269827393414,
      "loss": 1.1587,
      "step": 12120
    },
    {
      "epoch": 0.49148112882640144,
      "grad_norm": 0.6265568137168884,
      "learning_rate": 0.00018713187041233896,
      "loss": 1.1598,
      "step": 12130
    },
    {
      "epoch": 0.4918863070035048,
      "grad_norm": 0.5293442010879517,
      "learning_rate": 0.00018711102684238823,
      "loss": 1.1348,
      "step": 12140
    },
    {
      "epoch": 0.49229148518060817,
      "grad_norm": 0.6181290149688721,
      "learning_rate": 0.00018709016756783976,
      "loss": 1.1798,
      "step": 12150
    },
    {
      "epoch": 0.49269666335771156,
      "grad_norm": 0.7162533402442932,
      "learning_rate": 0.0001870692925924541,
      "loss": 1.1581,
      "step": 12160
    },
    {
      "epoch": 0.49310184153481496,
      "grad_norm": 0.5390025973320007,
      "learning_rate": 0.00018704840191999464,
      "loss": 1.1803,
      "step": 12170
    },
    {
      "epoch": 0.4935070197119183,
      "grad_norm": 0.6099391579627991,
      "learning_rate": 0.00018702749555422757,
      "loss": 1.155,
      "step": 12180
    },
    {
      "epoch": 0.4939121978890217,
      "grad_norm": 0.5739234089851379,
      "learning_rate": 0.00018700657349892202,
      "loss": 1.1659,
      "step": 12190
    },
    {
      "epoch": 0.4943173760661251,
      "grad_norm": 0.605988085269928,
      "learning_rate": 0.0001869856357578498,
      "loss": 1.1976,
      "step": 12200
    },
    {
      "epoch": 0.4947225542432285,
      "grad_norm": 0.5756819248199463,
      "learning_rate": 0.00018696468233478565,
      "loss": 1.1763,
      "step": 12210
    },
    {
      "epoch": 0.49512773242033187,
      "grad_norm": 0.577950656414032,
      "learning_rate": 0.0001869437132335071,
      "loss": 1.1647,
      "step": 12220
    },
    {
      "epoch": 0.4955329105974352,
      "grad_norm": 0.5449633598327637,
      "learning_rate": 0.00018692272845779448,
      "loss": 1.1568,
      "step": 12230
    },
    {
      "epoch": 0.4959380887745386,
      "grad_norm": 0.5313047766685486,
      "learning_rate": 0.00018690172801143104,
      "loss": 1.1394,
      "step": 12240
    },
    {
      "epoch": 0.496343266951642,
      "grad_norm": 0.6299227476119995,
      "learning_rate": 0.00018688071189820277,
      "loss": 1.1574,
      "step": 12250
    },
    {
      "epoch": 0.4967484451287454,
      "grad_norm": 0.5586907267570496,
      "learning_rate": 0.00018685968012189852,
      "loss": 1.1451,
      "step": 12260
    },
    {
      "epoch": 0.4971536233058487,
      "grad_norm": 0.6201167106628418,
      "learning_rate": 0.0001868386326863099,
      "loss": 1.1597,
      "step": 12270
    },
    {
      "epoch": 0.4975588014829521,
      "grad_norm": 0.5703408122062683,
      "learning_rate": 0.00018681756959523148,
      "loss": 1.1857,
      "step": 12280
    },
    {
      "epoch": 0.4979639796600555,
      "grad_norm": 0.5387154817581177,
      "learning_rate": 0.00018679649085246046,
      "loss": 1.1802,
      "step": 12290
    },
    {
      "epoch": 0.4983691578371589,
      "grad_norm": 0.5064283013343811,
      "learning_rate": 0.00018677539646179707,
      "loss": 1.1466,
      "step": 12300
    },
    {
      "epoch": 0.4987743360142623,
      "grad_norm": 0.6894688010215759,
      "learning_rate": 0.00018675428642704421,
      "loss": 1.1461,
      "step": 12310
    },
    {
      "epoch": 0.49917951419136564,
      "grad_norm": 0.5294455289840698,
      "learning_rate": 0.00018673316075200767,
      "loss": 1.1452,
      "step": 12320
    },
    {
      "epoch": 0.49958469236846903,
      "grad_norm": 0.5672134757041931,
      "learning_rate": 0.00018671201944049602,
      "loss": 1.1497,
      "step": 12330
    },
    {
      "epoch": 0.4999898705455724,
      "grad_norm": 0.5413781404495239,
      "learning_rate": 0.0001866908624963207,
      "loss": 1.1717,
      "step": 12340
    },
    {
      "epoch": 0.5003950487226758,
      "grad_norm": 0.5520235300064087,
      "learning_rate": 0.0001866696899232959,
      "loss": 1.1761,
      "step": 12350
    },
    {
      "epoch": 0.5008002268997792,
      "grad_norm": 0.6337764859199524,
      "learning_rate": 0.00018664850172523867,
      "loss": 1.1852,
      "step": 12360
    },
    {
      "epoch": 0.5012054050768826,
      "grad_norm": 0.5449768304824829,
      "learning_rate": 0.00018662729790596888,
      "loss": 1.1257,
      "step": 12370
    },
    {
      "epoch": 0.5016105832539859,
      "grad_norm": 0.5378426313400269,
      "learning_rate": 0.00018660607846930925,
      "loss": 1.1484,
      "step": 12380
    },
    {
      "epoch": 0.5020157614310893,
      "grad_norm": 0.575812041759491,
      "learning_rate": 0.00018658484341908525,
      "loss": 1.1805,
      "step": 12390
    },
    {
      "epoch": 0.5024209396081927,
      "grad_norm": 0.5462254285812378,
      "learning_rate": 0.00018656359275912515,
      "loss": 1.1361,
      "step": 12400
    },
    {
      "epoch": 0.5028261177852961,
      "grad_norm": 0.6146267056465149,
      "learning_rate": 0.0001865423264932601,
      "loss": 1.2038,
      "step": 12410
    },
    {
      "epoch": 0.5032312959623995,
      "grad_norm": 0.5097014904022217,
      "learning_rate": 0.00018652104462532404,
      "loss": 1.0941,
      "step": 12420
    },
    {
      "epoch": 0.5036364741395029,
      "grad_norm": 0.6515002250671387,
      "learning_rate": 0.0001864997471591537,
      "loss": 1.1839,
      "step": 12430
    },
    {
      "epoch": 0.5040416523166062,
      "grad_norm": 0.5486054420471191,
      "learning_rate": 0.00018647843409858869,
      "loss": 1.1699,
      "step": 12440
    },
    {
      "epoch": 0.5044468304937096,
      "grad_norm": 0.5791358351707458,
      "learning_rate": 0.00018645710544747132,
      "loss": 1.1497,
      "step": 12450
    },
    {
      "epoch": 0.504852008670813,
      "grad_norm": 0.5970382690429688,
      "learning_rate": 0.00018643576120964684,
      "loss": 1.1777,
      "step": 12460
    },
    {
      "epoch": 0.5052571868479163,
      "grad_norm": 0.5536314249038696,
      "learning_rate": 0.00018641440138896317,
      "loss": 1.1829,
      "step": 12470
    },
    {
      "epoch": 0.5056623650250197,
      "grad_norm": 0.5072109699249268,
      "learning_rate": 0.00018639302598927118,
      "loss": 1.1226,
      "step": 12480
    },
    {
      "epoch": 0.5060675432021231,
      "grad_norm": 0.5679654479026794,
      "learning_rate": 0.00018637163501442446,
      "loss": 1.1694,
      "step": 12490
    },
    {
      "epoch": 0.5064727213792265,
      "grad_norm": 0.5383331179618835,
      "learning_rate": 0.00018635022846827937,
      "loss": 1.1542,
      "step": 12500
    },
    {
      "epoch": 0.5068778995563299,
      "grad_norm": 0.5385815501213074,
      "learning_rate": 0.00018632880635469526,
      "loss": 1.153,
      "step": 12510
    },
    {
      "epoch": 0.5072830777334333,
      "grad_norm": 0.557794451713562,
      "learning_rate": 0.00018630736867753403,
      "loss": 1.1667,
      "step": 12520
    },
    {
      "epoch": 0.5076882559105367,
      "grad_norm": 0.5728976726531982,
      "learning_rate": 0.0001862859154406606,
      "loss": 1.1456,
      "step": 12530
    },
    {
      "epoch": 0.5080934340876401,
      "grad_norm": 0.6270129084587097,
      "learning_rate": 0.0001862644466479426,
      "loss": 1.1579,
      "step": 12540
    },
    {
      "epoch": 0.5084986122647435,
      "grad_norm": 0.5573577880859375,
      "learning_rate": 0.0001862429623032505,
      "loss": 1.1482,
      "step": 12550
    },
    {
      "epoch": 0.5089037904418467,
      "grad_norm": 0.5657044053077698,
      "learning_rate": 0.00018622146241045748,
      "loss": 1.1526,
      "step": 12560
    },
    {
      "epoch": 0.5093089686189501,
      "grad_norm": 0.4928171634674072,
      "learning_rate": 0.00018619994697343964,
      "loss": 1.1296,
      "step": 12570
    },
    {
      "epoch": 0.5097141467960535,
      "grad_norm": 0.6100425720214844,
      "learning_rate": 0.00018617841599607586,
      "loss": 1.1696,
      "step": 12580
    },
    {
      "epoch": 0.5101193249731569,
      "grad_norm": 0.5447854995727539,
      "learning_rate": 0.0001861568694822478,
      "loss": 1.153,
      "step": 12590
    },
    {
      "epoch": 0.5105245031502603,
      "grad_norm": 0.5850094556808472,
      "learning_rate": 0.00018613530743583987,
      "loss": 1.1509,
      "step": 12600
    },
    {
      "epoch": 0.5109296813273637,
      "grad_norm": 0.5541381239891052,
      "learning_rate": 0.00018611372986073935,
      "loss": 1.1481,
      "step": 12610
    },
    {
      "epoch": 0.5113348595044671,
      "grad_norm": 0.5685286521911621,
      "learning_rate": 0.00018609213676083633,
      "loss": 1.1391,
      "step": 12620
    },
    {
      "epoch": 0.5117400376815705,
      "grad_norm": 0.5182244777679443,
      "learning_rate": 0.00018607052814002367,
      "loss": 1.1592,
      "step": 12630
    },
    {
      "epoch": 0.5121452158586739,
      "grad_norm": 0.5227952599525452,
      "learning_rate": 0.00018604890400219696,
      "loss": 1.1369,
      "step": 12640
    },
    {
      "epoch": 0.5125503940357772,
      "grad_norm": 0.5442817211151123,
      "learning_rate": 0.00018602726435125474,
      "loss": 1.1552,
      "step": 12650
    },
    {
      "epoch": 0.5129555722128806,
      "grad_norm": 0.5654439926147461,
      "learning_rate": 0.0001860056091910982,
      "loss": 1.1286,
      "step": 12660
    },
    {
      "epoch": 0.513360750389984,
      "grad_norm": 0.6362553238868713,
      "learning_rate": 0.0001859839385256314,
      "loss": 1.1652,
      "step": 12670
    },
    {
      "epoch": 0.5137659285670874,
      "grad_norm": 0.5939968228340149,
      "learning_rate": 0.00018596225235876119,
      "loss": 1.167,
      "step": 12680
    },
    {
      "epoch": 0.5141711067441908,
      "grad_norm": 0.5879738926887512,
      "learning_rate": 0.00018594055069439725,
      "loss": 1.1829,
      "step": 12690
    },
    {
      "epoch": 0.5145762849212941,
      "grad_norm": 0.5609286427497864,
      "learning_rate": 0.00018591883353645194,
      "loss": 1.152,
      "step": 12700
    },
    {
      "epoch": 0.5149814630983975,
      "grad_norm": 0.5816230177879333,
      "learning_rate": 0.00018589710088884052,
      "loss": 1.1442,
      "step": 12710
    },
    {
      "epoch": 0.5153866412755009,
      "grad_norm": 0.7072398662567139,
      "learning_rate": 0.000185875352755481,
      "loss": 1.1159,
      "step": 12720
    },
    {
      "epoch": 0.5157918194526043,
      "grad_norm": 0.5122026801109314,
      "learning_rate": 0.0001858535891402942,
      "loss": 1.1245,
      "step": 12730
    },
    {
      "epoch": 0.5161969976297076,
      "grad_norm": 0.5479681491851807,
      "learning_rate": 0.0001858318100472037,
      "loss": 1.1676,
      "step": 12740
    },
    {
      "epoch": 0.516602175806811,
      "grad_norm": 0.6284963488578796,
      "learning_rate": 0.0001858100154801359,
      "loss": 1.168,
      "step": 12750
    },
    {
      "epoch": 0.5170073539839144,
      "grad_norm": 0.625220537185669,
      "learning_rate": 0.00018578820544302,
      "loss": 1.1407,
      "step": 12760
    },
    {
      "epoch": 0.5174125321610178,
      "grad_norm": 0.5802189707756042,
      "learning_rate": 0.0001857663799397879,
      "loss": 1.1362,
      "step": 12770
    },
    {
      "epoch": 0.5178177103381212,
      "grad_norm": 0.5422564744949341,
      "learning_rate": 0.00018574453897437447,
      "loss": 1.0968,
      "step": 12780
    },
    {
      "epoch": 0.5182228885152246,
      "grad_norm": 0.5511117577552795,
      "learning_rate": 0.00018572268255071718,
      "loss": 1.153,
      "step": 12790
    },
    {
      "epoch": 0.518628066692328,
      "grad_norm": 0.5455918908119202,
      "learning_rate": 0.00018570081067275637,
      "loss": 1.1644,
      "step": 12800
    },
    {
      "epoch": 0.5190332448694314,
      "grad_norm": 0.5813199281692505,
      "learning_rate": 0.00018567892334443517,
      "loss": 1.1324,
      "step": 12810
    },
    {
      "epoch": 0.5194384230465348,
      "grad_norm": 0.5217265486717224,
      "learning_rate": 0.00018565702056969944,
      "loss": 1.1354,
      "step": 12820
    },
    {
      "epoch": 0.519843601223638,
      "grad_norm": 0.5278756022453308,
      "learning_rate": 0.00018563510235249794,
      "loss": 1.1294,
      "step": 12830
    },
    {
      "epoch": 0.5202487794007414,
      "grad_norm": 0.58599454164505,
      "learning_rate": 0.0001856131686967821,
      "loss": 1.1416,
      "step": 12840
    },
    {
      "epoch": 0.5206539575778448,
      "grad_norm": 0.6153993606567383,
      "learning_rate": 0.00018559121960650616,
      "loss": 1.1725,
      "step": 12850
    },
    {
      "epoch": 0.5210591357549482,
      "grad_norm": 0.5844804048538208,
      "learning_rate": 0.0001855692550856272,
      "loss": 1.1586,
      "step": 12860
    },
    {
      "epoch": 0.5214643139320516,
      "grad_norm": 0.6225375533103943,
      "learning_rate": 0.00018554727513810503,
      "loss": 1.1902,
      "step": 12870
    },
    {
      "epoch": 0.521869492109155,
      "grad_norm": 0.5559949278831482,
      "learning_rate": 0.0001855252797679022,
      "loss": 1.1394,
      "step": 12880
    },
    {
      "epoch": 0.5222746702862584,
      "grad_norm": 0.5068804621696472,
      "learning_rate": 0.00018550326897898414,
      "loss": 1.1641,
      "step": 12890
    },
    {
      "epoch": 0.5226798484633618,
      "grad_norm": 0.5890830159187317,
      "learning_rate": 0.000185481242775319,
      "loss": 1.1666,
      "step": 12900
    },
    {
      "epoch": 0.5230850266404652,
      "grad_norm": 0.5581338405609131,
      "learning_rate": 0.00018545920116087776,
      "loss": 1.158,
      "step": 12910
    },
    {
      "epoch": 0.5234902048175685,
      "grad_norm": 0.56955885887146,
      "learning_rate": 0.00018543714413963406,
      "loss": 1.1306,
      "step": 12920
    },
    {
      "epoch": 0.5238953829946719,
      "grad_norm": 0.5089685320854187,
      "learning_rate": 0.00018541507171556445,
      "loss": 1.1788,
      "step": 12930
    },
    {
      "epoch": 0.5243005611717753,
      "grad_norm": 0.6204522848129272,
      "learning_rate": 0.0001853929838926482,
      "loss": 1.1322,
      "step": 12940
    },
    {
      "epoch": 0.5247057393488787,
      "grad_norm": 0.6265764832496643,
      "learning_rate": 0.00018537088067486734,
      "loss": 1.1564,
      "step": 12950
    },
    {
      "epoch": 0.525110917525982,
      "grad_norm": 0.5732021331787109,
      "learning_rate": 0.00018534876206620672,
      "loss": 1.132,
      "step": 12960
    },
    {
      "epoch": 0.5255160957030854,
      "grad_norm": 0.5585970878601074,
      "learning_rate": 0.0001853266280706539,
      "loss": 1.1663,
      "step": 12970
    },
    {
      "epoch": 0.5259212738801888,
      "grad_norm": 0.544829249382019,
      "learning_rate": 0.00018530447869219926,
      "loss": 1.1631,
      "step": 12980
    },
    {
      "epoch": 0.5263264520572922,
      "grad_norm": 0.5667039752006531,
      "learning_rate": 0.000185282313934836,
      "loss": 1.1432,
      "step": 12990
    },
    {
      "epoch": 0.5267316302343956,
      "grad_norm": 0.5751187801361084,
      "learning_rate": 0.00018526013380255999,
      "loss": 1.1192,
      "step": 13000
    },
    {
      "epoch": 0.5271368084114989,
      "grad_norm": 0.5418323874473572,
      "learning_rate": 0.0001852379382993699,
      "loss": 1.1606,
      "step": 13010
    },
    {
      "epoch": 0.5275419865886023,
      "grad_norm": 0.5098136067390442,
      "learning_rate": 0.00018521572742926726,
      "loss": 1.1413,
      "step": 13020
    },
    {
      "epoch": 0.5279471647657057,
      "grad_norm": 0.5731351375579834,
      "learning_rate": 0.00018519350119625624,
      "loss": 1.1242,
      "step": 13030
    },
    {
      "epoch": 0.5283523429428091,
      "grad_norm": 0.5401512980461121,
      "learning_rate": 0.00018517125960434386,
      "loss": 1.1628,
      "step": 13040
    },
    {
      "epoch": 0.5287575211199125,
      "grad_norm": 0.5491833686828613,
      "learning_rate": 0.00018514900265753994,
      "loss": 1.1488,
      "step": 13050
    },
    {
      "epoch": 0.5291626992970159,
      "grad_norm": 0.5310999155044556,
      "learning_rate": 0.00018512673035985696,
      "loss": 1.1197,
      "step": 13060
    },
    {
      "epoch": 0.5295678774741193,
      "grad_norm": 0.6302385926246643,
      "learning_rate": 0.00018510444271531022,
      "loss": 1.1336,
      "step": 13070
    },
    {
      "epoch": 0.5299730556512227,
      "grad_norm": 0.580278217792511,
      "learning_rate": 0.00018508213972791784,
      "loss": 1.1406,
      "step": 13080
    },
    {
      "epoch": 0.530378233828326,
      "grad_norm": 0.5810430645942688,
      "learning_rate": 0.00018505982140170062,
      "loss": 1.1507,
      "step": 13090
    },
    {
      "epoch": 0.5307834120054293,
      "grad_norm": 0.5094375014305115,
      "learning_rate": 0.0001850374877406822,
      "loss": 1.1523,
      "step": 13100
    },
    {
      "epoch": 0.5311885901825327,
      "grad_norm": 0.5088129043579102,
      "learning_rate": 0.0001850151387488889,
      "loss": 1.1184,
      "step": 13110
    },
    {
      "epoch": 0.5315937683596361,
      "grad_norm": 0.5770439505577087,
      "learning_rate": 0.0001849927744303499,
      "loss": 1.1474,
      "step": 13120
    },
    {
      "epoch": 0.5319989465367395,
      "grad_norm": 0.6115645170211792,
      "learning_rate": 0.0001849703947890971,
      "loss": 1.1579,
      "step": 13130
    },
    {
      "epoch": 0.5324041247138429,
      "grad_norm": 0.6145834922790527,
      "learning_rate": 0.0001849479998291651,
      "loss": 1.1488,
      "step": 13140
    },
    {
      "epoch": 0.5328093028909463,
      "grad_norm": 0.6391405463218689,
      "learning_rate": 0.00018492558955459138,
      "loss": 1.1493,
      "step": 13150
    },
    {
      "epoch": 0.5332144810680497,
      "grad_norm": 0.5791735053062439,
      "learning_rate": 0.0001849031639694161,
      "loss": 1.1865,
      "step": 13160
    },
    {
      "epoch": 0.5336196592451531,
      "grad_norm": 0.6057605147361755,
      "learning_rate": 0.0001848807230776822,
      "loss": 1.1034,
      "step": 13170
    },
    {
      "epoch": 0.5340248374222565,
      "grad_norm": 0.5941310524940491,
      "learning_rate": 0.00018485826688343537,
      "loss": 1.1475,
      "step": 13180
    },
    {
      "epoch": 0.5344300155993598,
      "grad_norm": 0.5050292611122131,
      "learning_rate": 0.0001848357953907241,
      "loss": 1.1597,
      "step": 13190
    },
    {
      "epoch": 0.5348351937764632,
      "grad_norm": 0.5622362494468689,
      "learning_rate": 0.0001848133086035996,
      "loss": 1.1202,
      "step": 13200
    },
    {
      "epoch": 0.5352403719535666,
      "grad_norm": 0.5024674534797668,
      "learning_rate": 0.00018479080652611583,
      "loss": 1.1175,
      "step": 13210
    },
    {
      "epoch": 0.53564555013067,
      "grad_norm": 0.7323898673057556,
      "learning_rate": 0.0001847682891623295,
      "loss": 1.149,
      "step": 13220
    },
    {
      "epoch": 0.5360507283077733,
      "grad_norm": 0.5820547938346863,
      "learning_rate": 0.00018474575651630012,
      "loss": 1.169,
      "step": 13230
    },
    {
      "epoch": 0.5364559064848767,
      "grad_norm": 0.5201345682144165,
      "learning_rate": 0.00018472320859208997,
      "loss": 1.1621,
      "step": 13240
    },
    {
      "epoch": 0.5368610846619801,
      "grad_norm": 0.5930827260017395,
      "learning_rate": 0.000184700645393764,
      "loss": 1.166,
      "step": 13250
    },
    {
      "epoch": 0.5372662628390835,
      "grad_norm": 0.610812246799469,
      "learning_rate": 0.00018467806692538998,
      "loss": 1.1436,
      "step": 13260
    },
    {
      "epoch": 0.5376714410161869,
      "grad_norm": 0.532146692276001,
      "learning_rate": 0.00018465547319103842,
      "loss": 1.1408,
      "step": 13270
    },
    {
      "epoch": 0.5380766191932902,
      "grad_norm": 0.5802783370018005,
      "learning_rate": 0.00018463286419478255,
      "loss": 1.1705,
      "step": 13280
    },
    {
      "epoch": 0.5384817973703936,
      "grad_norm": 0.5436097979545593,
      "learning_rate": 0.00018461023994069842,
      "loss": 1.1372,
      "step": 13290
    },
    {
      "epoch": 0.538886975547497,
      "grad_norm": 0.5737035274505615,
      "learning_rate": 0.00018458760043286475,
      "loss": 1.1684,
      "step": 13300
    },
    {
      "epoch": 0.5392921537246004,
      "grad_norm": 0.464033305644989,
      "learning_rate": 0.00018456494567536304,
      "loss": 1.1179,
      "step": 13310
    },
    {
      "epoch": 0.5396973319017038,
      "grad_norm": 0.5381934642791748,
      "learning_rate": 0.00018454227567227756,
      "loss": 1.1276,
      "step": 13320
    },
    {
      "epoch": 0.5401025100788072,
      "grad_norm": 0.49630072712898254,
      "learning_rate": 0.00018451959042769533,
      "loss": 1.0847,
      "step": 13330
    },
    {
      "epoch": 0.5405076882559106,
      "grad_norm": 0.5625321865081787,
      "learning_rate": 0.0001844968899457061,
      "loss": 1.1822,
      "step": 13340
    },
    {
      "epoch": 0.540912866433014,
      "grad_norm": 0.5433419942855835,
      "learning_rate": 0.0001844741742304024,
      "loss": 1.1343,
      "step": 13350
    },
    {
      "epoch": 0.5413180446101173,
      "grad_norm": 0.6659138202667236,
      "learning_rate": 0.0001844514432858794,
      "loss": 1.1386,
      "step": 13360
    },
    {
      "epoch": 0.5417232227872206,
      "grad_norm": 0.5874588489532471,
      "learning_rate": 0.00018442869711623517,
      "loss": 1.1622,
      "step": 13370
    },
    {
      "epoch": 0.542128400964324,
      "grad_norm": 0.6449431777000427,
      "learning_rate": 0.00018440593572557039,
      "loss": 1.147,
      "step": 13380
    },
    {
      "epoch": 0.5425335791414274,
      "grad_norm": 0.5579624772071838,
      "learning_rate": 0.00018438315911798857,
      "loss": 1.1343,
      "step": 13390
    },
    {
      "epoch": 0.5429387573185308,
      "grad_norm": 0.539602518081665,
      "learning_rate": 0.00018436036729759593,
      "loss": 1.1121,
      "step": 13400
    },
    {
      "epoch": 0.5433439354956342,
      "grad_norm": 0.5670642256736755,
      "learning_rate": 0.00018433756026850145,
      "loss": 1.1236,
      "step": 13410
    },
    {
      "epoch": 0.5437491136727376,
      "grad_norm": 0.5241008400917053,
      "learning_rate": 0.00018431473803481684,
      "loss": 1.1752,
      "step": 13420
    },
    {
      "epoch": 0.544154291849841,
      "grad_norm": 0.5444966554641724,
      "learning_rate": 0.0001842919006006565,
      "loss": 1.1612,
      "step": 13430
    },
    {
      "epoch": 0.5445594700269444,
      "grad_norm": 0.5885390639305115,
      "learning_rate": 0.00018426904797013766,
      "loss": 1.14,
      "step": 13440
    },
    {
      "epoch": 0.5449646482040478,
      "grad_norm": 0.5061194896697998,
      "learning_rate": 0.00018424618014738027,
      "loss": 1.1034,
      "step": 13450
    },
    {
      "epoch": 0.5453698263811511,
      "grad_norm": 0.6269741058349609,
      "learning_rate": 0.00018422329713650698,
      "loss": 1.1621,
      "step": 13460
    },
    {
      "epoch": 0.5457750045582545,
      "grad_norm": 0.5817946791648865,
      "learning_rate": 0.00018420039894164318,
      "loss": 1.1403,
      "step": 13470
    },
    {
      "epoch": 0.5461801827353578,
      "grad_norm": 0.5719766020774841,
      "learning_rate": 0.00018417748556691705,
      "loss": 1.164,
      "step": 13480
    },
    {
      "epoch": 0.5465853609124612,
      "grad_norm": 0.5554754137992859,
      "learning_rate": 0.00018415455701645942,
      "loss": 1.167,
      "step": 13490
    },
    {
      "epoch": 0.5469905390895646,
      "grad_norm": 0.615409255027771,
      "learning_rate": 0.00018413161329440398,
      "loss": 1.1255,
      "step": 13500
    },
    {
      "epoch": 0.547395717266668,
      "grad_norm": 0.5597519278526306,
      "learning_rate": 0.00018410865440488702,
      "loss": 1.1729,
      "step": 13510
    },
    {
      "epoch": 0.5478008954437714,
      "grad_norm": 0.614645779132843,
      "learning_rate": 0.00018408568035204766,
      "loss": 1.1717,
      "step": 13520
    },
    {
      "epoch": 0.5482060736208748,
      "grad_norm": 0.5625033974647522,
      "learning_rate": 0.0001840626911400277,
      "loss": 1.1369,
      "step": 13530
    },
    {
      "epoch": 0.5486112517979782,
      "grad_norm": 0.6904319524765015,
      "learning_rate": 0.00018403968677297168,
      "loss": 1.1513,
      "step": 13540
    },
    {
      "epoch": 0.5490164299750815,
      "grad_norm": 0.5377181172370911,
      "learning_rate": 0.00018401666725502693,
      "loss": 1.1358,
      "step": 13550
    },
    {
      "epoch": 0.5494216081521849,
      "grad_norm": 0.5708648562431335,
      "learning_rate": 0.00018399363259034347,
      "loss": 1.136,
      "step": 13560
    },
    {
      "epoch": 0.5498267863292883,
      "grad_norm": 0.613953709602356,
      "learning_rate": 0.00018397058278307398,
      "loss": 1.1359,
      "step": 13570
    },
    {
      "epoch": 0.5502319645063917,
      "grad_norm": 0.6092777848243713,
      "learning_rate": 0.000183947517837374,
      "loss": 1.1524,
      "step": 13580
    },
    {
      "epoch": 0.5506371426834951,
      "grad_norm": 0.6050844788551331,
      "learning_rate": 0.00018392443775740175,
      "loss": 1.1427,
      "step": 13590
    },
    {
      "epoch": 0.5510423208605985,
      "grad_norm": 0.48570799827575684,
      "learning_rate": 0.0001839013425473181,
      "loss": 1.1479,
      "step": 13600
    },
    {
      "epoch": 0.5514474990377018,
      "grad_norm": 0.4972561001777649,
      "learning_rate": 0.00018387823221128677,
      "loss": 1.1411,
      "step": 13610
    },
    {
      "epoch": 0.5518526772148052,
      "grad_norm": 0.5182533264160156,
      "learning_rate": 0.00018385510675347414,
      "loss": 1.1448,
      "step": 13620
    },
    {
      "epoch": 0.5522578553919086,
      "grad_norm": 0.4909082353115082,
      "learning_rate": 0.00018383196617804926,
      "loss": 1.1406,
      "step": 13630
    },
    {
      "epoch": 0.5526630335690119,
      "grad_norm": 0.6214056015014648,
      "learning_rate": 0.00018380881048918405,
      "loss": 1.164,
      "step": 13640
    },
    {
      "epoch": 0.5530682117461153,
      "grad_norm": 0.6163676381111145,
      "learning_rate": 0.0001837856396910531,
      "loss": 1.1549,
      "step": 13650
    },
    {
      "epoch": 0.5534733899232187,
      "grad_norm": 0.5856980085372925,
      "learning_rate": 0.0001837624537878336,
      "loss": 1.1774,
      "step": 13660
    },
    {
      "epoch": 0.5538785681003221,
      "grad_norm": 0.5481765866279602,
      "learning_rate": 0.00018373925278370565,
      "loss": 1.1463,
      "step": 13670
    },
    {
      "epoch": 0.5542837462774255,
      "grad_norm": 0.6342101693153381,
      "learning_rate": 0.00018371603668285196,
      "loss": 1.1586,
      "step": 13680
    },
    {
      "epoch": 0.5546889244545289,
      "grad_norm": 0.5508027672767639,
      "learning_rate": 0.00018369280548945796,
      "loss": 1.1611,
      "step": 13690
    },
    {
      "epoch": 0.5550941026316323,
      "grad_norm": 0.5576925873756409,
      "learning_rate": 0.00018366955920771184,
      "loss": 1.1929,
      "step": 13700
    },
    {
      "epoch": 0.5554992808087357,
      "grad_norm": 0.5599436163902283,
      "learning_rate": 0.00018364629784180453,
      "loss": 1.1461,
      "step": 13710
    },
    {
      "epoch": 0.5559044589858391,
      "grad_norm": 0.6529631018638611,
      "learning_rate": 0.00018362302139592962,
      "loss": 1.1406,
      "step": 13720
    },
    {
      "epoch": 0.5563096371629423,
      "grad_norm": 0.5731914043426514,
      "learning_rate": 0.00018359972987428344,
      "loss": 1.1229,
      "step": 13730
    },
    {
      "epoch": 0.5567148153400457,
      "grad_norm": 0.5896695852279663,
      "learning_rate": 0.0001835764232810651,
      "loss": 1.1307,
      "step": 13740
    },
    {
      "epoch": 0.5571199935171491,
      "grad_norm": 0.5126842856407166,
      "learning_rate": 0.00018355310162047628,
      "loss": 1.1545,
      "step": 13750
    },
    {
      "epoch": 0.5575251716942525,
      "grad_norm": 0.5623682141304016,
      "learning_rate": 0.00018352976489672157,
      "loss": 1.1301,
      "step": 13760
    },
    {
      "epoch": 0.5579303498713559,
      "grad_norm": 0.5503577589988708,
      "learning_rate": 0.00018350641311400812,
      "loss": 1.1487,
      "step": 13770
    },
    {
      "epoch": 0.5583355280484593,
      "grad_norm": 0.5301833748817444,
      "learning_rate": 0.00018348304627654582,
      "loss": 1.1688,
      "step": 13780
    },
    {
      "epoch": 0.5587407062255627,
      "grad_norm": 0.5980144143104553,
      "learning_rate": 0.00018345966438854734,
      "loss": 1.1617,
      "step": 13790
    },
    {
      "epoch": 0.5591458844026661,
      "grad_norm": 0.5282531380653381,
      "learning_rate": 0.00018343626745422805,
      "loss": 1.1519,
      "step": 13800
    },
    {
      "epoch": 0.5595510625797695,
      "grad_norm": 0.5064395666122437,
      "learning_rate": 0.00018341285547780593,
      "loss": 1.1387,
      "step": 13810
    },
    {
      "epoch": 0.5599562407568728,
      "grad_norm": 0.6021496057510376,
      "learning_rate": 0.00018338942846350188,
      "loss": 1.1582,
      "step": 13820
    },
    {
      "epoch": 0.5603614189339762,
      "grad_norm": 0.6732884645462036,
      "learning_rate": 0.00018336598641553923,
      "loss": 1.1392,
      "step": 13830
    },
    {
      "epoch": 0.5607665971110796,
      "grad_norm": 0.60038161277771,
      "learning_rate": 0.00018334252933814427,
      "loss": 1.1708,
      "step": 13840
    },
    {
      "epoch": 0.561171775288183,
      "grad_norm": 0.5192916393280029,
      "learning_rate": 0.00018331905723554588,
      "loss": 1.1338,
      "step": 13850
    },
    {
      "epoch": 0.5615769534652864,
      "grad_norm": 0.5365586280822754,
      "learning_rate": 0.00018329557011197568,
      "loss": 1.129,
      "step": 13860
    },
    {
      "epoch": 0.5619821316423897,
      "grad_norm": 0.537321925163269,
      "learning_rate": 0.00018327206797166796,
      "loss": 1.152,
      "step": 13870
    },
    {
      "epoch": 0.5623873098194931,
      "grad_norm": 0.5585379004478455,
      "learning_rate": 0.00018324855081885977,
      "loss": 1.1488,
      "step": 13880
    },
    {
      "epoch": 0.5627924879965965,
      "grad_norm": 0.5678502321243286,
      "learning_rate": 0.00018322501865779083,
      "loss": 1.1741,
      "step": 13890
    },
    {
      "epoch": 0.5631976661736999,
      "grad_norm": 0.5912489891052246,
      "learning_rate": 0.00018320147149270357,
      "loss": 1.1335,
      "step": 13900
    },
    {
      "epoch": 0.5636028443508032,
      "grad_norm": 0.514712393283844,
      "learning_rate": 0.00018317790932784317,
      "loss": 1.1521,
      "step": 13910
    },
    {
      "epoch": 0.5640080225279066,
      "grad_norm": 0.5337997078895569,
      "learning_rate": 0.00018315433216745743,
      "loss": 1.1395,
      "step": 13920
    },
    {
      "epoch": 0.56441320070501,
      "grad_norm": 0.5553361773490906,
      "learning_rate": 0.00018313074001579692,
      "loss": 1.1682,
      "step": 13930
    },
    {
      "epoch": 0.5648183788821134,
      "grad_norm": 0.5231690406799316,
      "learning_rate": 0.00018310713287711492,
      "loss": 1.1683,
      "step": 13940
    },
    {
      "epoch": 0.5652235570592168,
      "grad_norm": 0.4989342987537384,
      "learning_rate": 0.00018308351075566734,
      "loss": 1.1553,
      "step": 13950
    },
    {
      "epoch": 0.5656287352363202,
      "grad_norm": 0.5686271786689758,
      "learning_rate": 0.0001830598736557129,
      "loss": 1.1851,
      "step": 13960
    },
    {
      "epoch": 0.5660339134134236,
      "grad_norm": 0.5540581345558167,
      "learning_rate": 0.00018303622158151285,
      "loss": 1.138,
      "step": 13970
    },
    {
      "epoch": 0.566439091590527,
      "grad_norm": 0.4647706151008606,
      "learning_rate": 0.00018301255453733134,
      "loss": 1.1281,
      "step": 13980
    },
    {
      "epoch": 0.5668442697676304,
      "grad_norm": 0.5263564586639404,
      "learning_rate": 0.00018298887252743507,
      "loss": 1.1407,
      "step": 13990
    },
    {
      "epoch": 0.5672494479447336,
      "grad_norm": 0.5497360825538635,
      "learning_rate": 0.00018296517555609355,
      "loss": 1.1587,
      "step": 14000
    },
    {
      "epoch": 0.567654626121837,
      "grad_norm": 0.5530768632888794,
      "learning_rate": 0.0001829414636275789,
      "loss": 1.1702,
      "step": 14010
    },
    {
      "epoch": 0.5680598042989404,
      "grad_norm": 0.5131743550300598,
      "learning_rate": 0.00018291773674616594,
      "loss": 1.1456,
      "step": 14020
    },
    {
      "epoch": 0.5684649824760438,
      "grad_norm": 0.5370258688926697,
      "learning_rate": 0.00018289399491613225,
      "loss": 1.1191,
      "step": 14030
    },
    {
      "epoch": 0.5688701606531472,
      "grad_norm": 0.5787819027900696,
      "learning_rate": 0.00018287023814175807,
      "loss": 1.1389,
      "step": 14040
    },
    {
      "epoch": 0.5692753388302506,
      "grad_norm": 0.5509978532791138,
      "learning_rate": 0.0001828464664273263,
      "loss": 1.1443,
      "step": 14050
    },
    {
      "epoch": 0.569680517007354,
      "grad_norm": 0.5334867835044861,
      "learning_rate": 0.00018282267977712258,
      "loss": 1.1483,
      "step": 14060
    },
    {
      "epoch": 0.5700856951844574,
      "grad_norm": 0.6355846524238586,
      "learning_rate": 0.00018279887819543523,
      "loss": 1.1614,
      "step": 14070
    },
    {
      "epoch": 0.5704908733615608,
      "grad_norm": 0.538515567779541,
      "learning_rate": 0.00018277506168655526,
      "loss": 1.145,
      "step": 14080
    },
    {
      "epoch": 0.5708960515386641,
      "grad_norm": 0.6348052024841309,
      "learning_rate": 0.00018275123025477639,
      "loss": 1.0969,
      "step": 14090
    },
    {
      "epoch": 0.5713012297157675,
      "grad_norm": 0.5863112211227417,
      "learning_rate": 0.00018272738390439494,
      "loss": 1.1232,
      "step": 14100
    },
    {
      "epoch": 0.5717064078928709,
      "grad_norm": 0.622995138168335,
      "learning_rate": 0.00018270352263971012,
      "loss": 1.1313,
      "step": 14110
    },
    {
      "epoch": 0.5721115860699743,
      "grad_norm": 0.5736756920814514,
      "learning_rate": 0.00018267964646502357,
      "loss": 1.1403,
      "step": 14120
    },
    {
      "epoch": 0.5725167642470776,
      "grad_norm": 0.59429931640625,
      "learning_rate": 0.00018265575538463979,
      "loss": 1.1578,
      "step": 14130
    },
    {
      "epoch": 0.572921942424181,
      "grad_norm": 0.6435374617576599,
      "learning_rate": 0.00018263184940286596,
      "loss": 1.1485,
      "step": 14140
    },
    {
      "epoch": 0.5733271206012844,
      "grad_norm": 0.6171859502792358,
      "learning_rate": 0.00018260792852401188,
      "loss": 1.1336,
      "step": 14150
    },
    {
      "epoch": 0.5737322987783878,
      "grad_norm": 0.5482004284858704,
      "learning_rate": 0.00018258399275239006,
      "loss": 1.1347,
      "step": 14160
    },
    {
      "epoch": 0.5741374769554912,
      "grad_norm": 0.5166334509849548,
      "learning_rate": 0.00018256004209231573,
      "loss": 1.1538,
      "step": 14170
    },
    {
      "epoch": 0.5745426551325945,
      "grad_norm": 0.5549854040145874,
      "learning_rate": 0.0001825360765481068,
      "loss": 1.1672,
      "step": 14180
    },
    {
      "epoch": 0.5749478333096979,
      "grad_norm": 0.6041259169578552,
      "learning_rate": 0.00018251209612408373,
      "loss": 1.1424,
      "step": 14190
    },
    {
      "epoch": 0.5753530114868013,
      "grad_norm": 0.5467025637626648,
      "learning_rate": 0.00018248810082456988,
      "loss": 1.1467,
      "step": 14200
    },
    {
      "epoch": 0.5757581896639047,
      "grad_norm": 0.6007269620895386,
      "learning_rate": 0.00018246409065389113,
      "loss": 1.1214,
      "step": 14210
    },
    {
      "epoch": 0.5761633678410081,
      "grad_norm": 0.5384333729743958,
      "learning_rate": 0.0001824400656163761,
      "loss": 1.1045,
      "step": 14220
    },
    {
      "epoch": 0.5765685460181115,
      "grad_norm": 0.567457914352417,
      "learning_rate": 0.00018241602571635612,
      "loss": 1.1607,
      "step": 14230
    },
    {
      "epoch": 0.5769737241952149,
      "grad_norm": 0.6583143472671509,
      "learning_rate": 0.00018239197095816514,
      "loss": 1.1684,
      "step": 14240
    },
    {
      "epoch": 0.5773789023723183,
      "grad_norm": 0.5851271748542786,
      "learning_rate": 0.00018236790134613977,
      "loss": 1.1306,
      "step": 14250
    },
    {
      "epoch": 0.5777840805494217,
      "grad_norm": 0.528506338596344,
      "learning_rate": 0.00018234381688461942,
      "loss": 1.1869,
      "step": 14260
    },
    {
      "epoch": 0.5781892587265249,
      "grad_norm": 0.5153642892837524,
      "learning_rate": 0.00018231971757794608,
      "loss": 1.1689,
      "step": 14270
    },
    {
      "epoch": 0.5785944369036283,
      "grad_norm": 0.5718845725059509,
      "learning_rate": 0.00018229560343046438,
      "loss": 1.1647,
      "step": 14280
    },
    {
      "epoch": 0.5789996150807317,
      "grad_norm": 0.5913770794868469,
      "learning_rate": 0.00018227147444652172,
      "loss": 1.1217,
      "step": 14290
    },
    {
      "epoch": 0.5794047932578351,
      "grad_norm": 0.584455668926239,
      "learning_rate": 0.00018224733063046815,
      "loss": 1.1832,
      "step": 14300
    },
    {
      "epoch": 0.5798099714349385,
      "grad_norm": 0.5233607292175293,
      "learning_rate": 0.00018222317198665638,
      "loss": 1.1287,
      "step": 14310
    },
    {
      "epoch": 0.5802151496120419,
      "grad_norm": 0.551790177822113,
      "learning_rate": 0.0001821989985194417,
      "loss": 1.1754,
      "step": 14320
    },
    {
      "epoch": 0.5806203277891453,
      "grad_norm": 0.5858867764472961,
      "learning_rate": 0.0001821748102331823,
      "loss": 1.1156,
      "step": 14330
    },
    {
      "epoch": 0.5810255059662487,
      "grad_norm": 0.5082470178604126,
      "learning_rate": 0.0001821506071322388,
      "loss": 1.1676,
      "step": 14340
    },
    {
      "epoch": 0.5814306841433521,
      "grad_norm": 0.5247595906257629,
      "learning_rate": 0.00018212638922097466,
      "loss": 1.117,
      "step": 14350
    },
    {
      "epoch": 0.5818358623204554,
      "grad_norm": 0.5775333642959595,
      "learning_rate": 0.00018210215650375593,
      "loss": 1.1436,
      "step": 14360
    },
    {
      "epoch": 0.5822410404975588,
      "grad_norm": 0.5461657047271729,
      "learning_rate": 0.00018207790898495133,
      "loss": 1.1585,
      "step": 14370
    },
    {
      "epoch": 0.5826462186746622,
      "grad_norm": 0.6340778470039368,
      "learning_rate": 0.00018205364666893227,
      "loss": 1.1223,
      "step": 14380
    },
    {
      "epoch": 0.5830513968517655,
      "grad_norm": 0.5641189217567444,
      "learning_rate": 0.00018202936956007284,
      "loss": 1.1364,
      "step": 14390
    },
    {
      "epoch": 0.5834565750288689,
      "grad_norm": 0.5474604964256287,
      "learning_rate": 0.00018200507766274977,
      "loss": 1.1592,
      "step": 14400
    },
    {
      "epoch": 0.5838617532059723,
      "grad_norm": 0.5618220567703247,
      "learning_rate": 0.00018198077098134242,
      "loss": 1.1776,
      "step": 14410
    },
    {
      "epoch": 0.5842669313830757,
      "grad_norm": 0.5832874178886414,
      "learning_rate": 0.00018195644952023293,
      "loss": 1.1894,
      "step": 14420
    },
    {
      "epoch": 0.5846721095601791,
      "grad_norm": 0.4921233654022217,
      "learning_rate": 0.00018193211328380602,
      "loss": 1.164,
      "step": 14430
    },
    {
      "epoch": 0.5850772877372825,
      "grad_norm": 0.5238885283470154,
      "learning_rate": 0.00018190776227644908,
      "loss": 1.1272,
      "step": 14440
    },
    {
      "epoch": 0.5854824659143858,
      "grad_norm": 0.5255231857299805,
      "learning_rate": 0.00018188339650255215,
      "loss": 1.1297,
      "step": 14450
    },
    {
      "epoch": 0.5858876440914892,
      "grad_norm": 0.5367745757102966,
      "learning_rate": 0.00018185901596650794,
      "loss": 1.1524,
      "step": 14460
    },
    {
      "epoch": 0.5862928222685926,
      "grad_norm": 0.5279435515403748,
      "learning_rate": 0.0001818346206727119,
      "loss": 1.1319,
      "step": 14470
    },
    {
      "epoch": 0.586698000445696,
      "grad_norm": 0.5370720028877258,
      "learning_rate": 0.00018181021062556204,
      "loss": 1.145,
      "step": 14480
    },
    {
      "epoch": 0.5871031786227994,
      "grad_norm": 0.6028938293457031,
      "learning_rate": 0.00018178578582945908,
      "loss": 1.1159,
      "step": 14490
    },
    {
      "epoch": 0.5875083567999028,
      "grad_norm": 0.5964636206626892,
      "learning_rate": 0.00018176134628880637,
      "loss": 1.1158,
      "step": 14500
    },
    {
      "epoch": 0.5879135349770062,
      "grad_norm": 0.5652032494544983,
      "learning_rate": 0.0001817368920080099,
      "loss": 1.1428,
      "step": 14510
    },
    {
      "epoch": 0.5883187131541096,
      "grad_norm": 0.554445207118988,
      "learning_rate": 0.0001817124229914784,
      "loss": 1.1641,
      "step": 14520
    },
    {
      "epoch": 0.588723891331213,
      "grad_norm": 0.6095631122589111,
      "learning_rate": 0.0001816879392436232,
      "loss": 1.1294,
      "step": 14530
    },
    {
      "epoch": 0.5891290695083162,
      "grad_norm": 0.5790419578552246,
      "learning_rate": 0.00018166344076885827,
      "loss": 1.1317,
      "step": 14540
    },
    {
      "epoch": 0.5895342476854196,
      "grad_norm": 0.5796729326248169,
      "learning_rate": 0.0001816389275716003,
      "loss": 1.1767,
      "step": 14550
    },
    {
      "epoch": 0.589939425862523,
      "grad_norm": 0.6007521748542786,
      "learning_rate": 0.00018161439965626853,
      "loss": 1.1661,
      "step": 14560
    },
    {
      "epoch": 0.5903446040396264,
      "grad_norm": 0.5367429852485657,
      "learning_rate": 0.00018158985702728498,
      "loss": 1.1439,
      "step": 14570
    },
    {
      "epoch": 0.5907497822167298,
      "grad_norm": 0.5438438653945923,
      "learning_rate": 0.0001815652996890742,
      "loss": 1.1842,
      "step": 14580
    },
    {
      "epoch": 0.5911549603938332,
      "grad_norm": 0.5342624187469482,
      "learning_rate": 0.00018154072764606344,
      "loss": 1.1531,
      "step": 14590
    },
    {
      "epoch": 0.5915601385709366,
      "grad_norm": 0.5393738746643066,
      "learning_rate": 0.0001815161409026827,
      "loss": 1.1591,
      "step": 14600
    },
    {
      "epoch": 0.59196531674804,
      "grad_norm": 0.5078129768371582,
      "learning_rate": 0.00018149153946336446,
      "loss": 1.1211,
      "step": 14610
    },
    {
      "epoch": 0.5923704949251434,
      "grad_norm": 0.5616825819015503,
      "learning_rate": 0.00018146692333254397,
      "loss": 1.1572,
      "step": 14620
    },
    {
      "epoch": 0.5927756731022467,
      "grad_norm": 0.5537816286087036,
      "learning_rate": 0.00018144229251465906,
      "loss": 1.1588,
      "step": 14630
    },
    {
      "epoch": 0.59318085127935,
      "grad_norm": 0.6924319863319397,
      "learning_rate": 0.00018141764701415025,
      "loss": 1.1487,
      "step": 14640
    },
    {
      "epoch": 0.5935860294564534,
      "grad_norm": 0.5854458808898926,
      "learning_rate": 0.00018139298683546073,
      "loss": 1.1544,
      "step": 14650
    },
    {
      "epoch": 0.5939912076335568,
      "grad_norm": 0.5470637679100037,
      "learning_rate": 0.00018136831198303622,
      "loss": 1.1386,
      "step": 14660
    },
    {
      "epoch": 0.5943963858106602,
      "grad_norm": 0.5788532495498657,
      "learning_rate": 0.00018134362246132523,
      "loss": 1.1239,
      "step": 14670
    },
    {
      "epoch": 0.5948015639877636,
      "grad_norm": 0.5278974175453186,
      "learning_rate": 0.00018131891827477884,
      "loss": 1.1644,
      "step": 14680
    },
    {
      "epoch": 0.595206742164867,
      "grad_norm": 0.5592325329780579,
      "learning_rate": 0.00018129419942785076,
      "loss": 1.1606,
      "step": 14690
    },
    {
      "epoch": 0.5956119203419704,
      "grad_norm": 0.6049097776412964,
      "learning_rate": 0.00018126946592499738,
      "loss": 1.1458,
      "step": 14700
    },
    {
      "epoch": 0.5960170985190738,
      "grad_norm": 0.5599257349967957,
      "learning_rate": 0.00018124471777067772,
      "loss": 1.1155,
      "step": 14710
    },
    {
      "epoch": 0.5964222766961771,
      "grad_norm": 0.5290777683258057,
      "learning_rate": 0.00018121995496935342,
      "loss": 1.1031,
      "step": 14720
    },
    {
      "epoch": 0.5968274548732805,
      "grad_norm": 0.6298608779907227,
      "learning_rate": 0.00018119517752548882,
      "loss": 1.142,
      "step": 14730
    },
    {
      "epoch": 0.5972326330503839,
      "grad_norm": 0.527826189994812,
      "learning_rate": 0.00018117038544355083,
      "loss": 1.1613,
      "step": 14740
    },
    {
      "epoch": 0.5976378112274873,
      "grad_norm": 0.6178900599479675,
      "learning_rate": 0.00018114557872800905,
      "loss": 1.1375,
      "step": 14750
    },
    {
      "epoch": 0.5980429894045907,
      "grad_norm": 0.5640949606895447,
      "learning_rate": 0.00018112075738333568,
      "loss": 1.1586,
      "step": 14760
    },
    {
      "epoch": 0.5984481675816941,
      "grad_norm": 0.651132345199585,
      "learning_rate": 0.0001810959214140056,
      "loss": 1.134,
      "step": 14770
    },
    {
      "epoch": 0.5988533457587975,
      "grad_norm": 0.5843057632446289,
      "learning_rate": 0.00018107107082449627,
      "loss": 1.1405,
      "step": 14780
    },
    {
      "epoch": 0.5992585239359008,
      "grad_norm": 0.5334299802780151,
      "learning_rate": 0.00018104620561928782,
      "loss": 1.1429,
      "step": 14790
    },
    {
      "epoch": 0.5996637021130042,
      "grad_norm": 0.5296167135238647,
      "learning_rate": 0.00018102132580286305,
      "loss": 1.1563,
      "step": 14800
    },
    {
      "epoch": 0.6000688802901075,
      "grad_norm": 0.5192431211471558,
      "learning_rate": 0.00018099643137970732,
      "loss": 1.141,
      "step": 14810
    },
    {
      "epoch": 0.6004740584672109,
      "grad_norm": 0.5415987372398376,
      "learning_rate": 0.0001809715223543087,
      "loss": 1.1196,
      "step": 14820
    },
    {
      "epoch": 0.6008792366443143,
      "grad_norm": 0.574212908744812,
      "learning_rate": 0.00018094659873115778,
      "loss": 1.1308,
      "step": 14830
    },
    {
      "epoch": 0.6012844148214177,
      "grad_norm": 0.6291349530220032,
      "learning_rate": 0.00018092166051474792,
      "loss": 1.156,
      "step": 14840
    },
    {
      "epoch": 0.6016895929985211,
      "grad_norm": 0.6112820506095886,
      "learning_rate": 0.00018089670770957502,
      "loss": 1.167,
      "step": 14850
    },
    {
      "epoch": 0.6020947711756245,
      "grad_norm": 0.5630380511283875,
      "learning_rate": 0.00018087174032013764,
      "loss": 1.1664,
      "step": 14860
    },
    {
      "epoch": 0.6024999493527279,
      "grad_norm": 0.5931798815727234,
      "learning_rate": 0.00018084675835093697,
      "loss": 1.1028,
      "step": 14870
    },
    {
      "epoch": 0.6029051275298313,
      "grad_norm": 0.5221173167228699,
      "learning_rate": 0.0001808217618064768,
      "loss": 1.1428,
      "step": 14880
    },
    {
      "epoch": 0.6033103057069347,
      "grad_norm": 0.5705620646476746,
      "learning_rate": 0.0001807967506912636,
      "loss": 1.1086,
      "step": 14890
    },
    {
      "epoch": 0.603715483884038,
      "grad_norm": 0.542236328125,
      "learning_rate": 0.00018077172500980643,
      "loss": 1.145,
      "step": 14900
    },
    {
      "epoch": 0.6041206620611413,
      "grad_norm": 0.5989721417427063,
      "learning_rate": 0.00018074668476661697,
      "loss": 1.1577,
      "step": 14910
    },
    {
      "epoch": 0.6045258402382447,
      "grad_norm": 0.5641351342201233,
      "learning_rate": 0.00018072162996620955,
      "loss": 1.1352,
      "step": 14920
    },
    {
      "epoch": 0.6049310184153481,
      "grad_norm": 0.5481580495834351,
      "learning_rate": 0.00018069656061310113,
      "loss": 1.1369,
      "step": 14930
    },
    {
      "epoch": 0.6053361965924515,
      "grad_norm": 0.6762393712997437,
      "learning_rate": 0.00018067147671181123,
      "loss": 1.1267,
      "step": 14940
    },
    {
      "epoch": 0.6057413747695549,
      "grad_norm": 0.565818190574646,
      "learning_rate": 0.00018064637826686205,
      "loss": 1.1413,
      "step": 14950
    },
    {
      "epoch": 0.6061465529466583,
      "grad_norm": 0.5806300640106201,
      "learning_rate": 0.00018062126528277844,
      "loss": 1.167,
      "step": 14960
    },
    {
      "epoch": 0.6065517311237617,
      "grad_norm": 0.5843779444694519,
      "learning_rate": 0.0001805961377640878,
      "loss": 1.1453,
      "step": 14970
    },
    {
      "epoch": 0.6069569093008651,
      "grad_norm": 0.5848987102508545,
      "learning_rate": 0.00018057099571532024,
      "loss": 1.1298,
      "step": 14980
    },
    {
      "epoch": 0.6073620874779684,
      "grad_norm": 0.5248256921768188,
      "learning_rate": 0.00018054583914100833,
      "loss": 1.1547,
      "step": 14990
    },
    {
      "epoch": 0.6077672656550718,
      "grad_norm": 0.6042876243591309,
      "learning_rate": 0.00018052066804568747,
      "loss": 1.1484,
      "step": 15000
    },
    {
      "epoch": 0.6081724438321752,
      "grad_norm": 0.5553650259971619,
      "learning_rate": 0.00018049548243389546,
      "loss": 1.1111,
      "step": 15010
    },
    {
      "epoch": 0.6085776220092786,
      "grad_norm": 0.5859989523887634,
      "learning_rate": 0.00018047028231017294,
      "loss": 1.1862,
      "step": 15020
    },
    {
      "epoch": 0.608982800186382,
      "grad_norm": 0.5637743473052979,
      "learning_rate": 0.00018044506767906295,
      "loss": 1.1753,
      "step": 15030
    },
    {
      "epoch": 0.6093879783634853,
      "grad_norm": 0.5963983535766602,
      "learning_rate": 0.00018041983854511133,
      "loss": 1.1257,
      "step": 15040
    },
    {
      "epoch": 0.6097931565405887,
      "grad_norm": 0.6794252395629883,
      "learning_rate": 0.0001803945949128664,
      "loss": 1.1413,
      "step": 15050
    },
    {
      "epoch": 0.6101983347176921,
      "grad_norm": 0.5398178100585938,
      "learning_rate": 0.00018036933678687914,
      "loss": 1.1087,
      "step": 15060
    },
    {
      "epoch": 0.6106035128947955,
      "grad_norm": 0.5668941140174866,
      "learning_rate": 0.0001803440641717032,
      "loss": 1.134,
      "step": 15070
    },
    {
      "epoch": 0.6110086910718988,
      "grad_norm": 0.618416428565979,
      "learning_rate": 0.00018031877707189476,
      "loss": 1.1754,
      "step": 15080
    },
    {
      "epoch": 0.6114138692490022,
      "grad_norm": 0.4907599687576294,
      "learning_rate": 0.00018029347549201264,
      "loss": 1.0678,
      "step": 15090
    },
    {
      "epoch": 0.6118190474261056,
      "grad_norm": 0.6468308568000793,
      "learning_rate": 0.0001802681594366183,
      "loss": 1.1069,
      "step": 15100
    },
    {
      "epoch": 0.612224225603209,
      "grad_norm": 0.5647540092468262,
      "learning_rate": 0.00018024282891027573,
      "loss": 1.1099,
      "step": 15110
    },
    {
      "epoch": 0.6126294037803124,
      "grad_norm": 0.5360813736915588,
      "learning_rate": 0.00018021748391755163,
      "loss": 1.1484,
      "step": 15120
    },
    {
      "epoch": 0.6130345819574158,
      "grad_norm": 0.6311752200126648,
      "learning_rate": 0.00018019212446301522,
      "loss": 1.173,
      "step": 15130
    },
    {
      "epoch": 0.6134397601345192,
      "grad_norm": 0.6115313172340393,
      "learning_rate": 0.00018016675055123842,
      "loss": 1.141,
      "step": 15140
    },
    {
      "epoch": 0.6138449383116226,
      "grad_norm": 0.5799046158790588,
      "learning_rate": 0.00018014136218679567,
      "loss": 1.11,
      "step": 15150
    },
    {
      "epoch": 0.614250116488726,
      "grad_norm": 0.6033231616020203,
      "learning_rate": 0.00018011595937426407,
      "loss": 1.1602,
      "step": 15160
    },
    {
      "epoch": 0.6146552946658292,
      "grad_norm": 0.5540711283683777,
      "learning_rate": 0.00018009054211822324,
      "loss": 1.1135,
      "step": 15170
    },
    {
      "epoch": 0.6150604728429326,
      "grad_norm": 0.565997838973999,
      "learning_rate": 0.00018006511042325556,
      "loss": 1.1687,
      "step": 15180
    },
    {
      "epoch": 0.615465651020036,
      "grad_norm": 0.5036917328834534,
      "learning_rate": 0.00018003966429394586,
      "loss": 1.1958,
      "step": 15190
    },
    {
      "epoch": 0.6158708291971394,
      "grad_norm": 0.6302006840705872,
      "learning_rate": 0.00018001420373488165,
      "loss": 1.1742,
      "step": 15200
    },
    {
      "epoch": 0.6162760073742428,
      "grad_norm": 0.6456652879714966,
      "learning_rate": 0.00017998872875065302,
      "loss": 1.1241,
      "step": 15210
    },
    {
      "epoch": 0.6166811855513462,
      "grad_norm": 0.5494721531867981,
      "learning_rate": 0.00017996323934585265,
      "loss": 1.1236,
      "step": 15220
    },
    {
      "epoch": 0.6170863637284496,
      "grad_norm": 0.5605599284172058,
      "learning_rate": 0.0001799377355250759,
      "loss": 1.1685,
      "step": 15230
    },
    {
      "epoch": 0.617491541905553,
      "grad_norm": 0.6177376508712769,
      "learning_rate": 0.0001799122172929206,
      "loss": 1.1438,
      "step": 15240
    },
    {
      "epoch": 0.6178967200826564,
      "grad_norm": 0.5056686997413635,
      "learning_rate": 0.00017988668465398724,
      "loss": 1.1515,
      "step": 15250
    },
    {
      "epoch": 0.6183018982597597,
      "grad_norm": 0.5446740388870239,
      "learning_rate": 0.00017986113761287893,
      "loss": 1.1245,
      "step": 15260
    },
    {
      "epoch": 0.6187070764368631,
      "grad_norm": 0.48948976397514343,
      "learning_rate": 0.00017983557617420136,
      "loss": 1.1378,
      "step": 15270
    },
    {
      "epoch": 0.6191122546139665,
      "grad_norm": 0.5590505003929138,
      "learning_rate": 0.00017981000034256278,
      "loss": 1.1504,
      "step": 15280
    },
    {
      "epoch": 0.6195174327910699,
      "grad_norm": 0.5897467732429504,
      "learning_rate": 0.0001797844101225741,
      "loss": 1.1297,
      "step": 15290
    },
    {
      "epoch": 0.6199226109681732,
      "grad_norm": 0.49428442120552063,
      "learning_rate": 0.00017975880551884883,
      "loss": 1.1322,
      "step": 15300
    },
    {
      "epoch": 0.6203277891452766,
      "grad_norm": 0.5378120541572571,
      "learning_rate": 0.00017973318653600293,
      "loss": 1.1352,
      "step": 15310
    },
    {
      "epoch": 0.62073296732238,
      "grad_norm": 0.5658608078956604,
      "learning_rate": 0.00017970755317865512,
      "loss": 1.1679,
      "step": 15320
    },
    {
      "epoch": 0.6211381454994834,
      "grad_norm": 0.5416398644447327,
      "learning_rate": 0.00017968190545142665,
      "loss": 1.1375,
      "step": 15330
    },
    {
      "epoch": 0.6215433236765868,
      "grad_norm": 0.5597314834594727,
      "learning_rate": 0.00017965624335894136,
      "loss": 1.1756,
      "step": 15340
    },
    {
      "epoch": 0.6219485018536901,
      "grad_norm": 0.6340488791465759,
      "learning_rate": 0.00017963056690582563,
      "loss": 1.1395,
      "step": 15350
    },
    {
      "epoch": 0.6223536800307935,
      "grad_norm": 0.5552951693534851,
      "learning_rate": 0.00017960487609670848,
      "loss": 1.1627,
      "step": 15360
    },
    {
      "epoch": 0.6227588582078969,
      "grad_norm": 0.5535734295845032,
      "learning_rate": 0.0001795791709362216,
      "loss": 1.1385,
      "step": 15370
    },
    {
      "epoch": 0.6231640363850003,
      "grad_norm": 0.5048032999038696,
      "learning_rate": 0.0001795534514289991,
      "loss": 1.0982,
      "step": 15380
    },
    {
      "epoch": 0.6235692145621037,
      "grad_norm": 0.5777239799499512,
      "learning_rate": 0.00017952771757967777,
      "loss": 1.1467,
      "step": 15390
    },
    {
      "epoch": 0.6239743927392071,
      "grad_norm": 0.5462971925735474,
      "learning_rate": 0.000179501969392897,
      "loss": 1.1526,
      "step": 15400
    },
    {
      "epoch": 0.6243795709163105,
      "grad_norm": 0.6101622581481934,
      "learning_rate": 0.00017947620687329874,
      "loss": 1.1636,
      "step": 15410
    },
    {
      "epoch": 0.6247847490934139,
      "grad_norm": 0.5874033570289612,
      "learning_rate": 0.00017945043002552747,
      "loss": 1.1315,
      "step": 15420
    },
    {
      "epoch": 0.6251899272705173,
      "grad_norm": 0.6931918263435364,
      "learning_rate": 0.00017942463885423036,
      "loss": 1.1208,
      "step": 15430
    },
    {
      "epoch": 0.6255951054476205,
      "grad_norm": 0.565241813659668,
      "learning_rate": 0.00017939883336405712,
      "loss": 1.1379,
      "step": 15440
    },
    {
      "epoch": 0.6260002836247239,
      "grad_norm": 0.6057207584381104,
      "learning_rate": 0.00017937301355965996,
      "loss": 1.1026,
      "step": 15450
    },
    {
      "epoch": 0.6264054618018273,
      "grad_norm": 0.5951974987983704,
      "learning_rate": 0.0001793471794456938,
      "loss": 1.1145,
      "step": 15460
    },
    {
      "epoch": 0.6268106399789307,
      "grad_norm": 0.5702649354934692,
      "learning_rate": 0.00017932133102681607,
      "loss": 1.1386,
      "step": 15470
    },
    {
      "epoch": 0.6272158181560341,
      "grad_norm": 0.5532297492027283,
      "learning_rate": 0.00017929546830768676,
      "loss": 1.1169,
      "step": 15480
    },
    {
      "epoch": 0.6276209963331375,
      "grad_norm": 0.573919415473938,
      "learning_rate": 0.0001792695912929685,
      "loss": 1.1465,
      "step": 15490
    },
    {
      "epoch": 0.6280261745102409,
      "grad_norm": 0.5423048734664917,
      "learning_rate": 0.00017924369998732642,
      "loss": 1.1347,
      "step": 15500
    },
    {
      "epoch": 0.6284313526873443,
      "grad_norm": 0.5698812007904053,
      "learning_rate": 0.00017921779439542833,
      "loss": 1.1544,
      "step": 15510
    },
    {
      "epoch": 0.6288365308644477,
      "grad_norm": 0.5335921049118042,
      "learning_rate": 0.00017919187452194454,
      "loss": 1.1909,
      "step": 15520
    },
    {
      "epoch": 0.629241709041551,
      "grad_norm": 0.6562032699584961,
      "learning_rate": 0.0001791659403715479,
      "loss": 1.1415,
      "step": 15530
    },
    {
      "epoch": 0.6296468872186544,
      "grad_norm": 0.5788055062294006,
      "learning_rate": 0.00017913999194891394,
      "loss": 1.1825,
      "step": 15540
    },
    {
      "epoch": 0.6300520653957578,
      "grad_norm": 0.576626181602478,
      "learning_rate": 0.00017911402925872068,
      "loss": 1.1701,
      "step": 15550
    },
    {
      "epoch": 0.6304572435728611,
      "grad_norm": 0.6145191788673401,
      "learning_rate": 0.0001790880523056488,
      "loss": 1.1473,
      "step": 15560
    },
    {
      "epoch": 0.6308624217499645,
      "grad_norm": 0.6791607141494751,
      "learning_rate": 0.00017906206109438139,
      "loss": 1.1693,
      "step": 15570
    },
    {
      "epoch": 0.6312675999270679,
      "grad_norm": 0.5039661526679993,
      "learning_rate": 0.00017903605562960426,
      "loss": 1.1425,
      "step": 15580
    },
    {
      "epoch": 0.6316727781041713,
      "grad_norm": 0.5591131448745728,
      "learning_rate": 0.00017901003591600575,
      "loss": 1.1495,
      "step": 15590
    },
    {
      "epoch": 0.6320779562812747,
      "grad_norm": 0.5623401403427124,
      "learning_rate": 0.00017898400195827678,
      "loss": 1.1472,
      "step": 15600
    },
    {
      "epoch": 0.6324831344583781,
      "grad_norm": 0.531723141670227,
      "learning_rate": 0.00017895795376111077,
      "loss": 1.1445,
      "step": 15610
    },
    {
      "epoch": 0.6328883126354814,
      "grad_norm": 0.5673952698707581,
      "learning_rate": 0.0001789318913292038,
      "loss": 1.1264,
      "step": 15620
    },
    {
      "epoch": 0.6332934908125848,
      "grad_norm": 0.5644148588180542,
      "learning_rate": 0.00017890581466725443,
      "loss": 1.1351,
      "step": 15630
    },
    {
      "epoch": 0.6336986689896882,
      "grad_norm": 0.5325310230255127,
      "learning_rate": 0.00017887972377996384,
      "loss": 1.1872,
      "step": 15640
    },
    {
      "epoch": 0.6341038471667916,
      "grad_norm": 0.6867798566818237,
      "learning_rate": 0.00017885361867203578,
      "loss": 1.1052,
      "step": 15650
    },
    {
      "epoch": 0.634509025343895,
      "grad_norm": 0.5930383205413818,
      "learning_rate": 0.00017882749934817652,
      "loss": 1.1777,
      "step": 15660
    },
    {
      "epoch": 0.6349142035209984,
      "grad_norm": 0.684399425983429,
      "learning_rate": 0.00017880136581309496,
      "loss": 1.1691,
      "step": 15670
    },
    {
      "epoch": 0.6353193816981018,
      "grad_norm": 0.6148549318313599,
      "learning_rate": 0.00017877521807150247,
      "loss": 1.162,
      "step": 15680
    },
    {
      "epoch": 0.6357245598752052,
      "grad_norm": 0.5311169028282166,
      "learning_rate": 0.00017874905612811305,
      "loss": 1.1864,
      "step": 15690
    },
    {
      "epoch": 0.6361297380523085,
      "grad_norm": 0.5955155491828918,
      "learning_rate": 0.00017872287998764325,
      "loss": 1.1284,
      "step": 15700
    },
    {
      "epoch": 0.6365349162294118,
      "grad_norm": 0.5731568932533264,
      "learning_rate": 0.00017869668965481218,
      "loss": 1.1195,
      "step": 15710
    },
    {
      "epoch": 0.6369400944065152,
      "grad_norm": 0.5179499387741089,
      "learning_rate": 0.0001786704851343415,
      "loss": 1.1181,
      "step": 15720
    },
    {
      "epoch": 0.6373452725836186,
      "grad_norm": 0.5488529205322266,
      "learning_rate": 0.0001786442664309554,
      "loss": 1.0923,
      "step": 15730
    },
    {
      "epoch": 0.637750450760722,
      "grad_norm": 0.5430967807769775,
      "learning_rate": 0.00017861803354938066,
      "loss": 1.1297,
      "step": 15740
    },
    {
      "epoch": 0.6381556289378254,
      "grad_norm": 0.5350969433784485,
      "learning_rate": 0.00017859178649434662,
      "loss": 1.1502,
      "step": 15750
    },
    {
      "epoch": 0.6385608071149288,
      "grad_norm": 0.5680364370346069,
      "learning_rate": 0.00017856552527058516,
      "loss": 1.1076,
      "step": 15760
    },
    {
      "epoch": 0.6389659852920322,
      "grad_norm": 0.5959980487823486,
      "learning_rate": 0.00017853924988283073,
      "loss": 1.1146,
      "step": 15770
    },
    {
      "epoch": 0.6393711634691356,
      "grad_norm": 0.5964190363883972,
      "learning_rate": 0.00017851296033582033,
      "loss": 1.0999,
      "step": 15780
    },
    {
      "epoch": 0.639776341646239,
      "grad_norm": 0.6035036444664001,
      "learning_rate": 0.0001784866566342935,
      "loss": 1.1403,
      "step": 15790
    },
    {
      "epoch": 0.6401815198233423,
      "grad_norm": 0.5221298933029175,
      "learning_rate": 0.0001784603387829923,
      "loss": 1.1444,
      "step": 15800
    },
    {
      "epoch": 0.6405866980004457,
      "grad_norm": 0.5712006092071533,
      "learning_rate": 0.00017843400678666145,
      "loss": 1.1591,
      "step": 15810
    },
    {
      "epoch": 0.640991876177549,
      "grad_norm": 0.5557779669761658,
      "learning_rate": 0.0001784076606500481,
      "loss": 1.1548,
      "step": 15820
    },
    {
      "epoch": 0.6413970543546524,
      "grad_norm": 0.6515002250671387,
      "learning_rate": 0.00017838130037790202,
      "loss": 1.1572,
      "step": 15830
    },
    {
      "epoch": 0.6418022325317558,
      "grad_norm": 0.5972471237182617,
      "learning_rate": 0.0001783549259749755,
      "loss": 1.1097,
      "step": 15840
    },
    {
      "epoch": 0.6422074107088592,
      "grad_norm": 0.5221810340881348,
      "learning_rate": 0.00017832853744602335,
      "loss": 1.1112,
      "step": 15850
    },
    {
      "epoch": 0.6426125888859626,
      "grad_norm": 0.5441983342170715,
      "learning_rate": 0.00017830213479580302,
      "loss": 1.1642,
      "step": 15860
    },
    {
      "epoch": 0.643017767063066,
      "grad_norm": 0.5400417447090149,
      "learning_rate": 0.00017827571802907444,
      "loss": 1.1401,
      "step": 15870
    },
    {
      "epoch": 0.6434229452401694,
      "grad_norm": 0.5383133292198181,
      "learning_rate": 0.00017824928715060002,
      "loss": 1.0795,
      "step": 15880
    },
    {
      "epoch": 0.6438281234172727,
      "grad_norm": 0.4918181002140045,
      "learning_rate": 0.00017822284216514487,
      "loss": 1.1108,
      "step": 15890
    },
    {
      "epoch": 0.6442333015943761,
      "grad_norm": 0.5371662974357605,
      "learning_rate": 0.00017819638307747655,
      "loss": 1.171,
      "step": 15900
    },
    {
      "epoch": 0.6446384797714795,
      "grad_norm": 0.5239565372467041,
      "learning_rate": 0.00017816990989236511,
      "loss": 1.1661,
      "step": 15910
    },
    {
      "epoch": 0.6450436579485829,
      "grad_norm": 0.6125378608703613,
      "learning_rate": 0.00017814342261458327,
      "loss": 1.1443,
      "step": 15920
    },
    {
      "epoch": 0.6454488361256863,
      "grad_norm": 0.7451189160346985,
      "learning_rate": 0.0001781169212489062,
      "loss": 1.1376,
      "step": 15930
    },
    {
      "epoch": 0.6458540143027897,
      "grad_norm": 0.5332019925117493,
      "learning_rate": 0.00017809040580011164,
      "loss": 1.1646,
      "step": 15940
    },
    {
      "epoch": 0.646259192479893,
      "grad_norm": 0.5480294823646545,
      "learning_rate": 0.00017806387627297983,
      "loss": 1.1243,
      "step": 15950
    },
    {
      "epoch": 0.6466643706569964,
      "grad_norm": 0.5758212804794312,
      "learning_rate": 0.0001780373326722936,
      "loss": 1.1366,
      "step": 15960
    },
    {
      "epoch": 0.6470695488340998,
      "grad_norm": 0.5421702861785889,
      "learning_rate": 0.0001780107750028383,
      "loss": 1.1512,
      "step": 15970
    },
    {
      "epoch": 0.6474747270112031,
      "grad_norm": 0.5440150499343872,
      "learning_rate": 0.00017798420326940186,
      "loss": 1.1428,
      "step": 15980
    },
    {
      "epoch": 0.6478799051883065,
      "grad_norm": 0.5778823494911194,
      "learning_rate": 0.00017795761747677462,
      "loss": 1.1276,
      "step": 15990
    },
    {
      "epoch": 0.6482850833654099,
      "grad_norm": 0.5322615504264832,
      "learning_rate": 0.0001779310176297496,
      "loss": 1.1311,
      "step": 16000
    },
    {
      "epoch": 0.6486902615425133,
      "grad_norm": 0.6656062006950378,
      "learning_rate": 0.00017790440373312223,
      "loss": 1.133,
      "step": 16010
    },
    {
      "epoch": 0.6490954397196167,
      "grad_norm": 0.5421608686447144,
      "learning_rate": 0.00017787777579169055,
      "loss": 1.1636,
      "step": 16020
    },
    {
      "epoch": 0.6495006178967201,
      "grad_norm": 0.6413946151733398,
      "learning_rate": 0.00017785113381025514,
      "loss": 1.1153,
      "step": 16030
    },
    {
      "epoch": 0.6499057960738235,
      "grad_norm": 0.52801513671875,
      "learning_rate": 0.00017782447779361908,
      "loss": 1.1214,
      "step": 16040
    },
    {
      "epoch": 0.6503109742509269,
      "grad_norm": 0.5191813707351685,
      "learning_rate": 0.00017779780774658795,
      "loss": 1.138,
      "step": 16050
    },
    {
      "epoch": 0.6507161524280303,
      "grad_norm": 0.538918137550354,
      "learning_rate": 0.00017777112367396995,
      "loss": 1.1445,
      "step": 16060
    },
    {
      "epoch": 0.6511213306051336,
      "grad_norm": 0.619615375995636,
      "learning_rate": 0.0001777444255805757,
      "loss": 1.1683,
      "step": 16070
    },
    {
      "epoch": 0.651526508782237,
      "grad_norm": 0.5632796287536621,
      "learning_rate": 0.00017771771347121842,
      "loss": 1.1679,
      "step": 16080
    },
    {
      "epoch": 0.6519316869593403,
      "grad_norm": 0.5901058316230774,
      "learning_rate": 0.00017769098735071382,
      "loss": 1.1424,
      "step": 16090
    },
    {
      "epoch": 0.6523368651364437,
      "grad_norm": 0.6114177703857422,
      "learning_rate": 0.0001776642472238802,
      "loss": 1.1148,
      "step": 16100
    },
    {
      "epoch": 0.6527420433135471,
      "grad_norm": 0.5425405502319336,
      "learning_rate": 0.00017763749309553827,
      "loss": 1.1368,
      "step": 16110
    },
    {
      "epoch": 0.6531472214906505,
      "grad_norm": 0.5590180158615112,
      "learning_rate": 0.00017761072497051136,
      "loss": 1.1223,
      "step": 16120
    },
    {
      "epoch": 0.6535523996677539,
      "grad_norm": 0.57027667760849,
      "learning_rate": 0.00017758394285362537,
      "loss": 1.1432,
      "step": 16130
    },
    {
      "epoch": 0.6539575778448573,
      "grad_norm": 0.5484774112701416,
      "learning_rate": 0.00017755714674970853,
      "loss": 1.1389,
      "step": 16140
    },
    {
      "epoch": 0.6543627560219607,
      "grad_norm": 0.6346367001533508,
      "learning_rate": 0.00017753033666359177,
      "loss": 1.1266,
      "step": 16150
    },
    {
      "epoch": 0.654767934199064,
      "grad_norm": 0.5773330926895142,
      "learning_rate": 0.00017750351260010852,
      "loss": 1.1482,
      "step": 16160
    },
    {
      "epoch": 0.6551731123761674,
      "grad_norm": 0.550987958908081,
      "learning_rate": 0.0001774766745640946,
      "loss": 1.1673,
      "step": 16170
    },
    {
      "epoch": 0.6555782905532708,
      "grad_norm": 0.5816602110862732,
      "learning_rate": 0.0001774498225603885,
      "loss": 1.1307,
      "step": 16180
    },
    {
      "epoch": 0.6559834687303742,
      "grad_norm": 0.5995966196060181,
      "learning_rate": 0.00017742295659383113,
      "loss": 1.1336,
      "step": 16190
    },
    {
      "epoch": 0.6563886469074776,
      "grad_norm": 0.5778185725212097,
      "learning_rate": 0.00017739607666926598,
      "loss": 1.148,
      "step": 16200
    },
    {
      "epoch": 0.656793825084581,
      "grad_norm": 0.5841924548149109,
      "learning_rate": 0.00017736918279153904,
      "loss": 1.1808,
      "step": 16210
    },
    {
      "epoch": 0.6571990032616843,
      "grad_norm": 0.5658496618270874,
      "learning_rate": 0.0001773422749654988,
      "loss": 1.1356,
      "step": 16220
    },
    {
      "epoch": 0.6576041814387877,
      "grad_norm": 0.5653011798858643,
      "learning_rate": 0.00017731535319599626,
      "loss": 1.1607,
      "step": 16230
    },
    {
      "epoch": 0.6580093596158911,
      "grad_norm": 0.5546337962150574,
      "learning_rate": 0.00017728841748788495,
      "loss": 1.1201,
      "step": 16240
    },
    {
      "epoch": 0.6584145377929944,
      "grad_norm": 0.5673050880432129,
      "learning_rate": 0.0001772614678460209,
      "loss": 1.1373,
      "step": 16250
    },
    {
      "epoch": 0.6588197159700978,
      "grad_norm": 0.5388820171356201,
      "learning_rate": 0.00017723450427526268,
      "loss": 1.1427,
      "step": 16260
    },
    {
      "epoch": 0.6592248941472012,
      "grad_norm": 0.5974040627479553,
      "learning_rate": 0.00017720752678047138,
      "loss": 1.1507,
      "step": 16270
    },
    {
      "epoch": 0.6596300723243046,
      "grad_norm": 0.5815415382385254,
      "learning_rate": 0.00017718053536651044,
      "loss": 1.1564,
      "step": 16280
    },
    {
      "epoch": 0.660035250501408,
      "grad_norm": 0.5389411449432373,
      "learning_rate": 0.0001771535300382461,
      "loss": 1.1718,
      "step": 16290
    },
    {
      "epoch": 0.6604404286785114,
      "grad_norm": 0.5376825332641602,
      "learning_rate": 0.00017712651080054688,
      "loss": 1.1524,
      "step": 16300
    },
    {
      "epoch": 0.6608456068556148,
      "grad_norm": 0.6092084646224976,
      "learning_rate": 0.00017709947765828386,
      "loss": 1.1059,
      "step": 16310
    },
    {
      "epoch": 0.6612507850327182,
      "grad_norm": 0.5612658858299255,
      "learning_rate": 0.00017707243061633067,
      "loss": 1.1647,
      "step": 16320
    },
    {
      "epoch": 0.6616559632098216,
      "grad_norm": 0.6152335405349731,
      "learning_rate": 0.0001770453696795634,
      "loss": 1.1431,
      "step": 16330
    },
    {
      "epoch": 0.6620611413869248,
      "grad_norm": 0.5823425650596619,
      "learning_rate": 0.0001770182948528607,
      "loss": 1.1371,
      "step": 16340
    },
    {
      "epoch": 0.6624663195640282,
      "grad_norm": 0.5895346403121948,
      "learning_rate": 0.00017699120614110366,
      "loss": 1.1668,
      "step": 16350
    },
    {
      "epoch": 0.6628714977411316,
      "grad_norm": 0.5334838032722473,
      "learning_rate": 0.0001769641035491759,
      "loss": 1.1347,
      "step": 16360
    },
    {
      "epoch": 0.663276675918235,
      "grad_norm": 0.5610368251800537,
      "learning_rate": 0.00017693698708196354,
      "loss": 1.1312,
      "step": 16370
    },
    {
      "epoch": 0.6636818540953384,
      "grad_norm": 0.6461552381515503,
      "learning_rate": 0.00017690985674435523,
      "loss": 1.1647,
      "step": 16380
    },
    {
      "epoch": 0.6640870322724418,
      "grad_norm": 0.5891596674919128,
      "learning_rate": 0.00017688271254124208,
      "loss": 1.0986,
      "step": 16390
    },
    {
      "epoch": 0.6644922104495452,
      "grad_norm": 0.6188320517539978,
      "learning_rate": 0.00017685555447751773,
      "loss": 1.1369,
      "step": 16400
    },
    {
      "epoch": 0.6648973886266486,
      "grad_norm": 0.6021908521652222,
      "learning_rate": 0.0001768283825580783,
      "loss": 1.1114,
      "step": 16410
    },
    {
      "epoch": 0.665302566803752,
      "grad_norm": 0.5845525860786438,
      "learning_rate": 0.0001768011967878224,
      "loss": 1.1362,
      "step": 16420
    },
    {
      "epoch": 0.6657077449808553,
      "grad_norm": 0.5480151772499084,
      "learning_rate": 0.00017677399717165116,
      "loss": 1.1475,
      "step": 16430
    },
    {
      "epoch": 0.6661129231579587,
      "grad_norm": 0.5049716234207153,
      "learning_rate": 0.0001767467837144682,
      "loss": 1.1265,
      "step": 16440
    },
    {
      "epoch": 0.6665181013350621,
      "grad_norm": 0.49564483761787415,
      "learning_rate": 0.0001767195564211796,
      "loss": 1.1278,
      "step": 16450
    },
    {
      "epoch": 0.6669232795121655,
      "grad_norm": 0.5776987671852112,
      "learning_rate": 0.00017669231529669402,
      "loss": 1.1612,
      "step": 16460
    },
    {
      "epoch": 0.6673284576892689,
      "grad_norm": 0.524117112159729,
      "learning_rate": 0.00017666506034592252,
      "loss": 1.135,
      "step": 16470
    },
    {
      "epoch": 0.6677336358663722,
      "grad_norm": 0.555077075958252,
      "learning_rate": 0.0001766377915737787,
      "loss": 1.1678,
      "step": 16480
    },
    {
      "epoch": 0.6681388140434756,
      "grad_norm": 0.5465863943099976,
      "learning_rate": 0.00017661050898517863,
      "loss": 1.1185,
      "step": 16490
    },
    {
      "epoch": 0.668543992220579,
      "grad_norm": 0.5929831862449646,
      "learning_rate": 0.00017658321258504092,
      "loss": 1.1132,
      "step": 16500
    },
    {
      "epoch": 0.6689491703976824,
      "grad_norm": 0.6619282960891724,
      "learning_rate": 0.0001765559023782866,
      "loss": 1.1355,
      "step": 16510
    },
    {
      "epoch": 0.6693543485747857,
      "grad_norm": 0.6013734936714172,
      "learning_rate": 0.00017652857836983926,
      "loss": 1.1327,
      "step": 16520
    },
    {
      "epoch": 0.6697595267518891,
      "grad_norm": 0.5144134163856506,
      "learning_rate": 0.0001765012405646249,
      "loss": 1.145,
      "step": 16530
    },
    {
      "epoch": 0.6701647049289925,
      "grad_norm": 0.6188206076622009,
      "learning_rate": 0.0001764738889675721,
      "loss": 1.1427,
      "step": 16540
    },
    {
      "epoch": 0.6705698831060959,
      "grad_norm": 0.5501027703285217,
      "learning_rate": 0.00017644652358361182,
      "loss": 1.1638,
      "step": 16550
    },
    {
      "epoch": 0.6709750612831993,
      "grad_norm": 0.5230984091758728,
      "learning_rate": 0.00017641914441767758,
      "loss": 1.1554,
      "step": 16560
    },
    {
      "epoch": 0.6713802394603027,
      "grad_norm": 0.5615795254707336,
      "learning_rate": 0.00017639175147470538,
      "loss": 1.1367,
      "step": 16570
    },
    {
      "epoch": 0.6717854176374061,
      "grad_norm": 0.5082041025161743,
      "learning_rate": 0.0001763643447596337,
      "loss": 1.1365,
      "step": 16580
    },
    {
      "epoch": 0.6721905958145095,
      "grad_norm": 0.6520664095878601,
      "learning_rate": 0.00017633692427740345,
      "loss": 1.1475,
      "step": 16590
    },
    {
      "epoch": 0.6725957739916129,
      "grad_norm": 0.5827017426490784,
      "learning_rate": 0.0001763094900329581,
      "loss": 1.1493,
      "step": 16600
    },
    {
      "epoch": 0.6730009521687161,
      "grad_norm": 0.5913347005844116,
      "learning_rate": 0.00017628204203124355,
      "loss": 1.156,
      "step": 16610
    },
    {
      "epoch": 0.6734061303458195,
      "grad_norm": 0.5919530391693115,
      "learning_rate": 0.0001762545802772082,
      "loss": 1.1103,
      "step": 16620
    },
    {
      "epoch": 0.6738113085229229,
      "grad_norm": 0.6043758392333984,
      "learning_rate": 0.0001762271047758029,
      "loss": 1.1456,
      "step": 16630
    },
    {
      "epoch": 0.6742164867000263,
      "grad_norm": 0.5755093693733215,
      "learning_rate": 0.00017619961553198108,
      "loss": 1.142,
      "step": 16640
    },
    {
      "epoch": 0.6746216648771297,
      "grad_norm": 0.6099788546562195,
      "learning_rate": 0.00017617211255069846,
      "loss": 1.1339,
      "step": 16650
    },
    {
      "epoch": 0.6750268430542331,
      "grad_norm": 0.5540536046028137,
      "learning_rate": 0.00017614459583691346,
      "loss": 1.1472,
      "step": 16660
    },
    {
      "epoch": 0.6754320212313365,
      "grad_norm": 0.6255536675453186,
      "learning_rate": 0.00017611706539558675,
      "loss": 1.1476,
      "step": 16670
    },
    {
      "epoch": 0.6758371994084399,
      "grad_norm": 0.5504527688026428,
      "learning_rate": 0.00017608952123168172,
      "loss": 1.1509,
      "step": 16680
    },
    {
      "epoch": 0.6762423775855433,
      "grad_norm": 0.585088312625885,
      "learning_rate": 0.00017606196335016397,
      "loss": 1.1441,
      "step": 16690
    },
    {
      "epoch": 0.6766475557626466,
      "grad_norm": 0.5804777145385742,
      "learning_rate": 0.0001760343917560018,
      "loss": 1.1022,
      "step": 16700
    },
    {
      "epoch": 0.67705273393975,
      "grad_norm": 0.5427889823913574,
      "learning_rate": 0.00017600680645416583,
      "loss": 1.1271,
      "step": 16710
    },
    {
      "epoch": 0.6774579121168534,
      "grad_norm": 0.5819441080093384,
      "learning_rate": 0.00017597920744962924,
      "loss": 1.1332,
      "step": 16720
    },
    {
      "epoch": 0.6778630902939567,
      "grad_norm": 0.562942624092102,
      "learning_rate": 0.00017595159474736763,
      "loss": 1.1395,
      "step": 16730
    },
    {
      "epoch": 0.6782682684710601,
      "grad_norm": 0.5921532511711121,
      "learning_rate": 0.0001759239683523591,
      "loss": 1.1128,
      "step": 16740
    },
    {
      "epoch": 0.6786734466481635,
      "grad_norm": 0.6038963794708252,
      "learning_rate": 0.00017589632826958424,
      "loss": 1.1555,
      "step": 16750
    },
    {
      "epoch": 0.6790786248252669,
      "grad_norm": 0.6352828741073608,
      "learning_rate": 0.000175868674504026,
      "loss": 1.1464,
      "step": 16760
    },
    {
      "epoch": 0.6794838030023703,
      "grad_norm": 0.5518133640289307,
      "learning_rate": 0.00017584100706066993,
      "loss": 1.1432,
      "step": 16770
    },
    {
      "epoch": 0.6798889811794737,
      "grad_norm": 0.5416463017463684,
      "learning_rate": 0.00017581332594450392,
      "loss": 1.1101,
      "step": 16780
    },
    {
      "epoch": 0.680294159356577,
      "grad_norm": 0.5850431323051453,
      "learning_rate": 0.00017578563116051851,
      "loss": 1.1807,
      "step": 16790
    },
    {
      "epoch": 0.6806993375336804,
      "grad_norm": 0.5252676606178284,
      "learning_rate": 0.00017575792271370646,
      "loss": 1.1326,
      "step": 16800
    },
    {
      "epoch": 0.6811045157107838,
      "grad_norm": 0.5673633217811584,
      "learning_rate": 0.0001757302006090632,
      "loss": 1.1272,
      "step": 16810
    },
    {
      "epoch": 0.6815096938878872,
      "grad_norm": 0.6091298460960388,
      "learning_rate": 0.0001757024648515865,
      "loss": 1.1387,
      "step": 16820
    },
    {
      "epoch": 0.6819148720649906,
      "grad_norm": 0.6110486388206482,
      "learning_rate": 0.00017567471544627667,
      "loss": 1.1413,
      "step": 16830
    },
    {
      "epoch": 0.682320050242094,
      "grad_norm": 0.5540281534194946,
      "learning_rate": 0.00017564695239813644,
      "loss": 1.1488,
      "step": 16840
    },
    {
      "epoch": 0.6827252284191974,
      "grad_norm": 0.5677554607391357,
      "learning_rate": 0.00017561917571217093,
      "loss": 1.1695,
      "step": 16850
    },
    {
      "epoch": 0.6831304065963008,
      "grad_norm": 0.6077812314033508,
      "learning_rate": 0.0001755913853933879,
      "loss": 1.155,
      "step": 16860
    },
    {
      "epoch": 0.6835355847734041,
      "grad_norm": 0.5653752088546753,
      "learning_rate": 0.00017556358144679737,
      "loss": 1.1281,
      "step": 16870
    },
    {
      "epoch": 0.6839407629505074,
      "grad_norm": 0.6167461276054382,
      "learning_rate": 0.00017553576387741195,
      "loss": 1.1589,
      "step": 16880
    },
    {
      "epoch": 0.6843459411276108,
      "grad_norm": 0.5953029990196228,
      "learning_rate": 0.00017550793269024666,
      "loss": 1.1529,
      "step": 16890
    },
    {
      "epoch": 0.6847511193047142,
      "grad_norm": 0.5916039347648621,
      "learning_rate": 0.00017548008789031899,
      "loss": 1.1016,
      "step": 16900
    },
    {
      "epoch": 0.6851562974818176,
      "grad_norm": 0.6084550023078918,
      "learning_rate": 0.00017545222948264883,
      "loss": 1.1238,
      "step": 16910
    },
    {
      "epoch": 0.685561475658921,
      "grad_norm": 0.5103800296783447,
      "learning_rate": 0.0001754243574722586,
      "loss": 1.0871,
      "step": 16920
    },
    {
      "epoch": 0.6859666538360244,
      "grad_norm": 0.6715081334114075,
      "learning_rate": 0.00017539647186417311,
      "loss": 1.0782,
      "step": 16930
    },
    {
      "epoch": 0.6863718320131278,
      "grad_norm": 0.5086271166801453,
      "learning_rate": 0.0001753685726634197,
      "loss": 1.0871,
      "step": 16940
    },
    {
      "epoch": 0.6867770101902312,
      "grad_norm": 0.6798851490020752,
      "learning_rate": 0.00017534065987502803,
      "loss": 1.1481,
      "step": 16950
    },
    {
      "epoch": 0.6871821883673346,
      "grad_norm": 0.574505090713501,
      "learning_rate": 0.00017531273350403036,
      "loss": 1.1639,
      "step": 16960
    },
    {
      "epoch": 0.6875873665444379,
      "grad_norm": 0.6019743084907532,
      "learning_rate": 0.0001752847935554613,
      "loss": 1.1264,
      "step": 16970
    },
    {
      "epoch": 0.6879925447215413,
      "grad_norm": 0.5868002772331238,
      "learning_rate": 0.00017525684003435793,
      "loss": 1.1584,
      "step": 16980
    },
    {
      "epoch": 0.6883977228986446,
      "grad_norm": 0.5477590560913086,
      "learning_rate": 0.00017522887294575977,
      "loss": 1.1284,
      "step": 16990
    },
    {
      "epoch": 0.688802901075748,
      "grad_norm": 0.5964993834495544,
      "learning_rate": 0.00017520089229470883,
      "loss": 1.1014,
      "step": 17000
    },
    {
      "epoch": 0.6892080792528514,
      "grad_norm": 0.5745593905448914,
      "learning_rate": 0.00017517289808624952,
      "loss": 1.1502,
      "step": 17010
    },
    {
      "epoch": 0.6896132574299548,
      "grad_norm": 0.5281778573989868,
      "learning_rate": 0.00017514489032542869,
      "loss": 1.1608,
      "step": 17020
    },
    {
      "epoch": 0.6900184356070582,
      "grad_norm": 0.5827769637107849,
      "learning_rate": 0.0001751168690172957,
      "loss": 1.1745,
      "step": 17030
    },
    {
      "epoch": 0.6904236137841616,
      "grad_norm": 0.5310696959495544,
      "learning_rate": 0.00017508883416690222,
      "loss": 1.1475,
      "step": 17040
    },
    {
      "epoch": 0.690828791961265,
      "grad_norm": 0.6806015372276306,
      "learning_rate": 0.00017506078577930251,
      "loss": 1.1512,
      "step": 17050
    },
    {
      "epoch": 0.6912339701383683,
      "grad_norm": 0.5830384492874146,
      "learning_rate": 0.00017503272385955318,
      "loss": 1.1097,
      "step": 17060
    },
    {
      "epoch": 0.6916391483154717,
      "grad_norm": 0.5918838977813721,
      "learning_rate": 0.00017500464841271334,
      "loss": 1.1699,
      "step": 17070
    },
    {
      "epoch": 0.6920443264925751,
      "grad_norm": 0.48129186034202576,
      "learning_rate": 0.00017497655944384444,
      "loss": 1.15,
      "step": 17080
    },
    {
      "epoch": 0.6924495046696785,
      "grad_norm": 0.6589459776878357,
      "learning_rate": 0.00017494845695801049,
      "loss": 1.1493,
      "step": 17090
    },
    {
      "epoch": 0.6928546828467819,
      "grad_norm": 0.6072139739990234,
      "learning_rate": 0.00017492034096027782,
      "loss": 1.13,
      "step": 17100
    },
    {
      "epoch": 0.6932598610238853,
      "grad_norm": 0.5517972111701965,
      "learning_rate": 0.0001748922114557153,
      "loss": 1.1672,
      "step": 17110
    },
    {
      "epoch": 0.6936650392009887,
      "grad_norm": 0.5277575254440308,
      "learning_rate": 0.0001748640684493942,
      "loss": 1.1828,
      "step": 17120
    },
    {
      "epoch": 0.694070217378092,
      "grad_norm": 0.5691710710525513,
      "learning_rate": 0.00017483591194638817,
      "loss": 1.1122,
      "step": 17130
    },
    {
      "epoch": 0.6944753955551954,
      "grad_norm": 0.5930890440940857,
      "learning_rate": 0.00017480774195177337,
      "loss": 1.1302,
      "step": 17140
    },
    {
      "epoch": 0.6948805737322987,
      "grad_norm": 0.5476669073104858,
      "learning_rate": 0.0001747795584706283,
      "loss": 1.1282,
      "step": 17150
    },
    {
      "epoch": 0.6952857519094021,
      "grad_norm": 0.5614034533500671,
      "learning_rate": 0.00017475136150803405,
      "loss": 1.1601,
      "step": 17160
    },
    {
      "epoch": 0.6956909300865055,
      "grad_norm": 0.532279908657074,
      "learning_rate": 0.00017472315106907394,
      "loss": 1.1439,
      "step": 17170
    },
    {
      "epoch": 0.6960961082636089,
      "grad_norm": 0.6109350919723511,
      "learning_rate": 0.0001746949271588339,
      "loss": 1.1483,
      "step": 17180
    },
    {
      "epoch": 0.6965012864407123,
      "grad_norm": 0.551643967628479,
      "learning_rate": 0.00017466668978240212,
      "loss": 1.1551,
      "step": 17190
    },
    {
      "epoch": 0.6969064646178157,
      "grad_norm": 0.5155463218688965,
      "learning_rate": 0.00017463843894486937,
      "loss": 1.1563,
      "step": 17200
    },
    {
      "epoch": 0.6973116427949191,
      "grad_norm": 0.527687132358551,
      "learning_rate": 0.00017461017465132878,
      "loss": 1.0913,
      "step": 17210
    },
    {
      "epoch": 0.6977168209720225,
      "grad_norm": 0.6476535797119141,
      "learning_rate": 0.0001745818969068759,
      "loss": 1.1765,
      "step": 17220
    },
    {
      "epoch": 0.6981219991491259,
      "grad_norm": 0.5405405759811401,
      "learning_rate": 0.00017455360571660871,
      "loss": 1.1058,
      "step": 17230
    },
    {
      "epoch": 0.6985271773262292,
      "grad_norm": 0.5532897710800171,
      "learning_rate": 0.00017452530108562762,
      "loss": 1.1016,
      "step": 17240
    },
    {
      "epoch": 0.6989323555033325,
      "grad_norm": 0.5525125861167908,
      "learning_rate": 0.00017449698301903545,
      "loss": 1.1415,
      "step": 17250
    },
    {
      "epoch": 0.6993375336804359,
      "grad_norm": 0.6306107044219971,
      "learning_rate": 0.0001744686515219375,
      "loss": 1.1508,
      "step": 17260
    },
    {
      "epoch": 0.6997427118575393,
      "grad_norm": 0.5929704308509827,
      "learning_rate": 0.00017444030659944138,
      "loss": 1.144,
      "step": 17270
    },
    {
      "epoch": 0.7001478900346427,
      "grad_norm": 0.6231330633163452,
      "learning_rate": 0.00017441194825665722,
      "loss": 1.1441,
      "step": 17280
    },
    {
      "epoch": 0.7005530682117461,
      "grad_norm": 0.6167316436767578,
      "learning_rate": 0.0001743835764986975,
      "loss": 1.1354,
      "step": 17290
    },
    {
      "epoch": 0.7009582463888495,
      "grad_norm": 0.5757057666778564,
      "learning_rate": 0.00017435519133067722,
      "loss": 1.1574,
      "step": 17300
    },
    {
      "epoch": 0.7013634245659529,
      "grad_norm": 0.5963408946990967,
      "learning_rate": 0.00017432679275771368,
      "loss": 1.1693,
      "step": 17310
    },
    {
      "epoch": 0.7017686027430563,
      "grad_norm": 0.638995349407196,
      "learning_rate": 0.00017429838078492665,
      "loss": 1.1423,
      "step": 17320
    },
    {
      "epoch": 0.7021737809201596,
      "grad_norm": 0.5907160639762878,
      "learning_rate": 0.00017426995541743834,
      "loss": 1.129,
      "step": 17330
    },
    {
      "epoch": 0.702578959097263,
      "grad_norm": 0.5888515114784241,
      "learning_rate": 0.00017424151666037329,
      "loss": 1.1141,
      "step": 17340
    },
    {
      "epoch": 0.7029841372743664,
      "grad_norm": 0.5622380375862122,
      "learning_rate": 0.00017421306451885857,
      "loss": 1.1677,
      "step": 17350
    },
    {
      "epoch": 0.7033893154514698,
      "grad_norm": 0.5128424167633057,
      "learning_rate": 0.0001741845989980236,
      "loss": 1.1137,
      "step": 17360
    },
    {
      "epoch": 0.7037944936285732,
      "grad_norm": 0.5116934776306152,
      "learning_rate": 0.0001741561201030002,
      "loss": 1.1592,
      "step": 17370
    },
    {
      "epoch": 0.7041996718056766,
      "grad_norm": 0.570137619972229,
      "learning_rate": 0.0001741276278389226,
      "loss": 1.1401,
      "step": 17380
    },
    {
      "epoch": 0.70460484998278,
      "grad_norm": 0.5379998683929443,
      "learning_rate": 0.00017409912221092748,
      "loss": 1.1438,
      "step": 17390
    },
    {
      "epoch": 0.7050100281598833,
      "grad_norm": 0.5290420055389404,
      "learning_rate": 0.0001740706032241539,
      "loss": 1.1327,
      "step": 17400
    },
    {
      "epoch": 0.7054152063369867,
      "grad_norm": 0.5894034504890442,
      "learning_rate": 0.00017404207088374333,
      "loss": 1.1595,
      "step": 17410
    },
    {
      "epoch": 0.70582038451409,
      "grad_norm": 0.6148152947425842,
      "learning_rate": 0.0001740135251948397,
      "loss": 1.0904,
      "step": 17420
    },
    {
      "epoch": 0.7062255626911934,
      "grad_norm": 0.561274528503418,
      "learning_rate": 0.00017398496616258924,
      "loss": 1.141,
      "step": 17430
    },
    {
      "epoch": 0.7066307408682968,
      "grad_norm": 0.5067313313484192,
      "learning_rate": 0.00017395639379214065,
      "loss": 1.1599,
      "step": 17440
    },
    {
      "epoch": 0.7070359190454002,
      "grad_norm": 0.6468285322189331,
      "learning_rate": 0.00017392780808864506,
      "loss": 1.1581,
      "step": 17450
    },
    {
      "epoch": 0.7074410972225036,
      "grad_norm": 0.5342162847518921,
      "learning_rate": 0.00017389920905725597,
      "loss": 1.1261,
      "step": 17460
    },
    {
      "epoch": 0.707846275399607,
      "grad_norm": 0.5501462817192078,
      "learning_rate": 0.00017387059670312925,
      "loss": 1.1586,
      "step": 17470
    },
    {
      "epoch": 0.7082514535767104,
      "grad_norm": 0.6257019639015198,
      "learning_rate": 0.00017384197103142328,
      "loss": 1.1091,
      "step": 17480
    },
    {
      "epoch": 0.7086566317538138,
      "grad_norm": 0.5524917244911194,
      "learning_rate": 0.00017381333204729868,
      "loss": 1.1673,
      "step": 17490
    },
    {
      "epoch": 0.7090618099309172,
      "grad_norm": 0.528265118598938,
      "learning_rate": 0.00017378467975591862,
      "loss": 1.1869,
      "step": 17500
    },
    {
      "epoch": 0.7094669881080204,
      "grad_norm": 0.5930017232894897,
      "learning_rate": 0.0001737560141624486,
      "loss": 1.1533,
      "step": 17510
    },
    {
      "epoch": 0.7098721662851238,
      "grad_norm": 0.5824956297874451,
      "learning_rate": 0.0001737273352720565,
      "loss": 1.1393,
      "step": 17520
    },
    {
      "epoch": 0.7102773444622272,
      "grad_norm": 1.5536516904830933,
      "learning_rate": 0.00017369864308991266,
      "loss": 1.1403,
      "step": 17530
    },
    {
      "epoch": 0.7106825226393306,
      "grad_norm": 0.5552017688751221,
      "learning_rate": 0.00017366993762118972,
      "loss": 1.1092,
      "step": 17540
    },
    {
      "epoch": 0.711087700816434,
      "grad_norm": 0.5798711776733398,
      "learning_rate": 0.00017364121887106286,
      "loss": 1.1298,
      "step": 17550
    },
    {
      "epoch": 0.7114928789935374,
      "grad_norm": 0.6301178932189941,
      "learning_rate": 0.0001736124868447095,
      "loss": 1.1288,
      "step": 17560
    },
    {
      "epoch": 0.7118980571706408,
      "grad_norm": 0.5595502853393555,
      "learning_rate": 0.00017358374154730955,
      "loss": 1.151,
      "step": 17570
    },
    {
      "epoch": 0.7123032353477442,
      "grad_norm": 0.5434761643409729,
      "learning_rate": 0.00017355498298404526,
      "loss": 1.1645,
      "step": 17580
    },
    {
      "epoch": 0.7127084135248476,
      "grad_norm": 0.6034620404243469,
      "learning_rate": 0.00017352621116010136,
      "loss": 1.1615,
      "step": 17590
    },
    {
      "epoch": 0.7131135917019509,
      "grad_norm": 0.5743139982223511,
      "learning_rate": 0.00017349742608066483,
      "loss": 1.1138,
      "step": 17600
    },
    {
      "epoch": 0.7135187698790543,
      "grad_norm": 0.6469864249229431,
      "learning_rate": 0.00017346862775092522,
      "loss": 1.1273,
      "step": 17610
    },
    {
      "epoch": 0.7139239480561577,
      "grad_norm": 0.5544672608375549,
      "learning_rate": 0.00017343981617607424,
      "loss": 1.1315,
      "step": 17620
    },
    {
      "epoch": 0.7143291262332611,
      "grad_norm": 0.545799732208252,
      "learning_rate": 0.00017341099136130618,
      "loss": 1.1461,
      "step": 17630
    },
    {
      "epoch": 0.7147343044103645,
      "grad_norm": 0.6135145425796509,
      "learning_rate": 0.00017338215331181764,
      "loss": 1.1498,
      "step": 17640
    },
    {
      "epoch": 0.7151394825874678,
      "grad_norm": 0.6106663942337036,
      "learning_rate": 0.00017335330203280765,
      "loss": 1.1406,
      "step": 17650
    },
    {
      "epoch": 0.7155446607645712,
      "grad_norm": 0.606274425983429,
      "learning_rate": 0.00017332443752947753,
      "loss": 1.1162,
      "step": 17660
    },
    {
      "epoch": 0.7159498389416746,
      "grad_norm": 0.5891724228858948,
      "learning_rate": 0.00017329555980703112,
      "loss": 1.133,
      "step": 17670
    },
    {
      "epoch": 0.716355017118778,
      "grad_norm": 0.5282707214355469,
      "learning_rate": 0.0001732666688706745,
      "loss": 1.1119,
      "step": 17680
    },
    {
      "epoch": 0.7167601952958814,
      "grad_norm": 0.5621887445449829,
      "learning_rate": 0.00017323776472561627,
      "loss": 1.1351,
      "step": 17690
    },
    {
      "epoch": 0.7171653734729847,
      "grad_norm": 0.6223745942115784,
      "learning_rate": 0.00017320884737706725,
      "loss": 1.1477,
      "step": 17700
    },
    {
      "epoch": 0.7175705516500881,
      "grad_norm": 0.5834466218948364,
      "learning_rate": 0.00017317991683024083,
      "loss": 1.1074,
      "step": 17710
    },
    {
      "epoch": 0.7179757298271915,
      "grad_norm": 0.5502654314041138,
      "learning_rate": 0.0001731509730903526,
      "loss": 1.1419,
      "step": 17720
    },
    {
      "epoch": 0.7183809080042949,
      "grad_norm": 0.6260347962379456,
      "learning_rate": 0.00017312201616262072,
      "loss": 1.1206,
      "step": 17730
    },
    {
      "epoch": 0.7187860861813983,
      "grad_norm": 0.5147431492805481,
      "learning_rate": 0.0001730930460522655,
      "loss": 1.1526,
      "step": 17740
    },
    {
      "epoch": 0.7191912643585017,
      "grad_norm": 0.5982386469841003,
      "learning_rate": 0.0001730640627645098,
      "loss": 1.1444,
      "step": 17750
    },
    {
      "epoch": 0.7195964425356051,
      "grad_norm": 0.5677700042724609,
      "learning_rate": 0.0001730350663045788,
      "loss": 1.1365,
      "step": 17760
    },
    {
      "epoch": 0.7200016207127085,
      "grad_norm": 0.5780327320098877,
      "learning_rate": 0.00017300605667770004,
      "loss": 1.1738,
      "step": 17770
    },
    {
      "epoch": 0.7204067988898119,
      "grad_norm": 0.5017571449279785,
      "learning_rate": 0.0001729770338891035,
      "loss": 1.1141,
      "step": 17780
    },
    {
      "epoch": 0.7208119770669151,
      "grad_norm": 0.5772636532783508,
      "learning_rate": 0.00017294799794402143,
      "loss": 1.1634,
      "step": 17790
    },
    {
      "epoch": 0.7212171552440185,
      "grad_norm": 0.6200149059295654,
      "learning_rate": 0.0001729189488476885,
      "loss": 1.1513,
      "step": 17800
    },
    {
      "epoch": 0.7216223334211219,
      "grad_norm": 0.6855908632278442,
      "learning_rate": 0.0001728898866053418,
      "loss": 1.1682,
      "step": 17810
    },
    {
      "epoch": 0.7220275115982253,
      "grad_norm": 0.6057910323143005,
      "learning_rate": 0.00017286081122222069,
      "loss": 1.1292,
      "step": 17820
    },
    {
      "epoch": 0.7224326897753287,
      "grad_norm": 0.545036792755127,
      "learning_rate": 0.000172831722703567,
      "loss": 1.1254,
      "step": 17830
    },
    {
      "epoch": 0.7228378679524321,
      "grad_norm": 0.49617084860801697,
      "learning_rate": 0.00017280262105462483,
      "loss": 1.1596,
      "step": 17840
    },
    {
      "epoch": 0.7232430461295355,
      "grad_norm": 0.5577792525291443,
      "learning_rate": 0.00017277350628064076,
      "loss": 1.1235,
      "step": 17850
    },
    {
      "epoch": 0.7236482243066389,
      "grad_norm": 0.5978285670280457,
      "learning_rate": 0.00017274437838686368,
      "loss": 1.1434,
      "step": 17860
    },
    {
      "epoch": 0.7240534024837423,
      "grad_norm": 0.5576709508895874,
      "learning_rate": 0.0001727152373785448,
      "loss": 1.1743,
      "step": 17870
    },
    {
      "epoch": 0.7244585806608456,
      "grad_norm": 0.49729374051094055,
      "learning_rate": 0.0001726860832609377,
      "loss": 1.1461,
      "step": 17880
    },
    {
      "epoch": 0.724863758837949,
      "grad_norm": 0.563522458076477,
      "learning_rate": 0.00017265691603929843,
      "loss": 1.133,
      "step": 17890
    },
    {
      "epoch": 0.7252689370150524,
      "grad_norm": 0.5564418435096741,
      "learning_rate": 0.0001726277357188853,
      "loss": 1.1443,
      "step": 17900
    },
    {
      "epoch": 0.7256741151921557,
      "grad_norm": 0.6351546049118042,
      "learning_rate": 0.000172598542304959,
      "loss": 1.1654,
      "step": 17910
    },
    {
      "epoch": 0.7260792933692591,
      "grad_norm": 0.5955491065979004,
      "learning_rate": 0.00017256933580278268,
      "loss": 1.116,
      "step": 17920
    },
    {
      "epoch": 0.7264844715463625,
      "grad_norm": 0.6177263259887695,
      "learning_rate": 0.00017254011621762162,
      "loss": 1.1004,
      "step": 17930
    },
    {
      "epoch": 0.7268896497234659,
      "grad_norm": 0.7096222043037415,
      "learning_rate": 0.0001725108835547437,
      "loss": 1.174,
      "step": 17940
    },
    {
      "epoch": 0.7272948279005693,
      "grad_norm": 0.5860795378684998,
      "learning_rate": 0.00017248163781941904,
      "loss": 1.1288,
      "step": 17950
    },
    {
      "epoch": 0.7277000060776727,
      "grad_norm": 0.5619387626647949,
      "learning_rate": 0.00017245237901692014,
      "loss": 1.122,
      "step": 17960
    },
    {
      "epoch": 0.728105184254776,
      "grad_norm": 0.56015545129776,
      "learning_rate": 0.0001724231071525218,
      "loss": 1.1288,
      "step": 17970
    },
    {
      "epoch": 0.7285103624318794,
      "grad_norm": 0.5271645784378052,
      "learning_rate": 0.00017239382223150132,
      "loss": 1.1742,
      "step": 17980
    },
    {
      "epoch": 0.7289155406089828,
      "grad_norm": 0.5860309600830078,
      "learning_rate": 0.00017236452425913815,
      "loss": 1.1719,
      "step": 17990
    },
    {
      "epoch": 0.7293207187860862,
      "grad_norm": 0.6424168944358826,
      "learning_rate": 0.00017233521324071428,
      "loss": 1.1376,
      "step": 18000
    },
    {
      "epoch": 0.7297258969631896,
      "grad_norm": 0.5836331248283386,
      "learning_rate": 0.00017230588918151397,
      "loss": 1.1173,
      "step": 18010
    },
    {
      "epoch": 0.730131075140293,
      "grad_norm": 0.6537491083145142,
      "learning_rate": 0.00017227655208682384,
      "loss": 1.1146,
      "step": 18020
    },
    {
      "epoch": 0.7305362533173964,
      "grad_norm": 0.5470165610313416,
      "learning_rate": 0.0001722472019619328,
      "loss": 1.127,
      "step": 18030
    },
    {
      "epoch": 0.7309414314944997,
      "grad_norm": 0.5129386782646179,
      "learning_rate": 0.0001722178388121322,
      "loss": 1.1275,
      "step": 18040
    },
    {
      "epoch": 0.7313466096716031,
      "grad_norm": 0.605952262878418,
      "learning_rate": 0.00017218846264271573,
      "loss": 1.1502,
      "step": 18050
    },
    {
      "epoch": 0.7317517878487064,
      "grad_norm": 0.6324777603149414,
      "learning_rate": 0.00017215907345897934,
      "loss": 1.1178,
      "step": 18060
    },
    {
      "epoch": 0.7321569660258098,
      "grad_norm": 0.6020564436912537,
      "learning_rate": 0.0001721296712662215,
      "loss": 1.1727,
      "step": 18070
    },
    {
      "epoch": 0.7325621442029132,
      "grad_norm": 0.5403150916099548,
      "learning_rate": 0.00017210025606974276,
      "loss": 1.0907,
      "step": 18080
    },
    {
      "epoch": 0.7329673223800166,
      "grad_norm": 0.5958328247070312,
      "learning_rate": 0.00017207082787484626,
      "loss": 1.1111,
      "step": 18090
    },
    {
      "epoch": 0.73337250055712,
      "grad_norm": 0.5726959109306335,
      "learning_rate": 0.0001720413866868374,
      "loss": 1.1589,
      "step": 18100
    },
    {
      "epoch": 0.7337776787342234,
      "grad_norm": 0.8227291107177734,
      "learning_rate": 0.00017201193251102382,
      "loss": 1.1225,
      "step": 18110
    },
    {
      "epoch": 0.7341828569113268,
      "grad_norm": 0.5481386780738831,
      "learning_rate": 0.00017198246535271572,
      "loss": 1.148,
      "step": 18120
    },
    {
      "epoch": 0.7345880350884302,
      "grad_norm": 0.5329793095588684,
      "learning_rate": 0.00017195298521722544,
      "loss": 1.0939,
      "step": 18130
    },
    {
      "epoch": 0.7349932132655336,
      "grad_norm": 0.5352697372436523,
      "learning_rate": 0.00017192349210986774,
      "loss": 1.1187,
      "step": 18140
    },
    {
      "epoch": 0.7353983914426369,
      "grad_norm": 0.5483496189117432,
      "learning_rate": 0.00017189398603595973,
      "loss": 1.1011,
      "step": 18150
    },
    {
      "epoch": 0.7358035696197402,
      "grad_norm": 0.7455344200134277,
      "learning_rate": 0.0001718644670008208,
      "loss": 1.1516,
      "step": 18160
    },
    {
      "epoch": 0.7362087477968436,
      "grad_norm": 0.5520568490028381,
      "learning_rate": 0.00017183493500977278,
      "loss": 1.1198,
      "step": 18170
    },
    {
      "epoch": 0.736613925973947,
      "grad_norm": 0.5446961522102356,
      "learning_rate": 0.0001718053900681397,
      "loss": 1.1598,
      "step": 18180
    },
    {
      "epoch": 0.7370191041510504,
      "grad_norm": 0.5898950099945068,
      "learning_rate": 0.00017177583218124805,
      "loss": 1.1175,
      "step": 18190
    },
    {
      "epoch": 0.7374242823281538,
      "grad_norm": 0.5833065509796143,
      "learning_rate": 0.00017174626135442662,
      "loss": 1.1378,
      "step": 18200
    },
    {
      "epoch": 0.7378294605052572,
      "grad_norm": 0.5848112106323242,
      "learning_rate": 0.00017171667759300645,
      "loss": 1.0705,
      "step": 18210
    },
    {
      "epoch": 0.7382346386823606,
      "grad_norm": 0.5205530524253845,
      "learning_rate": 0.00017168708090232101,
      "loss": 1.103,
      "step": 18220
    },
    {
      "epoch": 0.738639816859464,
      "grad_norm": 0.5834738612174988,
      "learning_rate": 0.00017165747128770607,
      "loss": 1.113,
      "step": 18230
    },
    {
      "epoch": 0.7390449950365673,
      "grad_norm": 0.5417637825012207,
      "learning_rate": 0.00017162784875449972,
      "loss": 1.0948,
      "step": 18240
    },
    {
      "epoch": 0.7394501732136707,
      "grad_norm": 0.6023304462432861,
      "learning_rate": 0.00017159821330804236,
      "loss": 1.1266,
      "step": 18250
    },
    {
      "epoch": 0.7398553513907741,
      "grad_norm": 0.5816308259963989,
      "learning_rate": 0.0001715685649536768,
      "loss": 1.1634,
      "step": 18260
    },
    {
      "epoch": 0.7402605295678775,
      "grad_norm": 0.5546183586120605,
      "learning_rate": 0.00017153890369674807,
      "loss": 1.1348,
      "step": 18270
    },
    {
      "epoch": 0.7406657077449809,
      "grad_norm": 0.5743983387947083,
      "learning_rate": 0.00017150922954260365,
      "loss": 1.1311,
      "step": 18280
    },
    {
      "epoch": 0.7410708859220843,
      "grad_norm": 0.6121779680252075,
      "learning_rate": 0.00017147954249659316,
      "loss": 1.1438,
      "step": 18290
    },
    {
      "epoch": 0.7414760640991876,
      "grad_norm": 0.5894224047660828,
      "learning_rate": 0.00017144984256406876,
      "loss": 1.1593,
      "step": 18300
    },
    {
      "epoch": 0.741881242276291,
      "grad_norm": 0.5817199349403381,
      "learning_rate": 0.00017142012975038476,
      "loss": 1.1193,
      "step": 18310
    },
    {
      "epoch": 0.7422864204533944,
      "grad_norm": 0.6144255995750427,
      "learning_rate": 0.00017139040406089786,
      "loss": 1.1452,
      "step": 18320
    },
    {
      "epoch": 0.7426915986304977,
      "grad_norm": 0.5486526489257812,
      "learning_rate": 0.00017136066550096715,
      "loss": 1.1145,
      "step": 18330
    },
    {
      "epoch": 0.7430967768076011,
      "grad_norm": 0.6116833090782166,
      "learning_rate": 0.0001713309140759539,
      "loss": 1.1262,
      "step": 18340
    },
    {
      "epoch": 0.7435019549847045,
      "grad_norm": 0.5812392234802246,
      "learning_rate": 0.00017130114979122186,
      "loss": 1.1549,
      "step": 18350
    },
    {
      "epoch": 0.7439071331618079,
      "grad_norm": 0.6728997826576233,
      "learning_rate": 0.0001712713726521369,
      "loss": 1.1452,
      "step": 18360
    },
    {
      "epoch": 0.7443123113389113,
      "grad_norm": 0.5823251605033875,
      "learning_rate": 0.00017124158266406737,
      "loss": 1.1543,
      "step": 18370
    },
    {
      "epoch": 0.7447174895160147,
      "grad_norm": 0.6307849287986755,
      "learning_rate": 0.00017121177983238394,
      "loss": 1.1202,
      "step": 18380
    },
    {
      "epoch": 0.7451226676931181,
      "grad_norm": 0.5969803929328918,
      "learning_rate": 0.00017118196416245947,
      "loss": 1.1528,
      "step": 18390
    },
    {
      "epoch": 0.7455278458702215,
      "grad_norm": 0.6667117476463318,
      "learning_rate": 0.0001711521356596692,
      "loss": 1.1725,
      "step": 18400
    },
    {
      "epoch": 0.7459330240473249,
      "grad_norm": 0.61086106300354,
      "learning_rate": 0.00017112229432939078,
      "loss": 1.148,
      "step": 18410
    },
    {
      "epoch": 0.7463382022244281,
      "grad_norm": 0.6226322650909424,
      "learning_rate": 0.00017109244017700397,
      "loss": 1.1565,
      "step": 18420
    },
    {
      "epoch": 0.7467433804015315,
      "grad_norm": 0.6598115563392639,
      "learning_rate": 0.000171062573207891,
      "loss": 1.1527,
      "step": 18430
    },
    {
      "epoch": 0.7471485585786349,
      "grad_norm": 0.5530314445495605,
      "learning_rate": 0.00017103269342743635,
      "loss": 1.1739,
      "step": 18440
    },
    {
      "epoch": 0.7475537367557383,
      "grad_norm": 0.6161337494850159,
      "learning_rate": 0.00017100280084102684,
      "loss": 1.1236,
      "step": 18450
    },
    {
      "epoch": 0.7479589149328417,
      "grad_norm": 0.6366524696350098,
      "learning_rate": 0.0001709728954540516,
      "loss": 1.1192,
      "step": 18460
    },
    {
      "epoch": 0.7483640931099451,
      "grad_norm": 0.6361158490180969,
      "learning_rate": 0.000170942977271902,
      "loss": 1.1348,
      "step": 18470
    },
    {
      "epoch": 0.7487692712870485,
      "grad_norm": 0.5445762276649475,
      "learning_rate": 0.00017091304629997183,
      "loss": 1.123,
      "step": 18480
    },
    {
      "epoch": 0.7491744494641519,
      "grad_norm": 0.5994340181350708,
      "learning_rate": 0.00017088310254365703,
      "loss": 1.1075,
      "step": 18490
    },
    {
      "epoch": 0.7495796276412553,
      "grad_norm": 0.6032571196556091,
      "learning_rate": 0.000170853146008356,
      "loss": 1.1438,
      "step": 18500
    },
    {
      "epoch": 0.7499848058183586,
      "grad_norm": 0.5596930384635925,
      "learning_rate": 0.00017082317669946942,
      "loss": 1.1092,
      "step": 18510
    },
    {
      "epoch": 0.750389983995462,
      "grad_norm": 0.5684100389480591,
      "learning_rate": 0.00017079319462240012,
      "loss": 1.118,
      "step": 18520
    },
    {
      "epoch": 0.7507951621725654,
      "grad_norm": 0.5682975649833679,
      "learning_rate": 0.00017076319978255345,
      "loss": 1.1444,
      "step": 18530
    },
    {
      "epoch": 0.7512003403496688,
      "grad_norm": 0.6312071084976196,
      "learning_rate": 0.00017073319218533688,
      "loss": 1.1445,
      "step": 18540
    },
    {
      "epoch": 0.7516055185267722,
      "grad_norm": 0.6381691694259644,
      "learning_rate": 0.00017070317183616029,
      "loss": 1.1212,
      "step": 18550
    },
    {
      "epoch": 0.7520106967038755,
      "grad_norm": 0.5905799269676208,
      "learning_rate": 0.00017067313874043584,
      "loss": 1.1328,
      "step": 18560
    },
    {
      "epoch": 0.7524158748809789,
      "grad_norm": 0.5980437994003296,
      "learning_rate": 0.00017064309290357798,
      "loss": 1.14,
      "step": 18570
    },
    {
      "epoch": 0.7528210530580823,
      "grad_norm": 0.5886944532394409,
      "learning_rate": 0.00017061303433100341,
      "loss": 1.1588,
      "step": 18580
    },
    {
      "epoch": 0.7532262312351857,
      "grad_norm": 0.48990440368652344,
      "learning_rate": 0.00017058296302813116,
      "loss": 1.1397,
      "step": 18590
    },
    {
      "epoch": 0.753631409412289,
      "grad_norm": 0.570651650428772,
      "learning_rate": 0.00017055287900038263,
      "loss": 1.1306,
      "step": 18600
    },
    {
      "epoch": 0.7540365875893924,
      "grad_norm": 0.6325370073318481,
      "learning_rate": 0.00017052278225318136,
      "loss": 1.1546,
      "step": 18610
    },
    {
      "epoch": 0.7544417657664958,
      "grad_norm": 0.5436761379241943,
      "learning_rate": 0.00017049267279195337,
      "loss": 1.1357,
      "step": 18620
    },
    {
      "epoch": 0.7548469439435992,
      "grad_norm": 0.5659282207489014,
      "learning_rate": 0.00017046255062212677,
      "loss": 1.1182,
      "step": 18630
    },
    {
      "epoch": 0.7552521221207026,
      "grad_norm": 0.529324471950531,
      "learning_rate": 0.00017043241574913207,
      "loss": 1.1473,
      "step": 18640
    },
    {
      "epoch": 0.755657300297806,
      "grad_norm": 0.569097101688385,
      "learning_rate": 0.00017040226817840215,
      "loss": 1.1093,
      "step": 18650
    },
    {
      "epoch": 0.7560624784749094,
      "grad_norm": 0.5917137265205383,
      "learning_rate": 0.000170372107915372,
      "loss": 1.111,
      "step": 18660
    },
    {
      "epoch": 0.7564676566520128,
      "grad_norm": 0.5262924432754517,
      "learning_rate": 0.00017034193496547902,
      "loss": 1.1451,
      "step": 18670
    },
    {
      "epoch": 0.7568728348291162,
      "grad_norm": 0.5324811935424805,
      "learning_rate": 0.00017031174933416286,
      "loss": 1.1329,
      "step": 18680
    },
    {
      "epoch": 0.7572780130062194,
      "grad_norm": 0.6013408303260803,
      "learning_rate": 0.0001702815510268655,
      "loss": 1.1485,
      "step": 18690
    },
    {
      "epoch": 0.7576831911833228,
      "grad_norm": 0.6034098863601685,
      "learning_rate": 0.00017025134004903115,
      "loss": 1.1077,
      "step": 18700
    },
    {
      "epoch": 0.7580883693604262,
      "grad_norm": 0.556441605091095,
      "learning_rate": 0.00017022111640610627,
      "loss": 1.1346,
      "step": 18710
    },
    {
      "epoch": 0.7584935475375296,
      "grad_norm": 0.5036845207214355,
      "learning_rate": 0.0001701908801035397,
      "loss": 1.134,
      "step": 18720
    },
    {
      "epoch": 0.758898725714633,
      "grad_norm": 0.5721425414085388,
      "learning_rate": 0.00017016063114678253,
      "loss": 1.1221,
      "step": 18730
    },
    {
      "epoch": 0.7593039038917364,
      "grad_norm": 0.5754898190498352,
      "learning_rate": 0.0001701303695412881,
      "loss": 1.1835,
      "step": 18740
    },
    {
      "epoch": 0.7597090820688398,
      "grad_norm": 0.6188938617706299,
      "learning_rate": 0.00017010009529251207,
      "loss": 1.1241,
      "step": 18750
    },
    {
      "epoch": 0.7601142602459432,
      "grad_norm": 0.5127008557319641,
      "learning_rate": 0.00017006980840591232,
      "loss": 1.0814,
      "step": 18760
    },
    {
      "epoch": 0.7605194384230466,
      "grad_norm": 0.5939504504203796,
      "learning_rate": 0.00017003950888694908,
      "loss": 1.1491,
      "step": 18770
    },
    {
      "epoch": 0.7609246166001499,
      "grad_norm": 0.5786957144737244,
      "learning_rate": 0.00017000919674108479,
      "loss": 1.1457,
      "step": 18780
    },
    {
      "epoch": 0.7613297947772533,
      "grad_norm": 0.5518621206283569,
      "learning_rate": 0.00016997887197378424,
      "loss": 1.1384,
      "step": 18790
    },
    {
      "epoch": 0.7617349729543567,
      "grad_norm": 0.5375494956970215,
      "learning_rate": 0.00016994853459051444,
      "loss": 1.1017,
      "step": 18800
    },
    {
      "epoch": 0.76214015113146,
      "grad_norm": 0.5246190428733826,
      "learning_rate": 0.00016991818459674468,
      "loss": 1.1014,
      "step": 18810
    },
    {
      "epoch": 0.7625453293085634,
      "grad_norm": 0.5041666030883789,
      "learning_rate": 0.00016988782199794654,
      "loss": 1.1311,
      "step": 18820
    },
    {
      "epoch": 0.7629505074856668,
      "grad_norm": 0.58709317445755,
      "learning_rate": 0.0001698574467995939,
      "loss": 1.1546,
      "step": 18830
    },
    {
      "epoch": 0.7633556856627702,
      "grad_norm": 0.5709360837936401,
      "learning_rate": 0.00016982705900716282,
      "loss": 1.134,
      "step": 18840
    },
    {
      "epoch": 0.7637608638398736,
      "grad_norm": 0.6101324558258057,
      "learning_rate": 0.00016979665862613174,
      "loss": 1.1394,
      "step": 18850
    },
    {
      "epoch": 0.764166042016977,
      "grad_norm": 0.568415105342865,
      "learning_rate": 0.00016976624566198125,
      "loss": 1.0884,
      "step": 18860
    },
    {
      "epoch": 0.7645712201940803,
      "grad_norm": 0.5195732116699219,
      "learning_rate": 0.00016973582012019437,
      "loss": 1.1907,
      "step": 18870
    },
    {
      "epoch": 0.7649763983711837,
      "grad_norm": 0.5646130442619324,
      "learning_rate": 0.00016970538200625622,
      "loss": 1.1665,
      "step": 18880
    },
    {
      "epoch": 0.7653815765482871,
      "grad_norm": 0.5648688673973083,
      "learning_rate": 0.0001696749313256543,
      "loss": 1.1361,
      "step": 18890
    },
    {
      "epoch": 0.7657867547253905,
      "grad_norm": 0.5282736420631409,
      "learning_rate": 0.00016964446808387832,
      "loss": 1.1893,
      "step": 18900
    },
    {
      "epoch": 0.7661919329024939,
      "grad_norm": 0.5346330404281616,
      "learning_rate": 0.00016961399228642029,
      "loss": 1.1303,
      "step": 18910
    },
    {
      "epoch": 0.7665971110795973,
      "grad_norm": 0.6565847396850586,
      "learning_rate": 0.00016958350393877442,
      "loss": 1.122,
      "step": 18920
    },
    {
      "epoch": 0.7670022892567007,
      "grad_norm": 0.5982806086540222,
      "learning_rate": 0.00016955300304643728,
      "loss": 1.1763,
      "step": 18930
    },
    {
      "epoch": 0.7674074674338041,
      "grad_norm": 0.5785622000694275,
      "learning_rate": 0.0001695224896149076,
      "loss": 1.1453,
      "step": 18940
    },
    {
      "epoch": 0.7678126456109075,
      "grad_norm": 0.5215151309967041,
      "learning_rate": 0.00016949196364968646,
      "loss": 1.1431,
      "step": 18950
    },
    {
      "epoch": 0.7682178237880107,
      "grad_norm": 0.6402701735496521,
      "learning_rate": 0.00016946142515627716,
      "loss": 1.1233,
      "step": 18960
    },
    {
      "epoch": 0.7686230019651141,
      "grad_norm": 0.5478402972221375,
      "learning_rate": 0.00016943087414018524,
      "loss": 1.1283,
      "step": 18970
    },
    {
      "epoch": 0.7690281801422175,
      "grad_norm": 0.5129750967025757,
      "learning_rate": 0.00016940031060691847,
      "loss": 1.1088,
      "step": 18980
    },
    {
      "epoch": 0.7694333583193209,
      "grad_norm": 0.5910369157791138,
      "learning_rate": 0.000169369734561987,
      "loss": 1.1491,
      "step": 18990
    },
    {
      "epoch": 0.7698385364964243,
      "grad_norm": 0.53847336769104,
      "learning_rate": 0.00016933914601090312,
      "loss": 1.1369,
      "step": 19000
    },
    {
      "epoch": 0.7702437146735277,
      "grad_norm": 0.5384936332702637,
      "learning_rate": 0.0001693085449591814,
      "loss": 1.1446,
      "step": 19010
    },
    {
      "epoch": 0.7706488928506311,
      "grad_norm": 0.5781065225601196,
      "learning_rate": 0.00016927793141233868,
      "loss": 1.1328,
      "step": 19020
    },
    {
      "epoch": 0.7710540710277345,
      "grad_norm": 0.5569975972175598,
      "learning_rate": 0.00016924730537589406,
      "loss": 1.1172,
      "step": 19030
    },
    {
      "epoch": 0.7714592492048379,
      "grad_norm": 0.6110308766365051,
      "learning_rate": 0.0001692166668553689,
      "loss": 1.1204,
      "step": 19040
    },
    {
      "epoch": 0.7718644273819412,
      "grad_norm": 0.5499427318572998,
      "learning_rate": 0.00016918601585628679,
      "loss": 1.1136,
      "step": 19050
    },
    {
      "epoch": 0.7722696055590446,
      "grad_norm": 0.5418890118598938,
      "learning_rate": 0.0001691553523841735,
      "loss": 1.1174,
      "step": 19060
    },
    {
      "epoch": 0.772674783736148,
      "grad_norm": 0.5093777775764465,
      "learning_rate": 0.00016912467644455718,
      "loss": 1.1538,
      "step": 19070
    },
    {
      "epoch": 0.7730799619132513,
      "grad_norm": 0.5407425165176392,
      "learning_rate": 0.00016909398804296818,
      "loss": 1.1377,
      "step": 19080
    },
    {
      "epoch": 0.7734851400903547,
      "grad_norm": 0.5451932549476624,
      "learning_rate": 0.00016906328718493906,
      "loss": 1.1048,
      "step": 19090
    },
    {
      "epoch": 0.7738903182674581,
      "grad_norm": 0.6095989942550659,
      "learning_rate": 0.0001690325738760046,
      "loss": 1.0943,
      "step": 19100
    },
    {
      "epoch": 0.7742954964445615,
      "grad_norm": 0.4982737898826599,
      "learning_rate": 0.000169001848121702,
      "loss": 1.1375,
      "step": 19110
    },
    {
      "epoch": 0.7747006746216649,
      "grad_norm": 0.5359206795692444,
      "learning_rate": 0.0001689711099275704,
      "loss": 1.142,
      "step": 19120
    },
    {
      "epoch": 0.7751058527987683,
      "grad_norm": 0.5409841537475586,
      "learning_rate": 0.0001689403592991515,
      "loss": 1.1303,
      "step": 19130
    },
    {
      "epoch": 0.7755110309758716,
      "grad_norm": 0.5889129638671875,
      "learning_rate": 0.00016890959624198907,
      "loss": 1.1284,
      "step": 19140
    },
    {
      "epoch": 0.775916209152975,
      "grad_norm": 0.5303646326065063,
      "learning_rate": 0.00016887882076162914,
      "loss": 1.0902,
      "step": 19150
    },
    {
      "epoch": 0.7763213873300784,
      "grad_norm": 0.6085774898529053,
      "learning_rate": 0.00016884803286362,
      "loss": 1.0958,
      "step": 19160
    },
    {
      "epoch": 0.7767265655071818,
      "grad_norm": 0.5450120568275452,
      "learning_rate": 0.00016881723255351213,
      "loss": 1.1347,
      "step": 19170
    },
    {
      "epoch": 0.7771317436842852,
      "grad_norm": 0.5737848281860352,
      "learning_rate": 0.00016878641983685837,
      "loss": 1.1368,
      "step": 19180
    },
    {
      "epoch": 0.7775369218613886,
      "grad_norm": 0.5636813640594482,
      "learning_rate": 0.00016875559471921366,
      "loss": 1.1319,
      "step": 19190
    },
    {
      "epoch": 0.777942100038492,
      "grad_norm": 0.5992211699485779,
      "learning_rate": 0.00016872475720613517,
      "loss": 1.114,
      "step": 19200
    },
    {
      "epoch": 0.7783472782155954,
      "grad_norm": 0.5719745755195618,
      "learning_rate": 0.00016869390730318254,
      "loss": 1.1114,
      "step": 19210
    },
    {
      "epoch": 0.7787524563926987,
      "grad_norm": 0.6093538999557495,
      "learning_rate": 0.00016866304501591728,
      "loss": 1.1424,
      "step": 19220
    },
    {
      "epoch": 0.779157634569802,
      "grad_norm": 0.6229391098022461,
      "learning_rate": 0.00016863217034990342,
      "loss": 1.1609,
      "step": 19230
    },
    {
      "epoch": 0.7795628127469054,
      "grad_norm": 0.5779278874397278,
      "learning_rate": 0.00016860128331070714,
      "loss": 1.1515,
      "step": 19240
    },
    {
      "epoch": 0.7799679909240088,
      "grad_norm": 0.5748093724250793,
      "learning_rate": 0.00016857038390389678,
      "loss": 1.138,
      "step": 19250
    },
    {
      "epoch": 0.7803731691011122,
      "grad_norm": 0.5202934741973877,
      "learning_rate": 0.00016853947213504297,
      "loss": 1.0942,
      "step": 19260
    },
    {
      "epoch": 0.7807783472782156,
      "grad_norm": 0.6491340398788452,
      "learning_rate": 0.0001685085480097186,
      "loss": 1.1061,
      "step": 19270
    },
    {
      "epoch": 0.781183525455319,
      "grad_norm": 0.5768429636955261,
      "learning_rate": 0.00016847761153349872,
      "loss": 1.1282,
      "step": 19280
    },
    {
      "epoch": 0.7815887036324224,
      "grad_norm": 0.5525264143943787,
      "learning_rate": 0.00016844666271196066,
      "loss": 1.1374,
      "step": 19290
    },
    {
      "epoch": 0.7819938818095258,
      "grad_norm": 0.6252078413963318,
      "learning_rate": 0.0001684157015506839,
      "loss": 1.1169,
      "step": 19300
    },
    {
      "epoch": 0.7823990599866292,
      "grad_norm": 0.557466447353363,
      "learning_rate": 0.00016838472805525028,
      "loss": 1.1445,
      "step": 19310
    },
    {
      "epoch": 0.7828042381637325,
      "grad_norm": 0.5384408831596375,
      "learning_rate": 0.0001683537422312437,
      "loss": 1.1259,
      "step": 19320
    },
    {
      "epoch": 0.7832094163408359,
      "grad_norm": 0.5395480394363403,
      "learning_rate": 0.0001683227440842504,
      "loss": 1.1197,
      "step": 19330
    },
    {
      "epoch": 0.7836145945179392,
      "grad_norm": 0.5780560970306396,
      "learning_rate": 0.0001682917336198588,
      "loss": 1.1266,
      "step": 19340
    },
    {
      "epoch": 0.7840197726950426,
      "grad_norm": 0.5910555720329285,
      "learning_rate": 0.00016826071084365956,
      "loss": 1.1522,
      "step": 19350
    },
    {
      "epoch": 0.784424950872146,
      "grad_norm": 0.5599584579467773,
      "learning_rate": 0.00016822967576124556,
      "loss": 1.1559,
      "step": 19360
    },
    {
      "epoch": 0.7848301290492494,
      "grad_norm": 0.5346325635910034,
      "learning_rate": 0.00016819862837821181,
      "loss": 1.0914,
      "step": 19370
    },
    {
      "epoch": 0.7852353072263528,
      "grad_norm": 0.5599008202552795,
      "learning_rate": 0.0001681675687001557,
      "loss": 1.0898,
      "step": 19380
    },
    {
      "epoch": 0.7856404854034562,
      "grad_norm": 0.5999270677566528,
      "learning_rate": 0.00016813649673267668,
      "loss": 1.1271,
      "step": 19390
    },
    {
      "epoch": 0.7860456635805596,
      "grad_norm": 0.5360412001609802,
      "learning_rate": 0.0001681054124813765,
      "loss": 1.1735,
      "step": 19400
    },
    {
      "epoch": 0.7864508417576629,
      "grad_norm": 0.603269636631012,
      "learning_rate": 0.00016807431595185915,
      "loss": 1.0955,
      "step": 19410
    },
    {
      "epoch": 0.7868560199347663,
      "grad_norm": 0.5904859304428101,
      "learning_rate": 0.00016804320714973074,
      "loss": 1.1131,
      "step": 19420
    },
    {
      "epoch": 0.7872611981118697,
      "grad_norm": 0.5391505360603333,
      "learning_rate": 0.00016801208608059967,
      "loss": 1.1589,
      "step": 19430
    },
    {
      "epoch": 0.7876663762889731,
      "grad_norm": 0.5870198607444763,
      "learning_rate": 0.0001679809527500765,
      "loss": 1.1117,
      "step": 19440
    },
    {
      "epoch": 0.7880715544660765,
      "grad_norm": 0.6477810144424438,
      "learning_rate": 0.00016794980716377405,
      "loss": 1.1592,
      "step": 19450
    },
    {
      "epoch": 0.7884767326431799,
      "grad_norm": 0.5691637396812439,
      "learning_rate": 0.00016791864932730734,
      "loss": 1.1249,
      "step": 19460
    },
    {
      "epoch": 0.7888819108202832,
      "grad_norm": 0.6955838203430176,
      "learning_rate": 0.00016788747924629358,
      "loss": 1.1327,
      "step": 19470
    },
    {
      "epoch": 0.7892870889973866,
      "grad_norm": 0.6122684478759766,
      "learning_rate": 0.00016785629692635213,
      "loss": 1.1588,
      "step": 19480
    },
    {
      "epoch": 0.78969226717449,
      "grad_norm": 0.5720804333686829,
      "learning_rate": 0.0001678251023731047,
      "loss": 1.1531,
      "step": 19490
    },
    {
      "epoch": 0.7900974453515933,
      "grad_norm": 0.551045835018158,
      "learning_rate": 0.00016779389559217506,
      "loss": 1.1677,
      "step": 19500
    },
    {
      "epoch": 0.7905026235286967,
      "grad_norm": 0.5615679025650024,
      "learning_rate": 0.00016776267658918928,
      "loss": 1.1292,
      "step": 19510
    },
    {
      "epoch": 0.7909078017058001,
      "grad_norm": 0.6063175797462463,
      "learning_rate": 0.0001677314453697756,
      "loss": 1.1149,
      "step": 19520
    },
    {
      "epoch": 0.7913129798829035,
      "grad_norm": 0.5547340512275696,
      "learning_rate": 0.00016770020193956447,
      "loss": 1.1296,
      "step": 19530
    },
    {
      "epoch": 0.7917181580600069,
      "grad_norm": 0.5744422078132629,
      "learning_rate": 0.00016766894630418851,
      "loss": 1.1345,
      "step": 19540
    },
    {
      "epoch": 0.7921233362371103,
      "grad_norm": 0.5790544748306274,
      "learning_rate": 0.00016763767846928262,
      "loss": 1.1229,
      "step": 19550
    },
    {
      "epoch": 0.7925285144142137,
      "grad_norm": 0.5676174759864807,
      "learning_rate": 0.00016760639844048378,
      "loss": 1.164,
      "step": 19560
    },
    {
      "epoch": 0.7929336925913171,
      "grad_norm": 0.5591793060302734,
      "learning_rate": 0.00016757510622343125,
      "loss": 1.1502,
      "step": 19570
    },
    {
      "epoch": 0.7933388707684205,
      "grad_norm": 0.5487279295921326,
      "learning_rate": 0.0001675438018237665,
      "loss": 1.1121,
      "step": 19580
    },
    {
      "epoch": 0.7937440489455238,
      "grad_norm": 0.5827490091323853,
      "learning_rate": 0.00016751248524713312,
      "loss": 1.1824,
      "step": 19590
    },
    {
      "epoch": 0.7941492271226271,
      "grad_norm": 0.5725349187850952,
      "learning_rate": 0.00016748115649917698,
      "loss": 1.1395,
      "step": 19600
    },
    {
      "epoch": 0.7945544052997305,
      "grad_norm": 0.538130521774292,
      "learning_rate": 0.00016744981558554614,
      "loss": 1.1148,
      "step": 19610
    },
    {
      "epoch": 0.7949595834768339,
      "grad_norm": 0.5544222593307495,
      "learning_rate": 0.0001674184625118907,
      "loss": 1.1247,
      "step": 19620
    },
    {
      "epoch": 0.7953647616539373,
      "grad_norm": 0.5400269627571106,
      "learning_rate": 0.0001673870972838632,
      "loss": 1.1495,
      "step": 19630
    },
    {
      "epoch": 0.7957699398310407,
      "grad_norm": 0.6104006767272949,
      "learning_rate": 0.00016735571990711818,
      "loss": 1.1409,
      "step": 19640
    },
    {
      "epoch": 0.7961751180081441,
      "grad_norm": 0.5261937975883484,
      "learning_rate": 0.00016732433038731242,
      "loss": 1.131,
      "step": 19650
    },
    {
      "epoch": 0.7965802961852475,
      "grad_norm": 0.6801859140396118,
      "learning_rate": 0.00016729292873010496,
      "loss": 1.1396,
      "step": 19660
    },
    {
      "epoch": 0.7969854743623509,
      "grad_norm": 0.5963397026062012,
      "learning_rate": 0.00016726151494115695,
      "loss": 1.1015,
      "step": 19670
    },
    {
      "epoch": 0.7973906525394542,
      "grad_norm": 0.549649715423584,
      "learning_rate": 0.0001672300890261317,
      "loss": 1.1325,
      "step": 19680
    },
    {
      "epoch": 0.7977958307165576,
      "grad_norm": 0.5100482702255249,
      "learning_rate": 0.00016719865099069478,
      "loss": 1.1287,
      "step": 19690
    },
    {
      "epoch": 0.798201008893661,
      "grad_norm": 0.5495323538780212,
      "learning_rate": 0.00016716720084051397,
      "loss": 1.086,
      "step": 19700
    },
    {
      "epoch": 0.7986061870707644,
      "grad_norm": 0.5839194655418396,
      "learning_rate": 0.00016713573858125912,
      "loss": 1.116,
      "step": 19710
    },
    {
      "epoch": 0.7990113652478678,
      "grad_norm": 0.59615159034729,
      "learning_rate": 0.00016710426421860235,
      "loss": 1.1414,
      "step": 19720
    },
    {
      "epoch": 0.7994165434249711,
      "grad_norm": 0.6060029864311218,
      "learning_rate": 0.00016707277775821793,
      "loss": 1.1403,
      "step": 19730
    },
    {
      "epoch": 0.7998217216020745,
      "grad_norm": 0.5563827753067017,
      "learning_rate": 0.00016704127920578232,
      "loss": 1.1413,
      "step": 19740
    },
    {
      "epoch": 0.8002268997791779,
      "grad_norm": 0.6217523813247681,
      "learning_rate": 0.00016700976856697417,
      "loss": 1.1346,
      "step": 19750
    },
    {
      "epoch": 0.8006320779562813,
      "grad_norm": 0.6177238821983337,
      "learning_rate": 0.00016697824584747427,
      "loss": 1.1653,
      "step": 19760
    },
    {
      "epoch": 0.8010372561333846,
      "grad_norm": 0.5838682055473328,
      "learning_rate": 0.00016694671105296568,
      "loss": 1.1127,
      "step": 19770
    },
    {
      "epoch": 0.801442434310488,
      "grad_norm": 0.5305541753768921,
      "learning_rate": 0.00016691516418913352,
      "loss": 1.1078,
      "step": 19780
    },
    {
      "epoch": 0.8018476124875914,
      "grad_norm": 0.5796138048171997,
      "learning_rate": 0.00016688360526166514,
      "loss": 1.1232,
      "step": 19790
    },
    {
      "epoch": 0.8022527906646948,
      "grad_norm": 0.5393558144569397,
      "learning_rate": 0.0001668520342762501,
      "loss": 1.1518,
      "step": 19800
    },
    {
      "epoch": 0.8026579688417982,
      "grad_norm": 0.5799124240875244,
      "learning_rate": 0.00016682045123858006,
      "loss": 1.1469,
      "step": 19810
    },
    {
      "epoch": 0.8030631470189016,
      "grad_norm": 0.6174018383026123,
      "learning_rate": 0.00016678885615434888,
      "loss": 1.1559,
      "step": 19820
    },
    {
      "epoch": 0.803468325196005,
      "grad_norm": 0.540070116519928,
      "learning_rate": 0.00016675724902925264,
      "loss": 1.1422,
      "step": 19830
    },
    {
      "epoch": 0.8038735033731084,
      "grad_norm": 0.5456597805023193,
      "learning_rate": 0.00016672562986898957,
      "loss": 1.1145,
      "step": 19840
    },
    {
      "epoch": 0.8042786815502118,
      "grad_norm": 0.5183867812156677,
      "learning_rate": 0.00016669399867925996,
      "loss": 1.1258,
      "step": 19850
    },
    {
      "epoch": 0.804683859727315,
      "grad_norm": 0.5448383688926697,
      "learning_rate": 0.00016666235546576648,
      "loss": 1.1556,
      "step": 19860
    },
    {
      "epoch": 0.8050890379044184,
      "grad_norm": 0.7745888233184814,
      "learning_rate": 0.00016663070023421373,
      "loss": 1.1701,
      "step": 19870
    },
    {
      "epoch": 0.8054942160815218,
      "grad_norm": 0.5240235924720764,
      "learning_rate": 0.0001665990329903087,
      "loss": 1.1386,
      "step": 19880
    },
    {
      "epoch": 0.8058993942586252,
      "grad_norm": 0.7149744033813477,
      "learning_rate": 0.0001665673537397604,
      "loss": 1.1368,
      "step": 19890
    },
    {
      "epoch": 0.8063045724357286,
      "grad_norm": 0.5522542595863342,
      "learning_rate": 0.00016653566248828002,
      "loss": 1.1354,
      "step": 19900
    },
    {
      "epoch": 0.806709750612832,
      "grad_norm": 0.6175497174263,
      "learning_rate": 0.00016650395924158097,
      "loss": 1.1063,
      "step": 19910
    },
    {
      "epoch": 0.8071149287899354,
      "grad_norm": 0.5544633865356445,
      "learning_rate": 0.0001664722440053788,
      "loss": 1.1159,
      "step": 19920
    },
    {
      "epoch": 0.8075201069670388,
      "grad_norm": 0.542316734790802,
      "learning_rate": 0.0001664405167853912,
      "loss": 1.1457,
      "step": 19930
    },
    {
      "epoch": 0.8079252851441422,
      "grad_norm": 0.5749243497848511,
      "learning_rate": 0.00016640877758733802,
      "loss": 1.1188,
      "step": 19940
    },
    {
      "epoch": 0.8083304633212455,
      "grad_norm": 0.6128023862838745,
      "learning_rate": 0.0001663770264169413,
      "loss": 1.0955,
      "step": 19950
    },
    {
      "epoch": 0.8087356414983489,
      "grad_norm": 0.6449238061904907,
      "learning_rate": 0.00016634526327992524,
      "loss": 1.1285,
      "step": 19960
    },
    {
      "epoch": 0.8091408196754523,
      "grad_norm": 0.6129885911941528,
      "learning_rate": 0.00016631348818201612,
      "loss": 1.1215,
      "step": 19970
    },
    {
      "epoch": 0.8095459978525557,
      "grad_norm": 0.5799210071563721,
      "learning_rate": 0.00016628170112894253,
      "loss": 1.1067,
      "step": 19980
    },
    {
      "epoch": 0.809951176029659,
      "grad_norm": 0.5383392572402954,
      "learning_rate": 0.000166249902126435,
      "loss": 1.1142,
      "step": 19990
    },
    {
      "epoch": 0.8103563542067624,
      "grad_norm": 0.5351027250289917,
      "learning_rate": 0.00016621809118022647,
      "loss": 1.1437,
      "step": 20000
    },
    {
      "epoch": 0.8107615323838658,
      "grad_norm": 0.57749342918396,
      "learning_rate": 0.00016618626829605177,
      "loss": 1.1159,
      "step": 20010
    },
    {
      "epoch": 0.8111667105609692,
      "grad_norm": 0.737376868724823,
      "learning_rate": 0.0001661544334796481,
      "loss": 1.134,
      "step": 20020
    },
    {
      "epoch": 0.8115718887380726,
      "grad_norm": 0.6397894024848938,
      "learning_rate": 0.00016612258673675466,
      "loss": 1.1513,
      "step": 20030
    },
    {
      "epoch": 0.8119770669151759,
      "grad_norm": 0.49480140209198,
      "learning_rate": 0.0001660907280731129,
      "loss": 1.1434,
      "step": 20040
    },
    {
      "epoch": 0.8123822450922793,
      "grad_norm": 0.5297718644142151,
      "learning_rate": 0.0001660588574944664,
      "loss": 1.1233,
      "step": 20050
    },
    {
      "epoch": 0.8127874232693827,
      "grad_norm": 0.5480042695999146,
      "learning_rate": 0.0001660269750065608,
      "loss": 1.1219,
      "step": 20060
    },
    {
      "epoch": 0.8131926014464861,
      "grad_norm": 0.5497350692749023,
      "learning_rate": 0.00016599508061514404,
      "loss": 1.1603,
      "step": 20070
    },
    {
      "epoch": 0.8135977796235895,
      "grad_norm": 0.5655298233032227,
      "learning_rate": 0.00016596317432596605,
      "loss": 1.0808,
      "step": 20080
    },
    {
      "epoch": 0.8140029578006929,
      "grad_norm": 0.6081784963607788,
      "learning_rate": 0.00016593125614477905,
      "loss": 1.1392,
      "step": 20090
    },
    {
      "epoch": 0.8144081359777963,
      "grad_norm": 0.5941429138183594,
      "learning_rate": 0.00016589932607733726,
      "loss": 1.1537,
      "step": 20100
    },
    {
      "epoch": 0.8148133141548997,
      "grad_norm": 0.6162896156311035,
      "learning_rate": 0.00016586738412939712,
      "loss": 1.0942,
      "step": 20110
    },
    {
      "epoch": 0.815218492332003,
      "grad_norm": 0.5863626599311829,
      "learning_rate": 0.00016583543030671726,
      "loss": 1.0834,
      "step": 20120
    },
    {
      "epoch": 0.8156236705091063,
      "grad_norm": 0.5707448720932007,
      "learning_rate": 0.00016580346461505835,
      "loss": 1.1193,
      "step": 20130
    },
    {
      "epoch": 0.8160288486862097,
      "grad_norm": 0.5718323588371277,
      "learning_rate": 0.00016577148706018328,
      "loss": 1.1214,
      "step": 20140
    },
    {
      "epoch": 0.8164340268633131,
      "grad_norm": 0.58971107006073,
      "learning_rate": 0.000165739497647857,
      "loss": 1.1368,
      "step": 20150
    },
    {
      "epoch": 0.8168392050404165,
      "grad_norm": 0.5696049332618713,
      "learning_rate": 0.0001657074963838467,
      "loss": 1.1453,
      "step": 20160
    },
    {
      "epoch": 0.8172443832175199,
      "grad_norm": 0.6612564325332642,
      "learning_rate": 0.00016567548327392162,
      "loss": 1.1168,
      "step": 20170
    },
    {
      "epoch": 0.8176495613946233,
      "grad_norm": 0.5644662976264954,
      "learning_rate": 0.00016564345832385314,
      "loss": 1.1609,
      "step": 20180
    },
    {
      "epoch": 0.8180547395717267,
      "grad_norm": 0.5685643553733826,
      "learning_rate": 0.0001656114215394148,
      "loss": 1.1424,
      "step": 20190
    },
    {
      "epoch": 0.8184599177488301,
      "grad_norm": 0.5563279986381531,
      "learning_rate": 0.00016557937292638233,
      "loss": 1.1315,
      "step": 20200
    },
    {
      "epoch": 0.8188650959259335,
      "grad_norm": 0.6096606850624084,
      "learning_rate": 0.0001655473124905335,
      "loss": 1.0949,
      "step": 20210
    },
    {
      "epoch": 0.8192702741030368,
      "grad_norm": 0.5642665028572083,
      "learning_rate": 0.00016551524023764825,
      "loss": 1.151,
      "step": 20220
    },
    {
      "epoch": 0.8196754522801402,
      "grad_norm": 0.539894700050354,
      "learning_rate": 0.00016548315617350863,
      "loss": 1.1438,
      "step": 20230
    },
    {
      "epoch": 0.8200806304572436,
      "grad_norm": 0.5555012226104736,
      "learning_rate": 0.00016545106030389886,
      "loss": 1.1189,
      "step": 20240
    },
    {
      "epoch": 0.820485808634347,
      "grad_norm": 0.6500780582427979,
      "learning_rate": 0.00016541895263460528,
      "loss": 1.1161,
      "step": 20250
    },
    {
      "epoch": 0.8208909868114503,
      "grad_norm": 0.4794643819332123,
      "learning_rate": 0.0001653868331714163,
      "loss": 1.1322,
      "step": 20260
    },
    {
      "epoch": 0.8212961649885537,
      "grad_norm": 0.5552313923835754,
      "learning_rate": 0.00016535470192012252,
      "loss": 1.1256,
      "step": 20270
    },
    {
      "epoch": 0.8217013431656571,
      "grad_norm": 0.5281320810317993,
      "learning_rate": 0.00016532255888651666,
      "loss": 1.0918,
      "step": 20280
    },
    {
      "epoch": 0.8221065213427605,
      "grad_norm": 0.590743362903595,
      "learning_rate": 0.00016529040407639348,
      "loss": 1.1497,
      "step": 20290
    },
    {
      "epoch": 0.8225116995198639,
      "grad_norm": 0.5917896628379822,
      "learning_rate": 0.00016525823749555002,
      "loss": 1.1403,
      "step": 20300
    },
    {
      "epoch": 0.8229168776969672,
      "grad_norm": 0.5250363349914551,
      "learning_rate": 0.0001652260591497853,
      "loss": 1.1256,
      "step": 20310
    },
    {
      "epoch": 0.8233220558740706,
      "grad_norm": 0.6358888745307922,
      "learning_rate": 0.00016519386904490055,
      "loss": 1.1313,
      "step": 20320
    },
    {
      "epoch": 0.823727234051174,
      "grad_norm": 0.6302996277809143,
      "learning_rate": 0.00016516166718669905,
      "loss": 1.1152,
      "step": 20330
    },
    {
      "epoch": 0.8241324122282774,
      "grad_norm": 0.5789265632629395,
      "learning_rate": 0.00016512945358098627,
      "loss": 1.1649,
      "step": 20340
    },
    {
      "epoch": 0.8245375904053808,
      "grad_norm": 0.5040349364280701,
      "learning_rate": 0.0001650972282335697,
      "loss": 1.0755,
      "step": 20350
    },
    {
      "epoch": 0.8249427685824842,
      "grad_norm": 0.5588706135749817,
      "learning_rate": 0.00016506499115025907,
      "loss": 1.1216,
      "step": 20360
    },
    {
      "epoch": 0.8253479467595876,
      "grad_norm": 0.5400094985961914,
      "learning_rate": 0.00016503274233686612,
      "loss": 1.0912,
      "step": 20370
    },
    {
      "epoch": 0.825753124936691,
      "grad_norm": 0.5466423630714417,
      "learning_rate": 0.00016500048179920475,
      "loss": 1.1339,
      "step": 20380
    },
    {
      "epoch": 0.8261583031137943,
      "grad_norm": 0.5474914908409119,
      "learning_rate": 0.00016496820954309103,
      "loss": 1.1093,
      "step": 20390
    },
    {
      "epoch": 0.8265634812908976,
      "grad_norm": 0.5660754442214966,
      "learning_rate": 0.00016493592557434303,
      "loss": 1.123,
      "step": 20400
    },
    {
      "epoch": 0.826968659468001,
      "grad_norm": 0.6118622422218323,
      "learning_rate": 0.000164903629898781,
      "loss": 1.0946,
      "step": 20410
    },
    {
      "epoch": 0.8273738376451044,
      "grad_norm": 0.6386812925338745,
      "learning_rate": 0.00016487132252222727,
      "loss": 1.1191,
      "step": 20420
    },
    {
      "epoch": 0.8277790158222078,
      "grad_norm": 0.5753735303878784,
      "learning_rate": 0.00016483900345050633,
      "loss": 1.1298,
      "step": 20430
    },
    {
      "epoch": 0.8281841939993112,
      "grad_norm": 0.6585415601730347,
      "learning_rate": 0.0001648066726894447,
      "loss": 1.1362,
      "step": 20440
    },
    {
      "epoch": 0.8285893721764146,
      "grad_norm": 0.58055180311203,
      "learning_rate": 0.0001647743302448711,
      "loss": 1.1064,
      "step": 20450
    },
    {
      "epoch": 0.828994550353518,
      "grad_norm": 0.5101873874664307,
      "learning_rate": 0.0001647419761226163,
      "loss": 1.1301,
      "step": 20460
    },
    {
      "epoch": 0.8293997285306214,
      "grad_norm": 0.5896674394607544,
      "learning_rate": 0.0001647096103285131,
      "loss": 1.1442,
      "step": 20470
    },
    {
      "epoch": 0.8298049067077248,
      "grad_norm": 0.5313712358474731,
      "learning_rate": 0.0001646772328683966,
      "loss": 1.1368,
      "step": 20480
    },
    {
      "epoch": 0.8302100848848281,
      "grad_norm": 0.5883998870849609,
      "learning_rate": 0.0001646448437481039,
      "loss": 1.1605,
      "step": 20490
    },
    {
      "epoch": 0.8306152630619315,
      "grad_norm": 0.5703822374343872,
      "learning_rate": 0.00016461244297347408,
      "loss": 1.1805,
      "step": 20500
    },
    {
      "epoch": 0.8310204412390348,
      "grad_norm": 0.6017342805862427,
      "learning_rate": 0.00016458003055034852,
      "loss": 1.1167,
      "step": 20510
    },
    {
      "epoch": 0.8314256194161382,
      "grad_norm": 0.6095072627067566,
      "learning_rate": 0.00016454760648457057,
      "loss": 1.1267,
      "step": 20520
    },
    {
      "epoch": 0.8318307975932416,
      "grad_norm": 0.5859165787696838,
      "learning_rate": 0.00016451517078198578,
      "loss": 1.1376,
      "step": 20530
    },
    {
      "epoch": 0.832235975770345,
      "grad_norm": 0.6198223233222961,
      "learning_rate": 0.00016448272344844167,
      "loss": 1.1411,
      "step": 20540
    },
    {
      "epoch": 0.8326411539474484,
      "grad_norm": 0.5363257527351379,
      "learning_rate": 0.00016445026448978797,
      "loss": 1.1269,
      "step": 20550
    },
    {
      "epoch": 0.8330463321245518,
      "grad_norm": 0.5410638451576233,
      "learning_rate": 0.00016441779391187646,
      "loss": 1.1074,
      "step": 20560
    },
    {
      "epoch": 0.8334515103016552,
      "grad_norm": 0.5842058062553406,
      "learning_rate": 0.00016438531172056103,
      "loss": 1.13,
      "step": 20570
    },
    {
      "epoch": 0.8338566884787585,
      "grad_norm": 0.5461860299110413,
      "learning_rate": 0.00016435281792169762,
      "loss": 1.1448,
      "step": 20580
    },
    {
      "epoch": 0.8342618666558619,
      "grad_norm": 0.5332997441291809,
      "learning_rate": 0.0001643203125211443,
      "loss": 1.1455,
      "step": 20590
    },
    {
      "epoch": 0.8346670448329653,
      "grad_norm": 0.7184667587280273,
      "learning_rate": 0.00016428779552476127,
      "loss": 1.1807,
      "step": 20600
    },
    {
      "epoch": 0.8350722230100687,
      "grad_norm": 0.5381059646606445,
      "learning_rate": 0.00016425526693841073,
      "loss": 1.1354,
      "step": 20610
    },
    {
      "epoch": 0.8354774011871721,
      "grad_norm": 0.5337061285972595,
      "learning_rate": 0.00016422272676795701,
      "loss": 1.1703,
      "step": 20620
    },
    {
      "epoch": 0.8358825793642755,
      "grad_norm": 0.6402966380119324,
      "learning_rate": 0.00016419017501926656,
      "loss": 1.1483,
      "step": 20630
    },
    {
      "epoch": 0.8362877575413789,
      "grad_norm": 0.589629054069519,
      "learning_rate": 0.00016415761169820788,
      "loss": 1.1524,
      "step": 20640
    },
    {
      "epoch": 0.8366929357184822,
      "grad_norm": 0.536746621131897,
      "learning_rate": 0.00016412503681065163,
      "loss": 1.1463,
      "step": 20650
    },
    {
      "epoch": 0.8370981138955856,
      "grad_norm": 0.5475829839706421,
      "learning_rate": 0.00016409245036247042,
      "loss": 1.1576,
      "step": 20660
    },
    {
      "epoch": 0.8375032920726889,
      "grad_norm": 0.6952071189880371,
      "learning_rate": 0.000164059852359539,
      "loss": 1.117,
      "step": 20670
    },
    {
      "epoch": 0.8379084702497923,
      "grad_norm": 0.602559506893158,
      "learning_rate": 0.0001640272428077343,
      "loss": 1.1189,
      "step": 20680
    },
    {
      "epoch": 0.8383136484268957,
      "grad_norm": 0.5610623359680176,
      "learning_rate": 0.0001639946217129352,
      "loss": 1.1187,
      "step": 20690
    },
    {
      "epoch": 0.8387188266039991,
      "grad_norm": 0.6069028973579407,
      "learning_rate": 0.00016396198908102272,
      "loss": 1.1398,
      "step": 20700
    },
    {
      "epoch": 0.8391240047811025,
      "grad_norm": 0.5802363753318787,
      "learning_rate": 0.00016392934491787995,
      "loss": 1.1343,
      "step": 20710
    },
    {
      "epoch": 0.8395291829582059,
      "grad_norm": 0.4993867576122284,
      "learning_rate": 0.00016389668922939208,
      "loss": 1.1352,
      "step": 20720
    },
    {
      "epoch": 0.8399343611353093,
      "grad_norm": 0.609311044216156,
      "learning_rate": 0.00016386402202144635,
      "loss": 1.1459,
      "step": 20730
    },
    {
      "epoch": 0.8403395393124127,
      "grad_norm": 0.6170148253440857,
      "learning_rate": 0.0001638313432999321,
      "loss": 1.1355,
      "step": 20740
    },
    {
      "epoch": 0.8407447174895161,
      "grad_norm": 0.5945557355880737,
      "learning_rate": 0.0001637986530707407,
      "loss": 1.0898,
      "step": 20750
    },
    {
      "epoch": 0.8411498956666194,
      "grad_norm": 0.6043921113014221,
      "learning_rate": 0.00016376595133976571,
      "loss": 1.1224,
      "step": 20760
    },
    {
      "epoch": 0.8415550738437227,
      "grad_norm": 0.48154595494270325,
      "learning_rate": 0.0001637332381129026,
      "loss": 1.1368,
      "step": 20770
    },
    {
      "epoch": 0.8419602520208261,
      "grad_norm": 0.5165026187896729,
      "learning_rate": 0.00016370051339604898,
      "loss": 1.1318,
      "step": 20780
    },
    {
      "epoch": 0.8423654301979295,
      "grad_norm": 0.5780938863754272,
      "learning_rate": 0.00016366777719510462,
      "loss": 1.1173,
      "step": 20790
    },
    {
      "epoch": 0.8427706083750329,
      "grad_norm": 0.5239212512969971,
      "learning_rate": 0.00016363502951597124,
      "loss": 1.1123,
      "step": 20800
    },
    {
      "epoch": 0.8431757865521363,
      "grad_norm": 0.5353193879127502,
      "learning_rate": 0.00016360227036455266,
      "loss": 1.1007,
      "step": 20810
    },
    {
      "epoch": 0.8435809647292397,
      "grad_norm": 0.6468400955200195,
      "learning_rate": 0.00016356949974675485,
      "loss": 1.1265,
      "step": 20820
    },
    {
      "epoch": 0.8439861429063431,
      "grad_norm": 0.5773003101348877,
      "learning_rate": 0.00016353671766848572,
      "loss": 1.1217,
      "step": 20830
    },
    {
      "epoch": 0.8443913210834465,
      "grad_norm": 0.5425311326980591,
      "learning_rate": 0.0001635039241356553,
      "loss": 1.098,
      "step": 20840
    },
    {
      "epoch": 0.8447964992605498,
      "grad_norm": 0.6064731478691101,
      "learning_rate": 0.00016347111915417575,
      "loss": 1.1278,
      "step": 20850
    },
    {
      "epoch": 0.8452016774376532,
      "grad_norm": 0.5243592858314514,
      "learning_rate": 0.00016343830272996118,
      "loss": 1.1142,
      "step": 20860
    },
    {
      "epoch": 0.8456068556147566,
      "grad_norm": 0.6339744925498962,
      "learning_rate": 0.0001634054748689278,
      "loss": 1.1202,
      "step": 20870
    },
    {
      "epoch": 0.84601203379186,
      "grad_norm": 0.6001390814781189,
      "learning_rate": 0.000163372635576994,
      "loss": 1.1135,
      "step": 20880
    },
    {
      "epoch": 0.8464172119689634,
      "grad_norm": 0.5434259176254272,
      "learning_rate": 0.00016333978486008003,
      "loss": 1.1328,
      "step": 20890
    },
    {
      "epoch": 0.8468223901460668,
      "grad_norm": 0.5913121104240417,
      "learning_rate": 0.00016330692272410838,
      "loss": 1.1897,
      "step": 20900
    },
    {
      "epoch": 0.8472275683231701,
      "grad_norm": 0.6617525815963745,
      "learning_rate": 0.00016327404917500346,
      "loss": 1.1583,
      "step": 20910
    },
    {
      "epoch": 0.8476327465002735,
      "grad_norm": 0.6068620085716248,
      "learning_rate": 0.00016324116421869182,
      "loss": 1.1118,
      "step": 20920
    },
    {
      "epoch": 0.8480379246773769,
      "grad_norm": 0.646988570690155,
      "learning_rate": 0.00016320826786110203,
      "loss": 1.1522,
      "step": 20930
    },
    {
      "epoch": 0.8484431028544802,
      "grad_norm": 0.5944828987121582,
      "learning_rate": 0.0001631753601081648,
      "loss": 1.1445,
      "step": 20940
    },
    {
      "epoch": 0.8488482810315836,
      "grad_norm": 0.5717775225639343,
      "learning_rate": 0.00016314244096581268,
      "loss": 1.1345,
      "step": 20950
    },
    {
      "epoch": 0.849253459208687,
      "grad_norm": 0.5712586045265198,
      "learning_rate": 0.00016310951043998055,
      "loss": 1.1399,
      "step": 20960
    },
    {
      "epoch": 0.8496586373857904,
      "grad_norm": 0.5628635883331299,
      "learning_rate": 0.00016307656853660515,
      "loss": 1.1332,
      "step": 20970
    },
    {
      "epoch": 0.8500638155628938,
      "grad_norm": 0.5638338327407837,
      "learning_rate": 0.00016304361526162534,
      "loss": 1.1727,
      "step": 20980
    },
    {
      "epoch": 0.8504689937399972,
      "grad_norm": 0.5468073487281799,
      "learning_rate": 0.00016301065062098203,
      "loss": 1.1511,
      "step": 20990
    },
    {
      "epoch": 0.8508741719171006,
      "grad_norm": 0.5849620699882507,
      "learning_rate": 0.00016297767462061815,
      "loss": 1.1461,
      "step": 21000
    },
    {
      "epoch": 0.851279350094204,
      "grad_norm": 0.6264720559120178,
      "learning_rate": 0.0001629446872664787,
      "loss": 1.1378,
      "step": 21010
    },
    {
      "epoch": 0.8516845282713074,
      "grad_norm": 0.48682868480682373,
      "learning_rate": 0.00016291168856451073,
      "loss": 1.1221,
      "step": 21020
    },
    {
      "epoch": 0.8520897064484106,
      "grad_norm": 0.564090371131897,
      "learning_rate": 0.00016287867852066333,
      "loss": 1.1242,
      "step": 21030
    },
    {
      "epoch": 0.852494884625514,
      "grad_norm": 0.5451701879501343,
      "learning_rate": 0.00016284565714088766,
      "loss": 1.1284,
      "step": 21040
    },
    {
      "epoch": 0.8529000628026174,
      "grad_norm": 0.6579101085662842,
      "learning_rate": 0.0001628126244311369,
      "loss": 1.1306,
      "step": 21050
    },
    {
      "epoch": 0.8533052409797208,
      "grad_norm": 0.5571081042289734,
      "learning_rate": 0.00016277958039736618,
      "loss": 1.112,
      "step": 21060
    },
    {
      "epoch": 0.8537104191568242,
      "grad_norm": 0.5878406167030334,
      "learning_rate": 0.00016274652504553287,
      "loss": 1.1433,
      "step": 21070
    },
    {
      "epoch": 0.8541155973339276,
      "grad_norm": 0.6200227737426758,
      "learning_rate": 0.00016271345838159623,
      "loss": 1.1286,
      "step": 21080
    },
    {
      "epoch": 0.854520775511031,
      "grad_norm": 0.6276731491088867,
      "learning_rate": 0.00016268038041151762,
      "loss": 1.1697,
      "step": 21090
    },
    {
      "epoch": 0.8549259536881344,
      "grad_norm": 0.5544545650482178,
      "learning_rate": 0.00016264729114126035,
      "loss": 1.0987,
      "step": 21100
    },
    {
      "epoch": 0.8553311318652378,
      "grad_norm": 0.5557426810264587,
      "learning_rate": 0.00016261419057678994,
      "loss": 1.1238,
      "step": 21110
    },
    {
      "epoch": 0.8557363100423411,
      "grad_norm": 0.6090567111968994,
      "learning_rate": 0.00016258107872407375,
      "loss": 1.1333,
      "step": 21120
    },
    {
      "epoch": 0.8561414882194445,
      "grad_norm": 0.5773491859436035,
      "learning_rate": 0.0001625479555890814,
      "loss": 1.0837,
      "step": 21130
    },
    {
      "epoch": 0.8565466663965479,
      "grad_norm": 0.747291088104248,
      "learning_rate": 0.00016251482117778421,
      "loss": 1.1481,
      "step": 21140
    },
    {
      "epoch": 0.8569518445736513,
      "grad_norm": 0.5303491353988647,
      "learning_rate": 0.00016248167549615596,
      "loss": 1.1164,
      "step": 21150
    },
    {
      "epoch": 0.8573570227507546,
      "grad_norm": 0.592390775680542,
      "learning_rate": 0.0001624485185501721,
      "loss": 1.1648,
      "step": 21160
    },
    {
      "epoch": 0.857762200927858,
      "grad_norm": 0.5684775710105896,
      "learning_rate": 0.00016241535034581026,
      "loss": 1.128,
      "step": 21170
    },
    {
      "epoch": 0.8581673791049614,
      "grad_norm": 0.6151962876319885,
      "learning_rate": 0.00016238217088905015,
      "loss": 1.1906,
      "step": 21180
    },
    {
      "epoch": 0.8585725572820648,
      "grad_norm": 0.5841256976127625,
      "learning_rate": 0.00016234898018587337,
      "loss": 1.1456,
      "step": 21190
    },
    {
      "epoch": 0.8589777354591682,
      "grad_norm": 0.54506516456604,
      "learning_rate": 0.00016231577824226366,
      "loss": 1.1485,
      "step": 21200
    },
    {
      "epoch": 0.8593829136362715,
      "grad_norm": 0.6493832468986511,
      "learning_rate": 0.00016228256506420677,
      "loss": 1.1044,
      "step": 21210
    },
    {
      "epoch": 0.8597880918133749,
      "grad_norm": 0.5731949806213379,
      "learning_rate": 0.00016224934065769045,
      "loss": 1.1313,
      "step": 21220
    },
    {
      "epoch": 0.8601932699904783,
      "grad_norm": 0.5401985049247742,
      "learning_rate": 0.00016221610502870442,
      "loss": 1.1061,
      "step": 21230
    },
    {
      "epoch": 0.8605984481675817,
      "grad_norm": 0.6377050280570984,
      "learning_rate": 0.0001621828581832406,
      "loss": 1.121,
      "step": 21240
    },
    {
      "epoch": 0.8610036263446851,
      "grad_norm": 0.6223161220550537,
      "learning_rate": 0.00016214960012729272,
      "loss": 1.0922,
      "step": 21250
    },
    {
      "epoch": 0.8614088045217885,
      "grad_norm": 0.6274843215942383,
      "learning_rate": 0.00016211633086685664,
      "loss": 1.1591,
      "step": 21260
    },
    {
      "epoch": 0.8618139826988919,
      "grad_norm": 0.5670580863952637,
      "learning_rate": 0.00016208305040793026,
      "loss": 1.1542,
      "step": 21270
    },
    {
      "epoch": 0.8622191608759953,
      "grad_norm": 0.6003251671791077,
      "learning_rate": 0.00016204975875651345,
      "loss": 1.1058,
      "step": 21280
    },
    {
      "epoch": 0.8626243390530987,
      "grad_norm": 0.5238631963729858,
      "learning_rate": 0.0001620164559186081,
      "loss": 1.1098,
      "step": 21290
    },
    {
      "epoch": 0.8630295172302019,
      "grad_norm": 0.5897361040115356,
      "learning_rate": 0.00016198314190021815,
      "loss": 1.1109,
      "step": 21300
    },
    {
      "epoch": 0.8634346954073053,
      "grad_norm": 0.5555857419967651,
      "learning_rate": 0.00016194981670734953,
      "loss": 1.1075,
      "step": 21310
    },
    {
      "epoch": 0.8638398735844087,
      "grad_norm": 0.5815773606300354,
      "learning_rate": 0.0001619164803460102,
      "loss": 1.1825,
      "step": 21320
    },
    {
      "epoch": 0.8642450517615121,
      "grad_norm": 0.5751698017120361,
      "learning_rate": 0.00016188313282221008,
      "loss": 1.1657,
      "step": 21330
    },
    {
      "epoch": 0.8646502299386155,
      "grad_norm": 0.6503629684448242,
      "learning_rate": 0.00016184977414196122,
      "loss": 1.1038,
      "step": 21340
    },
    {
      "epoch": 0.8650554081157189,
      "grad_norm": 0.5651002526283264,
      "learning_rate": 0.00016181640431127753,
      "loss": 1.1325,
      "step": 21350
    },
    {
      "epoch": 0.8654605862928223,
      "grad_norm": 0.5746461749076843,
      "learning_rate": 0.0001617830233361751,
      "loss": 1.1318,
      "step": 21360
    },
    {
      "epoch": 0.8658657644699257,
      "grad_norm": 0.5757253766059875,
      "learning_rate": 0.00016174963122267183,
      "loss": 1.1437,
      "step": 21370
    },
    {
      "epoch": 0.8662709426470291,
      "grad_norm": 0.5803709626197815,
      "learning_rate": 0.00016171622797678782,
      "loss": 1.1277,
      "step": 21380
    },
    {
      "epoch": 0.8666761208241324,
      "grad_norm": 0.5524091124534607,
      "learning_rate": 0.00016168281360454505,
      "loss": 1.1587,
      "step": 21390
    },
    {
      "epoch": 0.8670812990012358,
      "grad_norm": 0.6410770416259766,
      "learning_rate": 0.00016164938811196757,
      "loss": 1.1161,
      "step": 21400
    },
    {
      "epoch": 0.8674864771783392,
      "grad_norm": 0.5896059274673462,
      "learning_rate": 0.00016161595150508137,
      "loss": 1.1038,
      "step": 21410
    },
    {
      "epoch": 0.8678916553554425,
      "grad_norm": 0.6336413621902466,
      "learning_rate": 0.00016158250378991458,
      "loss": 1.1428,
      "step": 21420
    },
    {
      "epoch": 0.8682968335325459,
      "grad_norm": 0.5821861028671265,
      "learning_rate": 0.00016154904497249712,
      "loss": 1.1434,
      "step": 21430
    },
    {
      "epoch": 0.8687020117096493,
      "grad_norm": 0.5858997106552124,
      "learning_rate": 0.0001615155750588611,
      "loss": 1.1447,
      "step": 21440
    },
    {
      "epoch": 0.8691071898867527,
      "grad_norm": 0.5922039151191711,
      "learning_rate": 0.00016148209405504057,
      "loss": 1.1371,
      "step": 21450
    },
    {
      "epoch": 0.8695123680638561,
      "grad_norm": 0.6145365834236145,
      "learning_rate": 0.00016144860196707156,
      "loss": 1.1267,
      "step": 21460
    },
    {
      "epoch": 0.8699175462409595,
      "grad_norm": 0.6407356858253479,
      "learning_rate": 0.00016141509880099206,
      "loss": 1.1242,
      "step": 21470
    },
    {
      "epoch": 0.8703227244180628,
      "grad_norm": 0.559052586555481,
      "learning_rate": 0.00016138158456284216,
      "loss": 1.1167,
      "step": 21480
    },
    {
      "epoch": 0.8707279025951662,
      "grad_norm": 0.5389677286148071,
      "learning_rate": 0.0001613480592586639,
      "loss": 1.0967,
      "step": 21490
    },
    {
      "epoch": 0.8711330807722696,
      "grad_norm": 0.6053739190101624,
      "learning_rate": 0.00016131452289450127,
      "loss": 1.1328,
      "step": 21500
    },
    {
      "epoch": 0.871538258949373,
      "grad_norm": 0.5404178500175476,
      "learning_rate": 0.00016128097547640029,
      "loss": 1.1012,
      "step": 21510
    },
    {
      "epoch": 0.8719434371264764,
      "grad_norm": 0.563388466835022,
      "learning_rate": 0.00016124741701040903,
      "loss": 1.1054,
      "step": 21520
    },
    {
      "epoch": 0.8723486153035798,
      "grad_norm": 0.615433394908905,
      "learning_rate": 0.00016121384750257742,
      "loss": 1.1518,
      "step": 21530
    },
    {
      "epoch": 0.8727537934806832,
      "grad_norm": 0.6065188050270081,
      "learning_rate": 0.0001611802669589575,
      "loss": 1.1069,
      "step": 21540
    },
    {
      "epoch": 0.8731589716577866,
      "grad_norm": 0.6055300235748291,
      "learning_rate": 0.00016114667538560324,
      "loss": 1.1158,
      "step": 21550
    },
    {
      "epoch": 0.87356414983489,
      "grad_norm": 0.6649710536003113,
      "learning_rate": 0.00016111307278857066,
      "loss": 1.1102,
      "step": 21560
    },
    {
      "epoch": 0.8739693280119932,
      "grad_norm": 0.5439127087593079,
      "learning_rate": 0.00016107945917391762,
      "loss": 1.1417,
      "step": 21570
    },
    {
      "epoch": 0.8743745061890966,
      "grad_norm": 0.5347424745559692,
      "learning_rate": 0.0001610458345477042,
      "loss": 1.1074,
      "step": 21580
    },
    {
      "epoch": 0.8747796843662,
      "grad_norm": 0.6412347555160522,
      "learning_rate": 0.00016101219891599225,
      "loss": 1.1324,
      "step": 21590
    },
    {
      "epoch": 0.8751848625433034,
      "grad_norm": 0.5170292854309082,
      "learning_rate": 0.00016097855228484567,
      "loss": 1.1367,
      "step": 21600
    },
    {
      "epoch": 0.8755900407204068,
      "grad_norm": 0.6159192323684692,
      "learning_rate": 0.00016094489466033043,
      "loss": 1.1558,
      "step": 21610
    },
    {
      "epoch": 0.8759952188975102,
      "grad_norm": 0.5975338816642761,
      "learning_rate": 0.00016091122604851434,
      "loss": 1.1544,
      "step": 21620
    },
    {
      "epoch": 0.8764003970746136,
      "grad_norm": 0.5894503593444824,
      "learning_rate": 0.00016087754645546733,
      "loss": 1.1233,
      "step": 21630
    },
    {
      "epoch": 0.876805575251717,
      "grad_norm": 0.5609699487686157,
      "learning_rate": 0.00016084385588726122,
      "loss": 1.0786,
      "step": 21640
    },
    {
      "epoch": 0.8772107534288204,
      "grad_norm": 0.5259765386581421,
      "learning_rate": 0.0001608101543499698,
      "loss": 1.1303,
      "step": 21650
    },
    {
      "epoch": 0.8776159316059237,
      "grad_norm": 0.5265316367149353,
      "learning_rate": 0.0001607764418496689,
      "loss": 1.1742,
      "step": 21660
    },
    {
      "epoch": 0.878021109783027,
      "grad_norm": 0.5825567841529846,
      "learning_rate": 0.00016074271839243626,
      "loss": 1.1269,
      "step": 21670
    },
    {
      "epoch": 0.8784262879601304,
      "grad_norm": 0.5706741809844971,
      "learning_rate": 0.00016070898398435167,
      "loss": 1.1501,
      "step": 21680
    },
    {
      "epoch": 0.8788314661372338,
      "grad_norm": 0.6116732954978943,
      "learning_rate": 0.00016067523863149688,
      "loss": 1.1433,
      "step": 21690
    },
    {
      "epoch": 0.8792366443143372,
      "grad_norm": 0.6332910656929016,
      "learning_rate": 0.00016064148233995554,
      "loss": 1.1405,
      "step": 21700
    },
    {
      "epoch": 0.8796418224914406,
      "grad_norm": 0.5967959761619568,
      "learning_rate": 0.00016060771511581327,
      "loss": 1.0858,
      "step": 21710
    },
    {
      "epoch": 0.880047000668544,
      "grad_norm": 0.5546802878379822,
      "learning_rate": 0.00016057393696515784,
      "loss": 1.11,
      "step": 21720
    },
    {
      "epoch": 0.8804521788456474,
      "grad_norm": 0.539550244808197,
      "learning_rate": 0.00016054014789407877,
      "loss": 1.1337,
      "step": 21730
    },
    {
      "epoch": 0.8808573570227508,
      "grad_norm": 0.6187642812728882,
      "learning_rate": 0.00016050634790866765,
      "loss": 1.122,
      "step": 21740
    },
    {
      "epoch": 0.8812625351998541,
      "grad_norm": 0.5544891357421875,
      "learning_rate": 0.00016047253701501808,
      "loss": 1.132,
      "step": 21750
    },
    {
      "epoch": 0.8816677133769575,
      "grad_norm": 0.5681062340736389,
      "learning_rate": 0.00016043871521922548,
      "loss": 1.1725,
      "step": 21760
    },
    {
      "epoch": 0.8820728915540609,
      "grad_norm": 0.6480683088302612,
      "learning_rate": 0.00016040488252738744,
      "loss": 1.1138,
      "step": 21770
    },
    {
      "epoch": 0.8824780697311643,
      "grad_norm": 0.5628143548965454,
      "learning_rate": 0.00016037103894560333,
      "loss": 1.1083,
      "step": 21780
    },
    {
      "epoch": 0.8828832479082677,
      "grad_norm": 0.5695332884788513,
      "learning_rate": 0.00016033718447997454,
      "loss": 1.1408,
      "step": 21790
    },
    {
      "epoch": 0.8832884260853711,
      "grad_norm": 0.5391930341720581,
      "learning_rate": 0.00016030331913660453,
      "loss": 1.1169,
      "step": 21800
    },
    {
      "epoch": 0.8836936042624745,
      "grad_norm": 0.5575686693191528,
      "learning_rate": 0.00016026944292159856,
      "loss": 1.1004,
      "step": 21810
    },
    {
      "epoch": 0.8840987824395778,
      "grad_norm": 0.5250729322433472,
      "learning_rate": 0.0001602355558410639,
      "loss": 1.1385,
      "step": 21820
    },
    {
      "epoch": 0.8845039606166812,
      "grad_norm": 0.710594117641449,
      "learning_rate": 0.00016020165790110987,
      "loss": 1.1268,
      "step": 21830
    },
    {
      "epoch": 0.8849091387937845,
      "grad_norm": 0.6024810075759888,
      "learning_rate": 0.00016016774910784765,
      "loss": 1.1457,
      "step": 21840
    },
    {
      "epoch": 0.8853143169708879,
      "grad_norm": 0.6588172912597656,
      "learning_rate": 0.00016013382946739038,
      "loss": 1.1239,
      "step": 21850
    },
    {
      "epoch": 0.8857194951479913,
      "grad_norm": 0.5623536705970764,
      "learning_rate": 0.00016009989898585318,
      "loss": 1.0993,
      "step": 21860
    },
    {
      "epoch": 0.8861246733250947,
      "grad_norm": 0.585561215877533,
      "learning_rate": 0.00016006595766935314,
      "loss": 1.131,
      "step": 21870
    },
    {
      "epoch": 0.8865298515021981,
      "grad_norm": 0.6566945314407349,
      "learning_rate": 0.0001600320055240093,
      "loss": 1.143,
      "step": 21880
    },
    {
      "epoch": 0.8869350296793015,
      "grad_norm": 0.5520895719528198,
      "learning_rate": 0.00015999804255594258,
      "loss": 1.1629,
      "step": 21890
    },
    {
      "epoch": 0.8873402078564049,
      "grad_norm": 0.5760812759399414,
      "learning_rate": 0.00015996406877127604,
      "loss": 1.1167,
      "step": 21900
    },
    {
      "epoch": 0.8877453860335083,
      "grad_norm": 0.6121431589126587,
      "learning_rate": 0.0001599300841761344,
      "loss": 1.1037,
      "step": 21910
    },
    {
      "epoch": 0.8881505642106117,
      "grad_norm": 0.5704744458198547,
      "learning_rate": 0.00015989608877664462,
      "loss": 1.1282,
      "step": 21920
    },
    {
      "epoch": 0.888555742387715,
      "grad_norm": 0.5487170219421387,
      "learning_rate": 0.00015986208257893537,
      "loss": 1.1396,
      "step": 21930
    },
    {
      "epoch": 0.8889609205648183,
      "grad_norm": 0.6009952425956726,
      "learning_rate": 0.00015982806558913746,
      "loss": 1.1124,
      "step": 21940
    },
    {
      "epoch": 0.8893660987419217,
      "grad_norm": 0.6672473549842834,
      "learning_rate": 0.0001597940378133835,
      "loss": 1.1461,
      "step": 21950
    },
    {
      "epoch": 0.8897712769190251,
      "grad_norm": 0.5762639045715332,
      "learning_rate": 0.00015975999925780813,
      "loss": 1.1086,
      "step": 21960
    },
    {
      "epoch": 0.8901764550961285,
      "grad_norm": 0.5763003826141357,
      "learning_rate": 0.00015972594992854792,
      "loss": 1.165,
      "step": 21970
    },
    {
      "epoch": 0.8905816332732319,
      "grad_norm": 0.5903466939926147,
      "learning_rate": 0.00015969188983174136,
      "loss": 1.1196,
      "step": 21980
    },
    {
      "epoch": 0.8909868114503353,
      "grad_norm": 0.55279541015625,
      "learning_rate": 0.00015965781897352887,
      "loss": 1.1114,
      "step": 21990
    },
    {
      "epoch": 0.8913919896274387,
      "grad_norm": 0.5702687501907349,
      "learning_rate": 0.00015962373736005288,
      "loss": 1.1135,
      "step": 22000
    },
    {
      "epoch": 0.8917971678045421,
      "grad_norm": 0.5635493993759155,
      "learning_rate": 0.00015958964499745764,
      "loss": 1.1634,
      "step": 22010
    },
    {
      "epoch": 0.8922023459816454,
      "grad_norm": 0.5770784616470337,
      "learning_rate": 0.00015955554189188946,
      "loss": 1.1625,
      "step": 22020
    },
    {
      "epoch": 0.8926075241587488,
      "grad_norm": 0.5738016366958618,
      "learning_rate": 0.00015952142804949652,
      "loss": 1.1369,
      "step": 22030
    },
    {
      "epoch": 0.8930127023358522,
      "grad_norm": 0.6038731336593628,
      "learning_rate": 0.00015948730347642895,
      "loss": 1.1139,
      "step": 22040
    },
    {
      "epoch": 0.8934178805129556,
      "grad_norm": 0.5604899525642395,
      "learning_rate": 0.00015945316817883885,
      "loss": 1.1101,
      "step": 22050
    },
    {
      "epoch": 0.893823058690059,
      "grad_norm": 0.595207929611206,
      "learning_rate": 0.00015941902216288015,
      "loss": 1.1145,
      "step": 22060
    },
    {
      "epoch": 0.8942282368671624,
      "grad_norm": 0.5854849219322205,
      "learning_rate": 0.0001593848654347088,
      "loss": 1.1197,
      "step": 22070
    },
    {
      "epoch": 0.8946334150442657,
      "grad_norm": 0.6397050023078918,
      "learning_rate": 0.00015935069800048271,
      "loss": 1.1831,
      "step": 22080
    },
    {
      "epoch": 0.8950385932213691,
      "grad_norm": 0.5074746608734131,
      "learning_rate": 0.00015931651986636163,
      "loss": 1.1144,
      "step": 22090
    },
    {
      "epoch": 0.8954437713984725,
      "grad_norm": 0.6112139821052551,
      "learning_rate": 0.0001592823310385073,
      "loss": 1.1315,
      "step": 22100
    },
    {
      "epoch": 0.8958489495755758,
      "grad_norm": 0.6483452320098877,
      "learning_rate": 0.00015924813152308332,
      "loss": 1.1469,
      "step": 22110
    },
    {
      "epoch": 0.8962541277526792,
      "grad_norm": 0.5848171710968018,
      "learning_rate": 0.00015921392132625534,
      "loss": 1.1236,
      "step": 22120
    },
    {
      "epoch": 0.8966593059297826,
      "grad_norm": 0.5982404351234436,
      "learning_rate": 0.00015917970045419084,
      "loss": 1.1267,
      "step": 22130
    },
    {
      "epoch": 0.897064484106886,
      "grad_norm": 0.5796005129814148,
      "learning_rate": 0.0001591454689130592,
      "loss": 1.1203,
      "step": 22140
    },
    {
      "epoch": 0.8974696622839894,
      "grad_norm": 0.5779382586479187,
      "learning_rate": 0.0001591112267090318,
      "loss": 1.1186,
      "step": 22150
    },
    {
      "epoch": 0.8978748404610928,
      "grad_norm": 0.5756171345710754,
      "learning_rate": 0.00015907697384828193,
      "loss": 1.1066,
      "step": 22160
    },
    {
      "epoch": 0.8982800186381962,
      "grad_norm": 0.637067437171936,
      "learning_rate": 0.0001590427103369848,
      "loss": 1.1368,
      "step": 22170
    },
    {
      "epoch": 0.8986851968152996,
      "grad_norm": 0.5158118009567261,
      "learning_rate": 0.00015900843618131742,
      "loss": 1.1351,
      "step": 22180
    },
    {
      "epoch": 0.899090374992403,
      "grad_norm": 0.5244913101196289,
      "learning_rate": 0.00015897415138745892,
      "loss": 1.1468,
      "step": 22190
    },
    {
      "epoch": 0.8994955531695062,
      "grad_norm": 0.552312970161438,
      "learning_rate": 0.00015893985596159024,
      "loss": 1.0785,
      "step": 22200
    },
    {
      "epoch": 0.8999007313466096,
      "grad_norm": 0.6025837659835815,
      "learning_rate": 0.00015890554990989424,
      "loss": 1.1803,
      "step": 22210
    },
    {
      "epoch": 0.900305909523713,
      "grad_norm": 0.5770346522331238,
      "learning_rate": 0.00015887123323855563,
      "loss": 1.1405,
      "step": 22220
    },
    {
      "epoch": 0.9007110877008164,
      "grad_norm": 0.5086196064949036,
      "learning_rate": 0.0001588369059537612,
      "loss": 1.1136,
      "step": 22230
    },
    {
      "epoch": 0.9011162658779198,
      "grad_norm": 0.5578281283378601,
      "learning_rate": 0.00015880256806169953,
      "loss": 1.1189,
      "step": 22240
    },
    {
      "epoch": 0.9015214440550232,
      "grad_norm": 0.6034433245658875,
      "learning_rate": 0.00015876821956856116,
      "loss": 1.1216,
      "step": 22250
    },
    {
      "epoch": 0.9019266222321266,
      "grad_norm": 0.581527054309845,
      "learning_rate": 0.00015873386048053848,
      "loss": 1.1558,
      "step": 22260
    },
    {
      "epoch": 0.90233180040923,
      "grad_norm": 0.5547844171524048,
      "learning_rate": 0.00015869949080382588,
      "loss": 1.139,
      "step": 22270
    },
    {
      "epoch": 0.9027369785863334,
      "grad_norm": 0.5638455152511597,
      "learning_rate": 0.00015866511054461957,
      "loss": 1.1279,
      "step": 22280
    },
    {
      "epoch": 0.9031421567634367,
      "grad_norm": 0.6095684766769409,
      "learning_rate": 0.00015863071970911773,
      "loss": 1.1368,
      "step": 22290
    },
    {
      "epoch": 0.9035473349405401,
      "grad_norm": 0.6138654947280884,
      "learning_rate": 0.00015859631830352045,
      "loss": 1.1478,
      "step": 22300
    },
    {
      "epoch": 0.9039525131176435,
      "grad_norm": 0.6253608465194702,
      "learning_rate": 0.00015856190633402968,
      "loss": 1.1205,
      "step": 22310
    },
    {
      "epoch": 0.9043576912947469,
      "grad_norm": 0.5470446944236755,
      "learning_rate": 0.0001585274838068493,
      "loss": 1.1153,
      "step": 22320
    },
    {
      "epoch": 0.9047628694718503,
      "grad_norm": 0.5418273210525513,
      "learning_rate": 0.00015849305072818508,
      "loss": 1.1101,
      "step": 22330
    },
    {
      "epoch": 0.9051680476489536,
      "grad_norm": 0.58623868227005,
      "learning_rate": 0.0001584586071042447,
      "loss": 1.1411,
      "step": 22340
    },
    {
      "epoch": 0.905573225826057,
      "grad_norm": 0.4886528253555298,
      "learning_rate": 0.00015842415294123774,
      "loss": 1.1486,
      "step": 22350
    },
    {
      "epoch": 0.9059784040031604,
      "grad_norm": 0.5207862257957458,
      "learning_rate": 0.00015838968824537577,
      "loss": 1.1154,
      "step": 22360
    },
    {
      "epoch": 0.9063835821802638,
      "grad_norm": 0.5261946320533752,
      "learning_rate": 0.00015835521302287207,
      "loss": 1.1248,
      "step": 22370
    },
    {
      "epoch": 0.9067887603573671,
      "grad_norm": 0.7131884098052979,
      "learning_rate": 0.00015832072727994193,
      "loss": 1.097,
      "step": 22380
    },
    {
      "epoch": 0.9071939385344705,
      "grad_norm": 0.5283380746841431,
      "learning_rate": 0.0001582862310228026,
      "loss": 1.1199,
      "step": 22390
    },
    {
      "epoch": 0.9075991167115739,
      "grad_norm": 0.7059460878372192,
      "learning_rate": 0.0001582517242576731,
      "loss": 1.1201,
      "step": 22400
    },
    {
      "epoch": 0.9080042948886773,
      "grad_norm": 0.6280739903450012,
      "learning_rate": 0.00015821720699077437,
      "loss": 1.1307,
      "step": 22410
    },
    {
      "epoch": 0.9084094730657807,
      "grad_norm": 0.6193175911903381,
      "learning_rate": 0.00015818267922832933,
      "loss": 1.1212,
      "step": 22420
    },
    {
      "epoch": 0.9088146512428841,
      "grad_norm": 0.6206029653549194,
      "learning_rate": 0.00015814814097656272,
      "loss": 1.1559,
      "step": 22430
    },
    {
      "epoch": 0.9092198294199875,
      "grad_norm": 0.5524816513061523,
      "learning_rate": 0.00015811359224170114,
      "loss": 1.1288,
      "step": 22440
    },
    {
      "epoch": 0.9096250075970909,
      "grad_norm": 0.5829833745956421,
      "learning_rate": 0.00015807903302997317,
      "loss": 1.1115,
      "step": 22450
    },
    {
      "epoch": 0.9100301857741943,
      "grad_norm": 0.6606831550598145,
      "learning_rate": 0.00015804446334760926,
      "loss": 1.126,
      "step": 22460
    },
    {
      "epoch": 0.9104353639512975,
      "grad_norm": 0.5584904551506042,
      "learning_rate": 0.00015800988320084162,
      "loss": 1.0793,
      "step": 22470
    },
    {
      "epoch": 0.9108405421284009,
      "grad_norm": 0.5341227054595947,
      "learning_rate": 0.00015797529259590451,
      "loss": 1.1081,
      "step": 22480
    },
    {
      "epoch": 0.9112457203055043,
      "grad_norm": 0.5852908492088318,
      "learning_rate": 0.00015794069153903407,
      "loss": 1.1446,
      "step": 22490
    },
    {
      "epoch": 0.9116508984826077,
      "grad_norm": 0.5491456985473633,
      "learning_rate": 0.00015790608003646815,
      "loss": 1.1191,
      "step": 22500
    },
    {
      "epoch": 0.9120560766597111,
      "grad_norm": 0.5139252543449402,
      "learning_rate": 0.00015787145809444665,
      "loss": 1.1267,
      "step": 22510
    },
    {
      "epoch": 0.9124612548368145,
      "grad_norm": 0.556383490562439,
      "learning_rate": 0.00015783682571921133,
      "loss": 1.1261,
      "step": 22520
    },
    {
      "epoch": 0.9128664330139179,
      "grad_norm": 0.6016799807548523,
      "learning_rate": 0.00015780218291700578,
      "loss": 1.141,
      "step": 22530
    },
    {
      "epoch": 0.9132716111910213,
      "grad_norm": 0.5491769313812256,
      "learning_rate": 0.00015776752969407545,
      "loss": 1.1144,
      "step": 22540
    },
    {
      "epoch": 0.9136767893681247,
      "grad_norm": 0.5616195797920227,
      "learning_rate": 0.0001577328660566678,
      "loss": 1.1082,
      "step": 22550
    },
    {
      "epoch": 0.914081967545228,
      "grad_norm": 0.6838018298149109,
      "learning_rate": 0.000157698192011032,
      "loss": 1.1272,
      "step": 22560
    },
    {
      "epoch": 0.9144871457223314,
      "grad_norm": 0.6232140064239502,
      "learning_rate": 0.00015766350756341924,
      "loss": 1.1104,
      "step": 22570
    },
    {
      "epoch": 0.9148923238994348,
      "grad_norm": 0.5936967730522156,
      "learning_rate": 0.00015762881272008247,
      "loss": 1.1172,
      "step": 22580
    },
    {
      "epoch": 0.9152975020765381,
      "grad_norm": 0.5563666224479675,
      "learning_rate": 0.00015759410748727662,
      "loss": 1.1395,
      "step": 22590
    },
    {
      "epoch": 0.9157026802536415,
      "grad_norm": 0.619244396686554,
      "learning_rate": 0.00015755939187125836,
      "loss": 1.1379,
      "step": 22600
    },
    {
      "epoch": 0.9161078584307449,
      "grad_norm": 0.6318624019622803,
      "learning_rate": 0.00015752466587828633,
      "loss": 1.1413,
      "step": 22610
    },
    {
      "epoch": 0.9165130366078483,
      "grad_norm": 0.6512631177902222,
      "learning_rate": 0.00015748992951462108,
      "loss": 1.1598,
      "step": 22620
    },
    {
      "epoch": 0.9169182147849517,
      "grad_norm": 0.661194920539856,
      "learning_rate": 0.00015745518278652493,
      "loss": 1.1403,
      "step": 22630
    },
    {
      "epoch": 0.9173233929620551,
      "grad_norm": 0.5859771370887756,
      "learning_rate": 0.0001574204257002621,
      "loss": 1.1452,
      "step": 22640
    },
    {
      "epoch": 0.9177285711391584,
      "grad_norm": 0.5514333844184875,
      "learning_rate": 0.00015738565826209868,
      "loss": 1.1388,
      "step": 22650
    },
    {
      "epoch": 0.9181337493162618,
      "grad_norm": 0.5758521556854248,
      "learning_rate": 0.00015735088047830268,
      "loss": 1.1225,
      "step": 22660
    },
    {
      "epoch": 0.9185389274933652,
      "grad_norm": 0.6365063786506653,
      "learning_rate": 0.00015731609235514386,
      "loss": 1.1244,
      "step": 22670
    },
    {
      "epoch": 0.9189441056704686,
      "grad_norm": 0.5764655470848083,
      "learning_rate": 0.000157281293898894,
      "loss": 1.1087,
      "step": 22680
    },
    {
      "epoch": 0.919349283847572,
      "grad_norm": 0.6204686164855957,
      "learning_rate": 0.00015724648511582654,
      "loss": 1.1044,
      "step": 22690
    },
    {
      "epoch": 0.9197544620246754,
      "grad_norm": 0.6116904020309448,
      "learning_rate": 0.00015721166601221698,
      "loss": 1.125,
      "step": 22700
    },
    {
      "epoch": 0.9201596402017788,
      "grad_norm": 0.5571669340133667,
      "learning_rate": 0.00015717683659434258,
      "loss": 1.1095,
      "step": 22710
    },
    {
      "epoch": 0.9205648183788822,
      "grad_norm": 0.5936890244483948,
      "learning_rate": 0.00015714199686848245,
      "loss": 1.1406,
      "step": 22720
    },
    {
      "epoch": 0.9209699965559855,
      "grad_norm": 0.5834794640541077,
      "learning_rate": 0.00015710714684091762,
      "loss": 1.1258,
      "step": 22730
    },
    {
      "epoch": 0.9213751747330888,
      "grad_norm": 0.6592190265655518,
      "learning_rate": 0.00015707228651793093,
      "loss": 1.1169,
      "step": 22740
    },
    {
      "epoch": 0.9217803529101922,
      "grad_norm": 0.6072849035263062,
      "learning_rate": 0.00015703741590580706,
      "loss": 1.1383,
      "step": 22750
    },
    {
      "epoch": 0.9221855310872956,
      "grad_norm": 0.6562568545341492,
      "learning_rate": 0.00015700253501083262,
      "loss": 1.1613,
      "step": 22760
    },
    {
      "epoch": 0.922590709264399,
      "grad_norm": 0.5737852454185486,
      "learning_rate": 0.000156967643839296,
      "loss": 1.122,
      "step": 22770
    },
    {
      "epoch": 0.9229958874415024,
      "grad_norm": 0.5962287187576294,
      "learning_rate": 0.0001569327423974875,
      "loss": 1.1086,
      "step": 22780
    },
    {
      "epoch": 0.9234010656186058,
      "grad_norm": 0.6251916885375977,
      "learning_rate": 0.00015689783069169916,
      "loss": 1.1458,
      "step": 22790
    },
    {
      "epoch": 0.9238062437957092,
      "grad_norm": 0.6200146079063416,
      "learning_rate": 0.00015686290872822504,
      "loss": 1.1483,
      "step": 22800
    },
    {
      "epoch": 0.9242114219728126,
      "grad_norm": 0.6361082792282104,
      "learning_rate": 0.0001568279765133609,
      "loss": 1.1294,
      "step": 22810
    },
    {
      "epoch": 0.924616600149916,
      "grad_norm": 0.6360374093055725,
      "learning_rate": 0.00015679303405340445,
      "loss": 1.1166,
      "step": 22820
    },
    {
      "epoch": 0.9250217783270193,
      "grad_norm": 0.6286635994911194,
      "learning_rate": 0.0001567580813546552,
      "loss": 1.1086,
      "step": 22830
    },
    {
      "epoch": 0.9254269565041227,
      "grad_norm": 0.5661700963973999,
      "learning_rate": 0.0001567231184234145,
      "loss": 1.139,
      "step": 22840
    },
    {
      "epoch": 0.925832134681226,
      "grad_norm": 0.5654430389404297,
      "learning_rate": 0.0001566881452659855,
      "loss": 1.0863,
      "step": 22850
    },
    {
      "epoch": 0.9262373128583294,
      "grad_norm": 0.5933253169059753,
      "learning_rate": 0.00015665316188867337,
      "loss": 1.087,
      "step": 22860
    },
    {
      "epoch": 0.9266424910354328,
      "grad_norm": 0.652202844619751,
      "learning_rate": 0.00015661816829778494,
      "loss": 1.0986,
      "step": 22870
    },
    {
      "epoch": 0.9270476692125362,
      "grad_norm": 0.6095232963562012,
      "learning_rate": 0.0001565831644996289,
      "loss": 1.1258,
      "step": 22880
    },
    {
      "epoch": 0.9274528473896396,
      "grad_norm": 0.6122059226036072,
      "learning_rate": 0.00015654815050051586,
      "loss": 1.1422,
      "step": 22890
    },
    {
      "epoch": 0.927858025566743,
      "grad_norm": 0.5635743141174316,
      "learning_rate": 0.0001565131263067583,
      "loss": 1.0975,
      "step": 22900
    },
    {
      "epoch": 0.9282632037438464,
      "grad_norm": 0.5341784358024597,
      "learning_rate": 0.0001564780919246704,
      "loss": 1.1105,
      "step": 22910
    },
    {
      "epoch": 0.9286683819209497,
      "grad_norm": 0.5135253667831421,
      "learning_rate": 0.00015644304736056822,
      "loss": 1.12,
      "step": 22920
    },
    {
      "epoch": 0.9290735600980531,
      "grad_norm": 0.5830720663070679,
      "learning_rate": 0.00015640799262076974,
      "loss": 1.1494,
      "step": 22930
    },
    {
      "epoch": 0.9294787382751565,
      "grad_norm": 0.5711979866027832,
      "learning_rate": 0.00015637292771159472,
      "loss": 1.11,
      "step": 22940
    },
    {
      "epoch": 0.9298839164522599,
      "grad_norm": 0.663058340549469,
      "learning_rate": 0.0001563378526393647,
      "loss": 1.0999,
      "step": 22950
    },
    {
      "epoch": 0.9302890946293633,
      "grad_norm": 0.5671886205673218,
      "learning_rate": 0.00015630276741040314,
      "loss": 1.1577,
      "step": 22960
    },
    {
      "epoch": 0.9306942728064667,
      "grad_norm": 0.5570474863052368,
      "learning_rate": 0.00015626767203103529,
      "loss": 1.1614,
      "step": 22970
    },
    {
      "epoch": 0.93109945098357,
      "grad_norm": 0.5648364424705505,
      "learning_rate": 0.00015623256650758825,
      "loss": 1.1136,
      "step": 22980
    },
    {
      "epoch": 0.9315046291606734,
      "grad_norm": 0.6342676877975464,
      "learning_rate": 0.00015619745084639088,
      "loss": 1.2044,
      "step": 22990
    },
    {
      "epoch": 0.9319098073377768,
      "grad_norm": 0.6454786062240601,
      "learning_rate": 0.00015616232505377396,
      "loss": 1.1418,
      "step": 23000
    },
    {
      "epoch": 0.9323149855148801,
      "grad_norm": 0.596078097820282,
      "learning_rate": 0.0001561271891360701,
      "loss": 1.1012,
      "step": 23010
    },
    {
      "epoch": 0.9327201636919835,
      "grad_norm": 0.5881653428077698,
      "learning_rate": 0.0001560920430996136,
      "loss": 1.0968,
      "step": 23020
    },
    {
      "epoch": 0.9331253418690869,
      "grad_norm": 0.7180758714675903,
      "learning_rate": 0.00015605688695074076,
      "loss": 1.1436,
      "step": 23030
    },
    {
      "epoch": 0.9335305200461903,
      "grad_norm": 0.5564097762107849,
      "learning_rate": 0.00015602172069578956,
      "loss": 1.1049,
      "step": 23040
    },
    {
      "epoch": 0.9339356982232937,
      "grad_norm": 0.6576531529426575,
      "learning_rate": 0.00015598654434109989,
      "loss": 1.1194,
      "step": 23050
    },
    {
      "epoch": 0.9343408764003971,
      "grad_norm": 0.6002846360206604,
      "learning_rate": 0.00015595135789301341,
      "loss": 1.1502,
      "step": 23060
    },
    {
      "epoch": 0.9347460545775005,
      "grad_norm": 0.6625397801399231,
      "learning_rate": 0.00015591616135787365,
      "loss": 1.1153,
      "step": 23070
    },
    {
      "epoch": 0.9351512327546039,
      "grad_norm": 0.6108580231666565,
      "learning_rate": 0.00015588095474202595,
      "loss": 1.1281,
      "step": 23080
    },
    {
      "epoch": 0.9355564109317073,
      "grad_norm": 0.5966638326644897,
      "learning_rate": 0.00015584573805181738,
      "loss": 1.1453,
      "step": 23090
    },
    {
      "epoch": 0.9359615891088106,
      "grad_norm": 0.5821145176887512,
      "learning_rate": 0.000155810511293597,
      "loss": 1.1301,
      "step": 23100
    },
    {
      "epoch": 0.936366767285914,
      "grad_norm": 0.5240674018859863,
      "learning_rate": 0.00015577527447371546,
      "loss": 1.1308,
      "step": 23110
    },
    {
      "epoch": 0.9367719454630173,
      "grad_norm": 0.5242542624473572,
      "learning_rate": 0.00015574002759852538,
      "loss": 1.173,
      "step": 23120
    },
    {
      "epoch": 0.9371771236401207,
      "grad_norm": 0.6226516366004944,
      "learning_rate": 0.00015570477067438122,
      "loss": 1.1223,
      "step": 23130
    },
    {
      "epoch": 0.9375823018172241,
      "grad_norm": 0.6081875562667847,
      "learning_rate": 0.00015566950370763912,
      "loss": 1.1243,
      "step": 23140
    },
    {
      "epoch": 0.9379874799943275,
      "grad_norm": 0.5547734498977661,
      "learning_rate": 0.00015563422670465712,
      "loss": 1.1537,
      "step": 23150
    },
    {
      "epoch": 0.9383926581714309,
      "grad_norm": 0.608329713344574,
      "learning_rate": 0.00015559893967179506,
      "loss": 1.1408,
      "step": 23160
    },
    {
      "epoch": 0.9387978363485343,
      "grad_norm": 0.58513343334198,
      "learning_rate": 0.00015556364261541455,
      "loss": 1.118,
      "step": 23170
    },
    {
      "epoch": 0.9392030145256377,
      "grad_norm": 0.5485582947731018,
      "learning_rate": 0.00015552833554187908,
      "loss": 1.1131,
      "step": 23180
    },
    {
      "epoch": 0.939608192702741,
      "grad_norm": 0.6036726832389832,
      "learning_rate": 0.00015549301845755383,
      "loss": 1.1119,
      "step": 23190
    },
    {
      "epoch": 0.9400133708798444,
      "grad_norm": 0.6056994795799255,
      "learning_rate": 0.00015545769136880591,
      "loss": 1.1001,
      "step": 23200
    },
    {
      "epoch": 0.9404185490569478,
      "grad_norm": 0.5495977401733398,
      "learning_rate": 0.00015542235428200416,
      "loss": 1.0985,
      "step": 23210
    },
    {
      "epoch": 0.9408237272340512,
      "grad_norm": 0.5525761246681213,
      "learning_rate": 0.00015538700720351924,
      "loss": 1.0959,
      "step": 23220
    },
    {
      "epoch": 0.9412289054111546,
      "grad_norm": 0.8031188249588013,
      "learning_rate": 0.0001553516501397236,
      "loss": 1.1489,
      "step": 23230
    },
    {
      "epoch": 0.941634083588258,
      "grad_norm": 0.5964778661727905,
      "learning_rate": 0.0001553162830969915,
      "loss": 1.1179,
      "step": 23240
    },
    {
      "epoch": 0.9420392617653613,
      "grad_norm": 0.5180347561836243,
      "learning_rate": 0.00015528090608169905,
      "loss": 1.1197,
      "step": 23250
    },
    {
      "epoch": 0.9424444399424647,
      "grad_norm": 0.629002034664154,
      "learning_rate": 0.00015524551910022405,
      "loss": 1.1334,
      "step": 23260
    },
    {
      "epoch": 0.9428496181195681,
      "grad_norm": 0.5472930073738098,
      "learning_rate": 0.00015521012215894618,
      "loss": 1.1499,
      "step": 23270
    },
    {
      "epoch": 0.9432547962966714,
      "grad_norm": 0.5180639028549194,
      "learning_rate": 0.0001551747152642469,
      "loss": 1.1297,
      "step": 23280
    },
    {
      "epoch": 0.9436599744737748,
      "grad_norm": 0.6137699484825134,
      "learning_rate": 0.0001551392984225094,
      "loss": 1.1323,
      "step": 23290
    },
    {
      "epoch": 0.9440651526508782,
      "grad_norm": 0.5518494844436646,
      "learning_rate": 0.0001551038716401188,
      "loss": 1.1004,
      "step": 23300
    },
    {
      "epoch": 0.9444703308279816,
      "grad_norm": 0.6367759108543396,
      "learning_rate": 0.00015506843492346187,
      "loss": 1.1036,
      "step": 23310
    },
    {
      "epoch": 0.944875509005085,
      "grad_norm": 0.5066690444946289,
      "learning_rate": 0.00015503298827892724,
      "loss": 1.1157,
      "step": 23320
    },
    {
      "epoch": 0.9452806871821884,
      "grad_norm": 0.562651515007019,
      "learning_rate": 0.00015499753171290537,
      "loss": 1.133,
      "step": 23330
    },
    {
      "epoch": 0.9456858653592918,
      "grad_norm": 0.6149501204490662,
      "learning_rate": 0.00015496206523178842,
      "loss": 1.1317,
      "step": 23340
    },
    {
      "epoch": 0.9460910435363952,
      "grad_norm": 0.5395510196685791,
      "learning_rate": 0.00015492658884197038,
      "loss": 1.1175,
      "step": 23350
    },
    {
      "epoch": 0.9464962217134986,
      "grad_norm": 0.5667554140090942,
      "learning_rate": 0.000154891102549847,
      "loss": 1.0909,
      "step": 23360
    },
    {
      "epoch": 0.9469013998906018,
      "grad_norm": 0.6792905926704407,
      "learning_rate": 0.0001548556063618159,
      "loss": 1.107,
      "step": 23370
    },
    {
      "epoch": 0.9473065780677052,
      "grad_norm": 0.54294753074646,
      "learning_rate": 0.00015482010028427634,
      "loss": 1.0949,
      "step": 23380
    },
    {
      "epoch": 0.9477117562448086,
      "grad_norm": 0.5989609360694885,
      "learning_rate": 0.00015478458432362954,
      "loss": 1.1107,
      "step": 23390
    },
    {
      "epoch": 0.948116934421912,
      "grad_norm": 0.5074239373207092,
      "learning_rate": 0.0001547490584862783,
      "loss": 1.0995,
      "step": 23400
    },
    {
      "epoch": 0.9485221125990154,
      "grad_norm": 0.5516952872276306,
      "learning_rate": 0.00015471352277862742,
      "loss": 1.1486,
      "step": 23410
    },
    {
      "epoch": 0.9489272907761188,
      "grad_norm": 0.5319008231163025,
      "learning_rate": 0.00015467797720708333,
      "loss": 1.151,
      "step": 23420
    },
    {
      "epoch": 0.9493324689532222,
      "grad_norm": 0.6122015714645386,
      "learning_rate": 0.00015464242177805422,
      "loss": 1.1397,
      "step": 23430
    },
    {
      "epoch": 0.9497376471303256,
      "grad_norm": 0.552744448184967,
      "learning_rate": 0.0001546068564979502,
      "loss": 1.1089,
      "step": 23440
    },
    {
      "epoch": 0.950142825307429,
      "grad_norm": 0.5724343061447144,
      "learning_rate": 0.000154571281373183,
      "loss": 1.091,
      "step": 23450
    },
    {
      "epoch": 0.9505480034845323,
      "grad_norm": 0.6356295943260193,
      "learning_rate": 0.0001545356964101662,
      "loss": 1.1183,
      "step": 23460
    },
    {
      "epoch": 0.9509531816616357,
      "grad_norm": 0.5559977293014526,
      "learning_rate": 0.00015450010161531522,
      "loss": 1.1027,
      "step": 23470
    },
    {
      "epoch": 0.9513583598387391,
      "grad_norm": 0.6448822617530823,
      "learning_rate": 0.00015446449699504708,
      "loss": 1.1379,
      "step": 23480
    },
    {
      "epoch": 0.9517635380158425,
      "grad_norm": 0.49885571002960205,
      "learning_rate": 0.00015442888255578073,
      "loss": 1.0813,
      "step": 23490
    },
    {
      "epoch": 0.9521687161929459,
      "grad_norm": 0.6018041372299194,
      "learning_rate": 0.00015439325830393687,
      "loss": 1.1388,
      "step": 23500
    },
    {
      "epoch": 0.9525738943700492,
      "grad_norm": 0.5455858111381531,
      "learning_rate": 0.00015435762424593783,
      "loss": 1.0866,
      "step": 23510
    },
    {
      "epoch": 0.9529790725471526,
      "grad_norm": 0.6223931312561035,
      "learning_rate": 0.0001543219803882079,
      "loss": 1.1057,
      "step": 23520
    },
    {
      "epoch": 0.953384250724256,
      "grad_norm": 0.5249074101448059,
      "learning_rate": 0.00015428632673717297,
      "loss": 1.0833,
      "step": 23530
    },
    {
      "epoch": 0.9537894289013594,
      "grad_norm": 0.5982576608657837,
      "learning_rate": 0.00015425066329926082,
      "loss": 1.0998,
      "step": 23540
    },
    {
      "epoch": 0.9541946070784627,
      "grad_norm": 0.5948882699012756,
      "learning_rate": 0.00015421499008090093,
      "loss": 1.1443,
      "step": 23550
    },
    {
      "epoch": 0.9545997852555661,
      "grad_norm": 0.5861117839813232,
      "learning_rate": 0.00015417930708852456,
      "loss": 1.1087,
      "step": 23560
    },
    {
      "epoch": 0.9550049634326695,
      "grad_norm": 0.6489322781562805,
      "learning_rate": 0.00015414361432856475,
      "loss": 1.124,
      "step": 23570
    },
    {
      "epoch": 0.9554101416097729,
      "grad_norm": 0.5419610142707825,
      "learning_rate": 0.00015410791180745626,
      "loss": 1.1626,
      "step": 23580
    },
    {
      "epoch": 0.9558153197868763,
      "grad_norm": 0.5806791186332703,
      "learning_rate": 0.0001540721995316356,
      "loss": 1.1303,
      "step": 23590
    },
    {
      "epoch": 0.9562204979639797,
      "grad_norm": 0.5579928159713745,
      "learning_rate": 0.00015403647750754113,
      "loss": 1.0835,
      "step": 23600
    },
    {
      "epoch": 0.9566256761410831,
      "grad_norm": 0.5264884829521179,
      "learning_rate": 0.00015400074574161291,
      "loss": 1.1517,
      "step": 23610
    },
    {
      "epoch": 0.9570308543181865,
      "grad_norm": 0.5743647217750549,
      "learning_rate": 0.0001539650042402927,
      "loss": 1.1343,
      "step": 23620
    },
    {
      "epoch": 0.9574360324952899,
      "grad_norm": 0.5993672609329224,
      "learning_rate": 0.00015392925301002411,
      "loss": 1.1279,
      "step": 23630
    },
    {
      "epoch": 0.9578412106723931,
      "grad_norm": 0.5674871802330017,
      "learning_rate": 0.00015389349205725242,
      "loss": 1.1825,
      "step": 23640
    },
    {
      "epoch": 0.9582463888494965,
      "grad_norm": 0.5991116166114807,
      "learning_rate": 0.00015385772138842477,
      "loss": 1.111,
      "step": 23650
    },
    {
      "epoch": 0.9586515670265999,
      "grad_norm": 0.5891130566596985,
      "learning_rate": 0.00015382194100998996,
      "loss": 1.1534,
      "step": 23660
    },
    {
      "epoch": 0.9590567452037033,
      "grad_norm": 0.6052838563919067,
      "learning_rate": 0.00015378615092839852,
      "loss": 1.1085,
      "step": 23670
    },
    {
      "epoch": 0.9594619233808067,
      "grad_norm": 0.8432464003562927,
      "learning_rate": 0.00015375035115010283,
      "loss": 1.1314,
      "step": 23680
    },
    {
      "epoch": 0.9598671015579101,
      "grad_norm": 0.5275864005088806,
      "learning_rate": 0.00015371454168155695,
      "loss": 1.1329,
      "step": 23690
    },
    {
      "epoch": 0.9602722797350135,
      "grad_norm": 0.5424835085868835,
      "learning_rate": 0.00015367872252921672,
      "loss": 1.1052,
      "step": 23700
    },
    {
      "epoch": 0.9606774579121169,
      "grad_norm": 0.5547723770141602,
      "learning_rate": 0.00015364289369953967,
      "loss": 1.1182,
      "step": 23710
    },
    {
      "epoch": 0.9610826360892203,
      "grad_norm": 0.5749952793121338,
      "learning_rate": 0.00015360705519898516,
      "loss": 1.1231,
      "step": 23720
    },
    {
      "epoch": 0.9614878142663236,
      "grad_norm": 0.5436679720878601,
      "learning_rate": 0.0001535712070340142,
      "loss": 1.1297,
      "step": 23730
    },
    {
      "epoch": 0.961892992443427,
      "grad_norm": 0.5528646111488342,
      "learning_rate": 0.0001535353492110896,
      "loss": 1.1516,
      "step": 23740
    },
    {
      "epoch": 0.9622981706205304,
      "grad_norm": 0.6077340245246887,
      "learning_rate": 0.00015349948173667594,
      "loss": 1.1543,
      "step": 23750
    },
    {
      "epoch": 0.9627033487976338,
      "grad_norm": 0.589695394039154,
      "learning_rate": 0.00015346360461723944,
      "loss": 1.1346,
      "step": 23760
    },
    {
      "epoch": 0.9631085269747371,
      "grad_norm": 0.5783637166023254,
      "learning_rate": 0.00015342771785924815,
      "loss": 1.1131,
      "step": 23770
    },
    {
      "epoch": 0.9635137051518405,
      "grad_norm": 0.5403435826301575,
      "learning_rate": 0.00015339182146917183,
      "loss": 1.1145,
      "step": 23780
    },
    {
      "epoch": 0.9639188833289439,
      "grad_norm": 0.5586273670196533,
      "learning_rate": 0.000153355915453482,
      "loss": 1.1089,
      "step": 23790
    },
    {
      "epoch": 0.9643240615060473,
      "grad_norm": 0.5466795563697815,
      "learning_rate": 0.00015331999981865177,
      "loss": 1.1625,
      "step": 23800
    },
    {
      "epoch": 0.9647292396831507,
      "grad_norm": 0.5802621841430664,
      "learning_rate": 0.00015328407457115625,
      "loss": 1.1085,
      "step": 23810
    },
    {
      "epoch": 0.965134417860254,
      "grad_norm": 0.5732377767562866,
      "learning_rate": 0.00015324813971747206,
      "loss": 1.119,
      "step": 23820
    },
    {
      "epoch": 0.9655395960373574,
      "grad_norm": 0.5200895667076111,
      "learning_rate": 0.00015321219526407765,
      "loss": 1.1037,
      "step": 23830
    },
    {
      "epoch": 0.9659447742144608,
      "grad_norm": 0.7075085639953613,
      "learning_rate": 0.00015317624121745314,
      "loss": 1.1677,
      "step": 23840
    },
    {
      "epoch": 0.9663499523915642,
      "grad_norm": 0.5916072130203247,
      "learning_rate": 0.00015314027758408044,
      "loss": 1.1362,
      "step": 23850
    },
    {
      "epoch": 0.9667551305686676,
      "grad_norm": 0.5884680151939392,
      "learning_rate": 0.00015310430437044317,
      "loss": 1.1451,
      "step": 23860
    },
    {
      "epoch": 0.967160308745771,
      "grad_norm": 0.5338191390037537,
      "learning_rate": 0.00015306832158302672,
      "loss": 1.1162,
      "step": 23870
    },
    {
      "epoch": 0.9675654869228744,
      "grad_norm": 0.6077600121498108,
      "learning_rate": 0.00015303232922831805,
      "loss": 1.1305,
      "step": 23880
    },
    {
      "epoch": 0.9679706650999778,
      "grad_norm": 0.5986672639846802,
      "learning_rate": 0.00015299632731280603,
      "loss": 1.1304,
      "step": 23890
    },
    {
      "epoch": 0.9683758432770811,
      "grad_norm": 0.5961667895317078,
      "learning_rate": 0.0001529603158429812,
      "loss": 1.1591,
      "step": 23900
    },
    {
      "epoch": 0.9687810214541844,
      "grad_norm": 0.604279100894928,
      "learning_rate": 0.00015292429482533567,
      "loss": 1.1341,
      "step": 23910
    },
    {
      "epoch": 0.9691861996312878,
      "grad_norm": 0.5562776923179626,
      "learning_rate": 0.00015288826426636354,
      "loss": 1.0892,
      "step": 23920
    },
    {
      "epoch": 0.9695913778083912,
      "grad_norm": 0.6004283428192139,
      "learning_rate": 0.00015285222417256043,
      "loss": 1.1454,
      "step": 23930
    },
    {
      "epoch": 0.9699965559854946,
      "grad_norm": 0.6238961219787598,
      "learning_rate": 0.0001528161745504238,
      "loss": 1.1119,
      "step": 23940
    },
    {
      "epoch": 0.970401734162598,
      "grad_norm": 0.5361995697021484,
      "learning_rate": 0.00015278011540645266,
      "loss": 1.1534,
      "step": 23950
    },
    {
      "epoch": 0.9708069123397014,
      "grad_norm": 0.6620064973831177,
      "learning_rate": 0.00015274404674714794,
      "loss": 1.1736,
      "step": 23960
    },
    {
      "epoch": 0.9712120905168048,
      "grad_norm": 0.5478693842887878,
      "learning_rate": 0.00015270796857901213,
      "loss": 1.1188,
      "step": 23970
    },
    {
      "epoch": 0.9716172686939082,
      "grad_norm": 0.6643903255462646,
      "learning_rate": 0.0001526718809085495,
      "loss": 1.1613,
      "step": 23980
    },
    {
      "epoch": 0.9720224468710116,
      "grad_norm": 0.5835068225860596,
      "learning_rate": 0.00015263578374226605,
      "loss": 1.1011,
      "step": 23990
    },
    {
      "epoch": 0.9724276250481149,
      "grad_norm": 0.543910801410675,
      "learning_rate": 0.00015259967708666948,
      "loss": 1.1426,
      "step": 24000
    },
    {
      "epoch": 0.9728328032252183,
      "grad_norm": 0.7228026390075684,
      "learning_rate": 0.00015256356094826918,
      "loss": 1.1405,
      "step": 24010
    },
    {
      "epoch": 0.9732379814023217,
      "grad_norm": 0.5572428107261658,
      "learning_rate": 0.00015252743533357622,
      "loss": 1.1381,
      "step": 24020
    },
    {
      "epoch": 0.973643159579425,
      "grad_norm": 0.5634073615074158,
      "learning_rate": 0.00015249130024910348,
      "loss": 1.1644,
      "step": 24030
    },
    {
      "epoch": 0.9740483377565284,
      "grad_norm": 0.5541849136352539,
      "learning_rate": 0.00015245515570136547,
      "loss": 1.1086,
      "step": 24040
    },
    {
      "epoch": 0.9744535159336318,
      "grad_norm": 0.5846569538116455,
      "learning_rate": 0.00015241900169687836,
      "loss": 1.1265,
      "step": 24050
    },
    {
      "epoch": 0.9748586941107352,
      "grad_norm": 0.5181740522384644,
      "learning_rate": 0.00015238283824216015,
      "loss": 1.0925,
      "step": 24060
    },
    {
      "epoch": 0.9752638722878386,
      "grad_norm": 0.6438547968864441,
      "learning_rate": 0.00015234666534373042,
      "loss": 1.1146,
      "step": 24070
    },
    {
      "epoch": 0.975669050464942,
      "grad_norm": 0.6909054517745972,
      "learning_rate": 0.00015231048300811062,
      "loss": 1.1378,
      "step": 24080
    },
    {
      "epoch": 0.9760742286420453,
      "grad_norm": 0.6766276359558105,
      "learning_rate": 0.00015227429124182372,
      "loss": 1.1544,
      "step": 24090
    },
    {
      "epoch": 0.9764794068191487,
      "grad_norm": 0.5997416973114014,
      "learning_rate": 0.00015223809005139444,
      "loss": 1.1148,
      "step": 24100
    },
    {
      "epoch": 0.9768845849962521,
      "grad_norm": 0.5866946578025818,
      "learning_rate": 0.00015220187944334926,
      "loss": 1.1272,
      "step": 24110
    },
    {
      "epoch": 0.9772897631733555,
      "grad_norm": 0.5481893420219421,
      "learning_rate": 0.00015216565942421633,
      "loss": 1.1172,
      "step": 24120
    },
    {
      "epoch": 0.9776949413504589,
      "grad_norm": 0.5752227306365967,
      "learning_rate": 0.00015212943000052545,
      "loss": 1.126,
      "step": 24130
    },
    {
      "epoch": 0.9781001195275623,
      "grad_norm": 0.5326637029647827,
      "learning_rate": 0.00015209319117880817,
      "loss": 1.1343,
      "step": 24140
    },
    {
      "epoch": 0.9785052977046657,
      "grad_norm": 0.6091455817222595,
      "learning_rate": 0.00015205694296559777,
      "loss": 1.1169,
      "step": 24150
    },
    {
      "epoch": 0.978910475881769,
      "grad_norm": 0.5976237654685974,
      "learning_rate": 0.00015202068536742907,
      "loss": 1.1499,
      "step": 24160
    },
    {
      "epoch": 0.9793156540588724,
      "grad_norm": 0.5387760400772095,
      "learning_rate": 0.00015198441839083874,
      "loss": 1.1101,
      "step": 24170
    },
    {
      "epoch": 0.9797208322359757,
      "grad_norm": 0.534932553768158,
      "learning_rate": 0.0001519481420423651,
      "loss": 1.1185,
      "step": 24180
    },
    {
      "epoch": 0.9801260104130791,
      "grad_norm": 0.5969175100326538,
      "learning_rate": 0.0001519118563285481,
      "loss": 1.1362,
      "step": 24190
    },
    {
      "epoch": 0.9805311885901825,
      "grad_norm": 0.5573731064796448,
      "learning_rate": 0.00015187556125592945,
      "loss": 1.1236,
      "step": 24200
    },
    {
      "epoch": 0.9809363667672859,
      "grad_norm": 0.5700602531433105,
      "learning_rate": 0.00015183925683105254,
      "loss": 1.1858,
      "step": 24210
    },
    {
      "epoch": 0.9813415449443893,
      "grad_norm": 0.5714423060417175,
      "learning_rate": 0.00015180294306046233,
      "loss": 1.102,
      "step": 24220
    },
    {
      "epoch": 0.9817467231214927,
      "grad_norm": 0.64220130443573,
      "learning_rate": 0.00015176661995070564,
      "loss": 1.1475,
      "step": 24230
    },
    {
      "epoch": 0.9821519012985961,
      "grad_norm": 0.5607056617736816,
      "learning_rate": 0.0001517302875083309,
      "loss": 1.1373,
      "step": 24240
    },
    {
      "epoch": 0.9825570794756995,
      "grad_norm": 0.5717743635177612,
      "learning_rate": 0.00015169394573988818,
      "loss": 1.1151,
      "step": 24250
    },
    {
      "epoch": 0.9829622576528029,
      "grad_norm": 0.6540253758430481,
      "learning_rate": 0.0001516575946519293,
      "loss": 1.1722,
      "step": 24260
    },
    {
      "epoch": 0.9833674358299062,
      "grad_norm": 0.5358489155769348,
      "learning_rate": 0.00015162123425100762,
      "loss": 1.106,
      "step": 24270
    },
    {
      "epoch": 0.9837726140070095,
      "grad_norm": 0.5921744704246521,
      "learning_rate": 0.00015158486454367845,
      "loss": 1.095,
      "step": 24280
    },
    {
      "epoch": 0.9841777921841129,
      "grad_norm": 0.5527446269989014,
      "learning_rate": 0.0001515484855364985,
      "loss": 1.1556,
      "step": 24290
    },
    {
      "epoch": 0.9845829703612163,
      "grad_norm": 0.6107127070426941,
      "learning_rate": 0.00015151209723602627,
      "loss": 1.111,
      "step": 24300
    },
    {
      "epoch": 0.9849881485383197,
      "grad_norm": 0.5697078704833984,
      "learning_rate": 0.00015147569964882201,
      "loss": 1.1187,
      "step": 24310
    },
    {
      "epoch": 0.9853933267154231,
      "grad_norm": 0.5480672121047974,
      "learning_rate": 0.00015143929278144756,
      "loss": 1.1416,
      "step": 24320
    },
    {
      "epoch": 0.9857985048925265,
      "grad_norm": 0.5146967172622681,
      "learning_rate": 0.00015140287664046635,
      "loss": 1.1428,
      "step": 24330
    },
    {
      "epoch": 0.9862036830696299,
      "grad_norm": 0.5591877102851868,
      "learning_rate": 0.00015136645123244366,
      "loss": 1.1182,
      "step": 24340
    },
    {
      "epoch": 0.9866088612467333,
      "grad_norm": 0.5727733373641968,
      "learning_rate": 0.00015133001656394632,
      "loss": 1.1335,
      "step": 24350
    },
    {
      "epoch": 0.9870140394238366,
      "grad_norm": 0.5802597999572754,
      "learning_rate": 0.00015129357264154287,
      "loss": 1.1033,
      "step": 24360
    },
    {
      "epoch": 0.98741921760094,
      "grad_norm": 0.516243577003479,
      "learning_rate": 0.00015125711947180354,
      "loss": 1.1557,
      "step": 24370
    },
    {
      "epoch": 0.9878243957780434,
      "grad_norm": 0.5809158086776733,
      "learning_rate": 0.00015122065706130012,
      "loss": 1.131,
      "step": 24380
    },
    {
      "epoch": 0.9882295739551468,
      "grad_norm": 0.5992233157157898,
      "learning_rate": 0.00015118418541660628,
      "loss": 1.1173,
      "step": 24390
    },
    {
      "epoch": 0.9886347521322502,
      "grad_norm": 0.5913904309272766,
      "learning_rate": 0.0001511477045442971,
      "loss": 1.1452,
      "step": 24400
    },
    {
      "epoch": 0.9890399303093536,
      "grad_norm": 0.6464443802833557,
      "learning_rate": 0.0001511112144509495,
      "loss": 1.1399,
      "step": 24410
    },
    {
      "epoch": 0.989445108486457,
      "grad_norm": 0.5625829100608826,
      "learning_rate": 0.000151074715143142,
      "loss": 1.1274,
      "step": 24420
    },
    {
      "epoch": 0.9898502866635603,
      "grad_norm": 0.6140351891517639,
      "learning_rate": 0.00015103820662745477,
      "loss": 1.0927,
      "step": 24430
    },
    {
      "epoch": 0.9902554648406637,
      "grad_norm": 0.5601906776428223,
      "learning_rate": 0.0001510016889104697,
      "loss": 1.0843,
      "step": 24440
    },
    {
      "epoch": 0.990660643017767,
      "grad_norm": 0.6657971143722534,
      "learning_rate": 0.00015096516199877025,
      "loss": 1.1203,
      "step": 24450
    },
    {
      "epoch": 0.9910658211948704,
      "grad_norm": 0.5688597559928894,
      "learning_rate": 0.0001509286258989416,
      "loss": 1.1526,
      "step": 24460
    },
    {
      "epoch": 0.9914709993719738,
      "grad_norm": 0.5483949184417725,
      "learning_rate": 0.0001508920806175706,
      "loss": 1.129,
      "step": 24470
    },
    {
      "epoch": 0.9918761775490772,
      "grad_norm": 0.5413434505462646,
      "learning_rate": 0.0001508555261612457,
      "loss": 1.1165,
      "step": 24480
    },
    {
      "epoch": 0.9922813557261806,
      "grad_norm": 0.580647885799408,
      "learning_rate": 0.00015081896253655704,
      "loss": 1.0991,
      "step": 24490
    },
    {
      "epoch": 0.992686533903284,
      "grad_norm": 0.5774683356285095,
      "learning_rate": 0.00015078238975009634,
      "loss": 1.1498,
      "step": 24500
    },
    {
      "epoch": 0.9930917120803874,
      "grad_norm": 0.6175000071525574,
      "learning_rate": 0.00015074580780845715,
      "loss": 1.1081,
      "step": 24510
    },
    {
      "epoch": 0.9934968902574908,
      "grad_norm": 0.5812476873397827,
      "learning_rate": 0.00015070921671823451,
      "loss": 1.1484,
      "step": 24520
    },
    {
      "epoch": 0.9939020684345942,
      "grad_norm": 0.5254901647567749,
      "learning_rate": 0.0001506726164860251,
      "loss": 1.1178,
      "step": 24530
    },
    {
      "epoch": 0.9943072466116974,
      "grad_norm": 0.5905852913856506,
      "learning_rate": 0.00015063600711842735,
      "loss": 1.1078,
      "step": 24540
    },
    {
      "epoch": 0.9947124247888008,
      "grad_norm": 0.561536967754364,
      "learning_rate": 0.00015059938862204127,
      "loss": 1.1261,
      "step": 24550
    },
    {
      "epoch": 0.9951176029659042,
      "grad_norm": 0.5274717211723328,
      "learning_rate": 0.00015056276100346855,
      "loss": 1.1076,
      "step": 24560
    },
    {
      "epoch": 0.9955227811430076,
      "grad_norm": 0.5821682214736938,
      "learning_rate": 0.00015052612426931256,
      "loss": 1.1307,
      "step": 24570
    },
    {
      "epoch": 0.995927959320111,
      "grad_norm": 0.576477587223053,
      "learning_rate": 0.00015048947842617817,
      "loss": 1.1133,
      "step": 24580
    },
    {
      "epoch": 0.9963331374972144,
      "grad_norm": 0.5961494445800781,
      "learning_rate": 0.00015045282348067205,
      "loss": 1.1544,
      "step": 24590
    },
    {
      "epoch": 0.9967383156743178,
      "grad_norm": 0.5334819555282593,
      "learning_rate": 0.0001504161594394024,
      "loss": 1.1381,
      "step": 24600
    },
    {
      "epoch": 0.9971434938514212,
      "grad_norm": 0.5655900239944458,
      "learning_rate": 0.0001503794863089792,
      "loss": 1.1122,
      "step": 24610
    },
    {
      "epoch": 0.9975486720285246,
      "grad_norm": 0.5691961050033569,
      "learning_rate": 0.00015034280409601385,
      "loss": 1.1193,
      "step": 24620
    },
    {
      "epoch": 0.9979538502056279,
      "grad_norm": 0.6199621558189392,
      "learning_rate": 0.00015030611280711958,
      "loss": 1.141,
      "step": 24630
    },
    {
      "epoch": 0.9983590283827313,
      "grad_norm": 0.5674569010734558,
      "learning_rate": 0.0001502694124489112,
      "loss": 1.1523,
      "step": 24640
    },
    {
      "epoch": 0.9987642065598347,
      "grad_norm": 0.6513556838035583,
      "learning_rate": 0.00015023270302800512,
      "loss": 1.1023,
      "step": 24650
    },
    {
      "epoch": 0.9991693847369381,
      "grad_norm": 0.5862869620323181,
      "learning_rate": 0.00015019598455101943,
      "loss": 1.1177,
      "step": 24660
    },
    {
      "epoch": 0.9995745629140415,
      "grad_norm": 0.585810124874115,
      "learning_rate": 0.00015015925702457377,
      "loss": 1.1331,
      "step": 24670
    },
    {
      "epoch": 0.9999797410911448,
      "grad_norm": 0.5561410784721375,
      "learning_rate": 0.00015012252045528953,
      "loss": 1.128,
      "step": 24680
    },
    {
      "epoch": 0.9999797410911448,
      "eval_loss": 1.1248676776885986,
      "eval_runtime": 952.6485,
      "eval_samples_per_second": 11.515,
      "eval_steps_per_second": 1.44,
      "step": 24680
    },
    {
      "epoch": 1.0003849192682481,
      "grad_norm": 0.6232626438140869,
      "learning_rate": 0.00015008577484978966,
      "loss": 1.1227,
      "step": 24690
    },
    {
      "epoch": 1.0007900974453516,
      "grad_norm": 0.5889710187911987,
      "learning_rate": 0.00015004902021469878,
      "loss": 1.1479,
      "step": 24700
    },
    {
      "epoch": 1.001195275622455,
      "grad_norm": 0.6364986896514893,
      "learning_rate": 0.00015001225655664305,
      "loss": 1.0924,
      "step": 24710
    },
    {
      "epoch": 1.0016004537995584,
      "grad_norm": 0.5676513910293579,
      "learning_rate": 0.00014997548388225032,
      "loss": 1.0714,
      "step": 24720
    },
    {
      "epoch": 1.0020056319766617,
      "grad_norm": 0.5715664625167847,
      "learning_rate": 0.00014993870219815014,
      "loss": 1.0948,
      "step": 24730
    },
    {
      "epoch": 1.0024108101537652,
      "grad_norm": 0.5738197565078735,
      "learning_rate": 0.00014990191151097353,
      "loss": 1.1006,
      "step": 24740
    },
    {
      "epoch": 1.0028159883308685,
      "grad_norm": 0.6074545383453369,
      "learning_rate": 0.0001498651118273532,
      "loss": 1.1047,
      "step": 24750
    },
    {
      "epoch": 1.0032211665079718,
      "grad_norm": 0.5813114643096924,
      "learning_rate": 0.00014982830315392358,
      "loss": 1.0963,
      "step": 24760
    },
    {
      "epoch": 1.0036263446850753,
      "grad_norm": 0.6005133390426636,
      "learning_rate": 0.00014979148549732052,
      "loss": 1.075,
      "step": 24770
    },
    {
      "epoch": 1.0040315228621786,
      "grad_norm": 0.5829113721847534,
      "learning_rate": 0.0001497546588641817,
      "loss": 1.0987,
      "step": 24780
    },
    {
      "epoch": 1.004436701039282,
      "grad_norm": 0.6546384692192078,
      "learning_rate": 0.00014971782326114622,
      "loss": 1.0866,
      "step": 24790
    },
    {
      "epoch": 1.0048418792163853,
      "grad_norm": 0.5699377655982971,
      "learning_rate": 0.00014968097869485497,
      "loss": 1.0918,
      "step": 24800
    },
    {
      "epoch": 1.0052470573934889,
      "grad_norm": 0.5743865370750427,
      "learning_rate": 0.00014964412517195038,
      "loss": 1.1209,
      "step": 24810
    },
    {
      "epoch": 1.0056522355705921,
      "grad_norm": 0.5655612349510193,
      "learning_rate": 0.00014960726269907645,
      "loss": 1.102,
      "step": 24820
    },
    {
      "epoch": 1.0060574137476956,
      "grad_norm": 0.6294947266578674,
      "learning_rate": 0.00014957039128287892,
      "loss": 1.0906,
      "step": 24830
    },
    {
      "epoch": 1.006462591924799,
      "grad_norm": 0.5854537487030029,
      "learning_rate": 0.00014953351093000494,
      "loss": 1.1367,
      "step": 24840
    },
    {
      "epoch": 1.0068677701019022,
      "grad_norm": 0.5814386010169983,
      "learning_rate": 0.00014949662164710352,
      "loss": 1.1018,
      "step": 24850
    },
    {
      "epoch": 1.0072729482790057,
      "grad_norm": 0.6016951203346252,
      "learning_rate": 0.00014945972344082507,
      "loss": 1.0876,
      "step": 24860
    },
    {
      "epoch": 1.007678126456109,
      "grad_norm": 0.5629173517227173,
      "learning_rate": 0.00014942281631782174,
      "loss": 1.083,
      "step": 24870
    },
    {
      "epoch": 1.0080833046332125,
      "grad_norm": 0.6297256350517273,
      "learning_rate": 0.0001493859002847472,
      "loss": 1.0865,
      "step": 24880
    },
    {
      "epoch": 1.0084884828103158,
      "grad_norm": 0.6169393062591553,
      "learning_rate": 0.0001493489753482568,
      "loss": 1.064,
      "step": 24890
    },
    {
      "epoch": 1.0088936609874193,
      "grad_norm": 0.578269898891449,
      "learning_rate": 0.00014931204151500747,
      "loss": 1.0619,
      "step": 24900
    },
    {
      "epoch": 1.0092988391645226,
      "grad_norm": 0.6479613184928894,
      "learning_rate": 0.0001492750987916577,
      "loss": 1.1266,
      "step": 24910
    },
    {
      "epoch": 1.009704017341626,
      "grad_norm": 0.5299257636070251,
      "learning_rate": 0.00014923814718486762,
      "loss": 1.0488,
      "step": 24920
    },
    {
      "epoch": 1.0101091955187294,
      "grad_norm": 0.7489888072013855,
      "learning_rate": 0.000149201186701299,
      "loss": 1.0804,
      "step": 24930
    },
    {
      "epoch": 1.0105143736958326,
      "grad_norm": 0.665690541267395,
      "learning_rate": 0.0001491642173476151,
      "loss": 1.1046,
      "step": 24940
    },
    {
      "epoch": 1.0109195518729361,
      "grad_norm": 0.6294705271720886,
      "learning_rate": 0.00014912723913048092,
      "loss": 1.0886,
      "step": 24950
    },
    {
      "epoch": 1.0113247300500394,
      "grad_norm": 0.646873950958252,
      "learning_rate": 0.00014909025205656298,
      "loss": 1.114,
      "step": 24960
    },
    {
      "epoch": 1.011729908227143,
      "grad_norm": 0.5919949412345886,
      "learning_rate": 0.00014905325613252937,
      "loss": 1.0991,
      "step": 24970
    },
    {
      "epoch": 1.0121350864042462,
      "grad_norm": 0.6285662651062012,
      "learning_rate": 0.00014901625136504978,
      "loss": 1.0766,
      "step": 24980
    },
    {
      "epoch": 1.0125402645813497,
      "grad_norm": 0.5448628067970276,
      "learning_rate": 0.00014897923776079563,
      "loss": 1.1046,
      "step": 24990
    },
    {
      "epoch": 1.012945442758453,
      "grad_norm": 0.5806989073753357,
      "learning_rate": 0.00014894221532643973,
      "loss": 1.0758,
      "step": 25000
    },
    {
      "epoch": 1.0133506209355565,
      "grad_norm": 0.5501072406768799,
      "learning_rate": 0.00014890518406865662,
      "loss": 1.0946,
      "step": 25010
    },
    {
      "epoch": 1.0137557991126598,
      "grad_norm": 0.6766510605812073,
      "learning_rate": 0.00014886814399412242,
      "loss": 1.0692,
      "step": 25020
    },
    {
      "epoch": 1.014160977289763,
      "grad_norm": 0.547247052192688,
      "learning_rate": 0.0001488310951095147,
      "loss": 1.0885,
      "step": 25030
    },
    {
      "epoch": 1.0145661554668666,
      "grad_norm": 0.5400190949440002,
      "learning_rate": 0.00014879403742151283,
      "loss": 1.0978,
      "step": 25040
    },
    {
      "epoch": 1.0149713336439699,
      "grad_norm": 0.5550058484077454,
      "learning_rate": 0.00014875697093679765,
      "loss": 1.0491,
      "step": 25050
    },
    {
      "epoch": 1.0153765118210734,
      "grad_norm": 0.5529043078422546,
      "learning_rate": 0.00014871989566205156,
      "loss": 1.0968,
      "step": 25060
    },
    {
      "epoch": 1.0157816899981766,
      "grad_norm": 0.5977621078491211,
      "learning_rate": 0.00014868281160395862,
      "loss": 1.0892,
      "step": 25070
    },
    {
      "epoch": 1.0161868681752801,
      "grad_norm": 0.6497211456298828,
      "learning_rate": 0.0001486457187692044,
      "loss": 1.0475,
      "step": 25080
    },
    {
      "epoch": 1.0165920463523834,
      "grad_norm": 0.5151035189628601,
      "learning_rate": 0.00014860861716447615,
      "loss": 1.1058,
      "step": 25090
    },
    {
      "epoch": 1.016997224529487,
      "grad_norm": 0.5852901935577393,
      "learning_rate": 0.00014857150679646258,
      "loss": 1.1073,
      "step": 25100
    },
    {
      "epoch": 1.0174024027065902,
      "grad_norm": 0.6446422338485718,
      "learning_rate": 0.00014853438767185412,
      "loss": 1.1101,
      "step": 25110
    },
    {
      "epoch": 1.0178075808836935,
      "grad_norm": 0.6521576642990112,
      "learning_rate": 0.00014849725979734259,
      "loss": 1.1074,
      "step": 25120
    },
    {
      "epoch": 1.018212759060797,
      "grad_norm": 0.6277870535850525,
      "learning_rate": 0.00014846012317962158,
      "loss": 1.101,
      "step": 25130
    },
    {
      "epoch": 1.0186179372379003,
      "grad_norm": 0.6501975655555725,
      "learning_rate": 0.0001484229778253862,
      "loss": 1.0798,
      "step": 25140
    },
    {
      "epoch": 1.0190231154150038,
      "grad_norm": 0.5206642150878906,
      "learning_rate": 0.000148385823741333,
      "loss": 1.1004,
      "step": 25150
    },
    {
      "epoch": 1.019428293592107,
      "grad_norm": 0.6591925024986267,
      "learning_rate": 0.0001483486609341603,
      "loss": 1.1,
      "step": 25160
    },
    {
      "epoch": 1.0198334717692106,
      "grad_norm": 0.5795564651489258,
      "learning_rate": 0.00014831148941056786,
      "loss": 1.106,
      "step": 25170
    },
    {
      "epoch": 1.0202386499463139,
      "grad_norm": 0.5996028780937195,
      "learning_rate": 0.00014827430917725712,
      "loss": 1.0967,
      "step": 25180
    },
    {
      "epoch": 1.0206438281234174,
      "grad_norm": 0.5471130013465881,
      "learning_rate": 0.00014823712024093094,
      "loss": 1.057,
      "step": 25190
    },
    {
      "epoch": 1.0210490063005206,
      "grad_norm": 0.567160427570343,
      "learning_rate": 0.0001481999226082939,
      "loss": 1.1012,
      "step": 25200
    },
    {
      "epoch": 1.021454184477624,
      "grad_norm": 0.5555440187454224,
      "learning_rate": 0.00014816271628605206,
      "loss": 1.0787,
      "step": 25210
    },
    {
      "epoch": 1.0218593626547274,
      "grad_norm": 0.604796826839447,
      "learning_rate": 0.00014812550128091308,
      "loss": 1.0429,
      "step": 25220
    },
    {
      "epoch": 1.0222645408318307,
      "grad_norm": 0.5330144166946411,
      "learning_rate": 0.00014808827759958614,
      "loss": 1.1114,
      "step": 25230
    },
    {
      "epoch": 1.0226697190089342,
      "grad_norm": 0.6424878239631653,
      "learning_rate": 0.00014805104524878208,
      "loss": 1.1012,
      "step": 25240
    },
    {
      "epoch": 1.0230748971860375,
      "grad_norm": 0.5790878534317017,
      "learning_rate": 0.00014801380423521324,
      "loss": 1.1548,
      "step": 25250
    },
    {
      "epoch": 1.023480075363141,
      "grad_norm": 0.6536170840263367,
      "learning_rate": 0.00014797655456559345,
      "loss": 1.1264,
      "step": 25260
    },
    {
      "epoch": 1.0238852535402443,
      "grad_norm": 0.6013097763061523,
      "learning_rate": 0.00014793929624663826,
      "loss": 1.0727,
      "step": 25270
    },
    {
      "epoch": 1.0242904317173478,
      "grad_norm": 0.6046758890151978,
      "learning_rate": 0.00014790202928506466,
      "loss": 1.0755,
      "step": 25280
    },
    {
      "epoch": 1.024695609894451,
      "grad_norm": 0.5363589525222778,
      "learning_rate": 0.00014786475368759117,
      "loss": 1.1015,
      "step": 25290
    },
    {
      "epoch": 1.0251007880715544,
      "grad_norm": 0.5781187415122986,
      "learning_rate": 0.00014782746946093807,
      "loss": 1.1453,
      "step": 25300
    },
    {
      "epoch": 1.0255059662486579,
      "grad_norm": 0.6460751295089722,
      "learning_rate": 0.0001477901766118269,
      "loss": 1.1018,
      "step": 25310
    },
    {
      "epoch": 1.0259111444257611,
      "grad_norm": 0.5941773653030396,
      "learning_rate": 0.00014775287514698105,
      "loss": 1.1322,
      "step": 25320
    },
    {
      "epoch": 1.0263163226028647,
      "grad_norm": 0.6228278279304504,
      "learning_rate": 0.00014771556507312523,
      "loss": 1.1334,
      "step": 25330
    },
    {
      "epoch": 1.026721500779968,
      "grad_norm": 0.5537286996841431,
      "learning_rate": 0.00014767824639698581,
      "loss": 1.1012,
      "step": 25340
    },
    {
      "epoch": 1.0271266789570714,
      "grad_norm": 0.6210373044013977,
      "learning_rate": 0.00014764091912529073,
      "loss": 1.0625,
      "step": 25350
    },
    {
      "epoch": 1.0275318571341747,
      "grad_norm": 0.5313713550567627,
      "learning_rate": 0.00014760358326476944,
      "loss": 1.0787,
      "step": 25360
    },
    {
      "epoch": 1.0279370353112782,
      "grad_norm": 0.6662773489952087,
      "learning_rate": 0.0001475662388221529,
      "loss": 1.0864,
      "step": 25370
    },
    {
      "epoch": 1.0283422134883815,
      "grad_norm": 0.5555450916290283,
      "learning_rate": 0.00014752888580417373,
      "loss": 1.0942,
      "step": 25380
    },
    {
      "epoch": 1.0287473916654848,
      "grad_norm": 0.708221971988678,
      "learning_rate": 0.00014749152421756595,
      "loss": 1.1055,
      "step": 25390
    },
    {
      "epoch": 1.0291525698425883,
      "grad_norm": 0.6691581606864929,
      "learning_rate": 0.00014745415406906525,
      "loss": 1.0722,
      "step": 25400
    },
    {
      "epoch": 1.0295577480196916,
      "grad_norm": 0.6159145832061768,
      "learning_rate": 0.0001474167753654088,
      "loss": 1.0917,
      "step": 25410
    },
    {
      "epoch": 1.029962926196795,
      "grad_norm": 0.6291682124137878,
      "learning_rate": 0.00014737938811333533,
      "loss": 1.1239,
      "step": 25420
    },
    {
      "epoch": 1.0303681043738984,
      "grad_norm": 0.5817980766296387,
      "learning_rate": 0.00014734199231958512,
      "loss": 1.0591,
      "step": 25430
    },
    {
      "epoch": 1.0307732825510019,
      "grad_norm": 0.6223018765449524,
      "learning_rate": 0.00014730458799089999,
      "loss": 1.1068,
      "step": 25440
    },
    {
      "epoch": 1.0311784607281052,
      "grad_norm": 0.6332880258560181,
      "learning_rate": 0.00014726717513402325,
      "loss": 1.0984,
      "step": 25450
    },
    {
      "epoch": 1.0315836389052087,
      "grad_norm": 0.5692764520645142,
      "learning_rate": 0.00014722975375569978,
      "loss": 1.0865,
      "step": 25460
    },
    {
      "epoch": 1.031988817082312,
      "grad_norm": 0.594914972782135,
      "learning_rate": 0.00014719232386267608,
      "loss": 1.0973,
      "step": 25470
    },
    {
      "epoch": 1.0323939952594152,
      "grad_norm": 0.626420259475708,
      "learning_rate": 0.0001471548854617,
      "loss": 1.1124,
      "step": 25480
    },
    {
      "epoch": 1.0327991734365187,
      "grad_norm": 0.7143211960792542,
      "learning_rate": 0.00014711743855952113,
      "loss": 1.0816,
      "step": 25490
    },
    {
      "epoch": 1.033204351613622,
      "grad_norm": 0.5982559323310852,
      "learning_rate": 0.00014707998316289042,
      "loss": 1.11,
      "step": 25500
    },
    {
      "epoch": 1.0336095297907255,
      "grad_norm": 0.5567126870155334,
      "learning_rate": 0.00014704251927856045,
      "loss": 1.1109,
      "step": 25510
    },
    {
      "epoch": 1.0340147079678288,
      "grad_norm": 0.6198965311050415,
      "learning_rate": 0.00014700504691328532,
      "loss": 1.0947,
      "step": 25520
    },
    {
      "epoch": 1.0344198861449323,
      "grad_norm": 0.618492841720581,
      "learning_rate": 0.0001469675660738206,
      "loss": 1.1167,
      "step": 25530
    },
    {
      "epoch": 1.0348250643220356,
      "grad_norm": 0.6697366237640381,
      "learning_rate": 0.00014693007676692347,
      "loss": 1.0689,
      "step": 25540
    },
    {
      "epoch": 1.035230242499139,
      "grad_norm": 0.5943389534950256,
      "learning_rate": 0.0001468925789993526,
      "loss": 1.1202,
      "step": 25550
    },
    {
      "epoch": 1.0356354206762424,
      "grad_norm": 0.6378085613250732,
      "learning_rate": 0.00014685507277786817,
      "loss": 1.1308,
      "step": 25560
    },
    {
      "epoch": 1.0360405988533459,
      "grad_norm": 0.537905216217041,
      "learning_rate": 0.00014681755810923186,
      "loss": 1.0879,
      "step": 25570
    },
    {
      "epoch": 1.0364457770304492,
      "grad_norm": 0.5340473055839539,
      "learning_rate": 0.00014678003500020697,
      "loss": 1.0861,
      "step": 25580
    },
    {
      "epoch": 1.0368509552075524,
      "grad_norm": 0.5431529879570007,
      "learning_rate": 0.00014674250345755825,
      "loss": 1.0898,
      "step": 25590
    },
    {
      "epoch": 1.037256133384656,
      "grad_norm": 0.7042612433433533,
      "learning_rate": 0.00014670496348805195,
      "loss": 1.0851,
      "step": 25600
    },
    {
      "epoch": 1.0376613115617592,
      "grad_norm": 0.6087439060211182,
      "learning_rate": 0.00014666741509845592,
      "loss": 1.1636,
      "step": 25610
    },
    {
      "epoch": 1.0380664897388627,
      "grad_norm": 0.6178156137466431,
      "learning_rate": 0.0001466298582955394,
      "loss": 1.1104,
      "step": 25620
    },
    {
      "epoch": 1.038471667915966,
      "grad_norm": 0.5544847249984741,
      "learning_rate": 0.0001465922930860733,
      "loss": 1.0877,
      "step": 25630
    },
    {
      "epoch": 1.0388768460930695,
      "grad_norm": 0.6292670369148254,
      "learning_rate": 0.00014655471947683,
      "loss": 1.1022,
      "step": 25640
    },
    {
      "epoch": 1.0392820242701728,
      "grad_norm": 0.5986620187759399,
      "learning_rate": 0.00014651713747458324,
      "loss": 1.076,
      "step": 25650
    },
    {
      "epoch": 1.039687202447276,
      "grad_norm": 0.5988556146621704,
      "learning_rate": 0.00014647954708610852,
      "loss": 1.0886,
      "step": 25660
    },
    {
      "epoch": 1.0400923806243796,
      "grad_norm": 0.5785689949989319,
      "learning_rate": 0.00014644194831818266,
      "loss": 1.1109,
      "step": 25670
    },
    {
      "epoch": 1.0404975588014829,
      "grad_norm": 0.5992876887321472,
      "learning_rate": 0.0001464043411775841,
      "loss": 1.1082,
      "step": 25680
    },
    {
      "epoch": 1.0409027369785864,
      "grad_norm": 0.6243312954902649,
      "learning_rate": 0.00014636672567109273,
      "loss": 1.1239,
      "step": 25690
    },
    {
      "epoch": 1.0413079151556897,
      "grad_norm": 0.6140483617782593,
      "learning_rate": 0.00014632910180549,
      "loss": 1.0722,
      "step": 25700
    },
    {
      "epoch": 1.0417130933327932,
      "grad_norm": 0.5540391802787781,
      "learning_rate": 0.00014629146958755878,
      "loss": 1.0817,
      "step": 25710
    },
    {
      "epoch": 1.0421182715098964,
      "grad_norm": 0.5935832858085632,
      "learning_rate": 0.00014625382902408356,
      "loss": 1.1187,
      "step": 25720
    },
    {
      "epoch": 1.042523449687,
      "grad_norm": 0.5896474719047546,
      "learning_rate": 0.00014621618012185023,
      "loss": 1.0872,
      "step": 25730
    },
    {
      "epoch": 1.0429286278641032,
      "grad_norm": 0.5500726699829102,
      "learning_rate": 0.00014617852288764625,
      "loss": 1.1135,
      "step": 25740
    },
    {
      "epoch": 1.0433338060412067,
      "grad_norm": 0.6229005455970764,
      "learning_rate": 0.00014614085732826058,
      "loss": 1.0861,
      "step": 25750
    },
    {
      "epoch": 1.04373898421831,
      "grad_norm": 0.5866906642913818,
      "learning_rate": 0.0001461031834504836,
      "loss": 1.1099,
      "step": 25760
    },
    {
      "epoch": 1.0441441623954133,
      "grad_norm": 0.6447248458862305,
      "learning_rate": 0.00014606550126110737,
      "loss": 1.1172,
      "step": 25770
    },
    {
      "epoch": 1.0445493405725168,
      "grad_norm": 0.5524841547012329,
      "learning_rate": 0.0001460278107669252,
      "loss": 1.1127,
      "step": 25780
    },
    {
      "epoch": 1.04495451874962,
      "grad_norm": 0.6079192757606506,
      "learning_rate": 0.00014599011197473214,
      "loss": 1.0822,
      "step": 25790
    },
    {
      "epoch": 1.0453596969267236,
      "grad_norm": 0.5664893984794617,
      "learning_rate": 0.00014595240489132454,
      "loss": 1.0855,
      "step": 25800
    },
    {
      "epoch": 1.0457648751038269,
      "grad_norm": 0.598901629447937,
      "learning_rate": 0.0001459146895235004,
      "loss": 1.0724,
      "step": 25810
    },
    {
      "epoch": 1.0461700532809304,
      "grad_norm": 0.622755765914917,
      "learning_rate": 0.00014587696587805904,
      "loss": 1.1339,
      "step": 25820
    },
    {
      "epoch": 1.0465752314580337,
      "grad_norm": 0.6206207871437073,
      "learning_rate": 0.00014583923396180147,
      "loss": 1.0763,
      "step": 25830
    },
    {
      "epoch": 1.046980409635137,
      "grad_norm": 0.618492603302002,
      "learning_rate": 0.0001458014937815301,
      "loss": 1.05,
      "step": 25840
    },
    {
      "epoch": 1.0473855878122404,
      "grad_norm": 0.6017782688140869,
      "learning_rate": 0.00014576374534404875,
      "loss": 1.0853,
      "step": 25850
    },
    {
      "epoch": 1.0477907659893437,
      "grad_norm": 0.6504418849945068,
      "learning_rate": 0.00014572598865616288,
      "loss": 1.0715,
      "step": 25860
    },
    {
      "epoch": 1.0481959441664472,
      "grad_norm": 0.5810089707374573,
      "learning_rate": 0.0001456882237246793,
      "loss": 1.1039,
      "step": 25870
    },
    {
      "epoch": 1.0486011223435505,
      "grad_norm": 0.6084761023521423,
      "learning_rate": 0.00014565045055640638,
      "loss": 1.1515,
      "step": 25880
    },
    {
      "epoch": 1.049006300520654,
      "grad_norm": 0.6219791769981384,
      "learning_rate": 0.00014561266915815398,
      "loss": 1.0934,
      "step": 25890
    },
    {
      "epoch": 1.0494114786977573,
      "grad_norm": 0.6240952610969543,
      "learning_rate": 0.00014557487953673347,
      "loss": 1.1141,
      "step": 25900
    },
    {
      "epoch": 1.0498166568748608,
      "grad_norm": 0.6455026268959045,
      "learning_rate": 0.00014553708169895758,
      "loss": 1.1092,
      "step": 25910
    },
    {
      "epoch": 1.050221835051964,
      "grad_norm": 0.626740574836731,
      "learning_rate": 0.00014549927565164064,
      "loss": 1.1081,
      "step": 25920
    },
    {
      "epoch": 1.0506270132290676,
      "grad_norm": 0.5690827369689941,
      "learning_rate": 0.00014546146140159843,
      "loss": 1.0527,
      "step": 25930
    },
    {
      "epoch": 1.0510321914061709,
      "grad_norm": 0.5910043120384216,
      "learning_rate": 0.00014542363895564818,
      "loss": 1.0913,
      "step": 25940
    },
    {
      "epoch": 1.0514373695832742,
      "grad_norm": 0.6275461316108704,
      "learning_rate": 0.0001453858083206086,
      "loss": 1.1269,
      "step": 25950
    },
    {
      "epoch": 1.0518425477603777,
      "grad_norm": 0.6067894697189331,
      "learning_rate": 0.0001453479695032999,
      "loss": 1.1096,
      "step": 25960
    },
    {
      "epoch": 1.052247725937481,
      "grad_norm": 1.1135737895965576,
      "learning_rate": 0.0001453101225105438,
      "loss": 1.1177,
      "step": 25970
    },
    {
      "epoch": 1.0526529041145845,
      "grad_norm": 0.5861806869506836,
      "learning_rate": 0.00014527226734916345,
      "loss": 1.0862,
      "step": 25980
    },
    {
      "epoch": 1.0530580822916877,
      "grad_norm": 0.5264201164245605,
      "learning_rate": 0.0001452344040259834,
      "loss": 1.1091,
      "step": 25990
    },
    {
      "epoch": 1.0534632604687912,
      "grad_norm": 0.6767321825027466,
      "learning_rate": 0.00014519653254782983,
      "loss": 1.1286,
      "step": 26000
    },
    {
      "epoch": 1.0538684386458945,
      "grad_norm": 0.5980548858642578,
      "learning_rate": 0.00014515865292153029,
      "loss": 1.0732,
      "step": 26010
    },
    {
      "epoch": 1.0542736168229978,
      "grad_norm": 0.6148680448532104,
      "learning_rate": 0.00014512076515391375,
      "loss": 1.1402,
      "step": 26020
    },
    {
      "epoch": 1.0546787950001013,
      "grad_norm": 0.5174545049667358,
      "learning_rate": 0.0001450828692518108,
      "loss": 1.077,
      "step": 26030
    },
    {
      "epoch": 1.0550839731772046,
      "grad_norm": 0.6677821278572083,
      "learning_rate": 0.00014504496522205334,
      "loss": 1.1114,
      "step": 26040
    },
    {
      "epoch": 1.055489151354308,
      "grad_norm": 0.5404249429702759,
      "learning_rate": 0.00014500705307147484,
      "loss": 1.0955,
      "step": 26050
    },
    {
      "epoch": 1.0558943295314114,
      "grad_norm": 0.5936490297317505,
      "learning_rate": 0.00014496913280691022,
      "loss": 1.0816,
      "step": 26060
    },
    {
      "epoch": 1.0562995077085149,
      "grad_norm": 0.6479759216308594,
      "learning_rate": 0.0001449312044351958,
      "loss": 1.1185,
      "step": 26070
    },
    {
      "epoch": 1.0567046858856182,
      "grad_norm": 0.6287264227867126,
      "learning_rate": 0.00014489326796316942,
      "loss": 1.1421,
      "step": 26080
    },
    {
      "epoch": 1.0571098640627217,
      "grad_norm": 0.5583555102348328,
      "learning_rate": 0.00014485532339767037,
      "loss": 1.077,
      "step": 26090
    },
    {
      "epoch": 1.057515042239825,
      "grad_norm": 0.6310145258903503,
      "learning_rate": 0.00014481737074553938,
      "loss": 1.1253,
      "step": 26100
    },
    {
      "epoch": 1.0579202204169285,
      "grad_norm": 0.6145515441894531,
      "learning_rate": 0.00014477941001361864,
      "loss": 1.0783,
      "step": 26110
    },
    {
      "epoch": 1.0583253985940317,
      "grad_norm": 0.5708408951759338,
      "learning_rate": 0.00014474144120875187,
      "loss": 1.0748,
      "step": 26120
    },
    {
      "epoch": 1.058730576771135,
      "grad_norm": 0.5614081621170044,
      "learning_rate": 0.0001447034643377841,
      "loss": 1.1399,
      "step": 26130
    },
    {
      "epoch": 1.0591357549482385,
      "grad_norm": 0.6080864667892456,
      "learning_rate": 0.00014466547940756197,
      "loss": 1.0862,
      "step": 26140
    },
    {
      "epoch": 1.0595409331253418,
      "grad_norm": 0.6461595892906189,
      "learning_rate": 0.00014462748642493344,
      "loss": 1.082,
      "step": 26150
    },
    {
      "epoch": 1.0599461113024453,
      "grad_norm": 0.6487632989883423,
      "learning_rate": 0.000144589485396748,
      "loss": 1.0889,
      "step": 26160
    },
    {
      "epoch": 1.0603512894795486,
      "grad_norm": 0.5825628638267517,
      "learning_rate": 0.0001445514763298566,
      "loss": 1.1382,
      "step": 26170
    },
    {
      "epoch": 1.060756467656652,
      "grad_norm": 0.5799226760864258,
      "learning_rate": 0.0001445134592311116,
      "loss": 1.0921,
      "step": 26180
    },
    {
      "epoch": 1.0611616458337554,
      "grad_norm": 0.5897616147994995,
      "learning_rate": 0.00014447543410736678,
      "loss": 1.1368,
      "step": 26190
    },
    {
      "epoch": 1.0615668240108587,
      "grad_norm": 0.5683260560035706,
      "learning_rate": 0.00014443740096547745,
      "loss": 1.1228,
      "step": 26200
    },
    {
      "epoch": 1.0619720021879622,
      "grad_norm": 0.6233024001121521,
      "learning_rate": 0.00014439935981230034,
      "loss": 1.1206,
      "step": 26210
    },
    {
      "epoch": 1.0623771803650655,
      "grad_norm": 0.5543897151947021,
      "learning_rate": 0.00014436131065469353,
      "loss": 1.1014,
      "step": 26220
    },
    {
      "epoch": 1.062782358542169,
      "grad_norm": 0.634801983833313,
      "learning_rate": 0.00014432325349951667,
      "loss": 1.1207,
      "step": 26230
    },
    {
      "epoch": 1.0631875367192722,
      "grad_norm": 0.6338844895362854,
      "learning_rate": 0.00014428518835363078,
      "loss": 1.1005,
      "step": 26240
    },
    {
      "epoch": 1.0635927148963757,
      "grad_norm": 0.5372252464294434,
      "learning_rate": 0.0001442471152238984,
      "loss": 1.0563,
      "step": 26250
    },
    {
      "epoch": 1.063997893073479,
      "grad_norm": 0.6208821535110474,
      "learning_rate": 0.00014420903411718338,
      "loss": 1.1307,
      "step": 26260
    },
    {
      "epoch": 1.0644030712505825,
      "grad_norm": 0.6101475358009338,
      "learning_rate": 0.0001441709450403511,
      "loss": 1.1105,
      "step": 26270
    },
    {
      "epoch": 1.0648082494276858,
      "grad_norm": 0.5925567746162415,
      "learning_rate": 0.00014413284800026836,
      "loss": 1.0753,
      "step": 26280
    },
    {
      "epoch": 1.0652134276047893,
      "grad_norm": 0.5899381637573242,
      "learning_rate": 0.00014409474300380341,
      "loss": 1.0624,
      "step": 26290
    },
    {
      "epoch": 1.0656186057818926,
      "grad_norm": 0.5864571332931519,
      "learning_rate": 0.0001440566300578259,
      "loss": 1.0928,
      "step": 26300
    },
    {
      "epoch": 1.0660237839589959,
      "grad_norm": 0.5230423212051392,
      "learning_rate": 0.0001440185091692069,
      "loss": 1.115,
      "step": 26310
    },
    {
      "epoch": 1.0664289621360994,
      "grad_norm": 0.585584282875061,
      "learning_rate": 0.00014398038034481897,
      "loss": 1.0617,
      "step": 26320
    },
    {
      "epoch": 1.0668341403132027,
      "grad_norm": 0.5654427409172058,
      "learning_rate": 0.0001439422435915361,
      "loss": 1.1174,
      "step": 26330
    },
    {
      "epoch": 1.0672393184903062,
      "grad_norm": 0.6408030390739441,
      "learning_rate": 0.00014390409891623359,
      "loss": 1.1134,
      "step": 26340
    },
    {
      "epoch": 1.0676444966674095,
      "grad_norm": 0.6206217408180237,
      "learning_rate": 0.0001438659463257884,
      "loss": 1.0946,
      "step": 26350
    },
    {
      "epoch": 1.068049674844513,
      "grad_norm": 0.54759281873703,
      "learning_rate": 0.00014382778582707864,
      "loss": 1.0886,
      "step": 26360
    },
    {
      "epoch": 1.0684548530216162,
      "grad_norm": 0.6493293642997742,
      "learning_rate": 0.000143789617426984,
      "loss": 1.038,
      "step": 26370
    },
    {
      "epoch": 1.0688600311987195,
      "grad_norm": 0.5322364568710327,
      "learning_rate": 0.0001437514411323857,
      "loss": 1.0919,
      "step": 26380
    },
    {
      "epoch": 1.069265209375823,
      "grad_norm": 0.5385836958885193,
      "learning_rate": 0.00014371325695016613,
      "loss": 1.1174,
      "step": 26390
    },
    {
      "epoch": 1.0696703875529263,
      "grad_norm": 0.5388386845588684,
      "learning_rate": 0.0001436750648872093,
      "loss": 1.0984,
      "step": 26400
    },
    {
      "epoch": 1.0700755657300298,
      "grad_norm": 0.7017953991889954,
      "learning_rate": 0.00014363686495040052,
      "loss": 1.0986,
      "step": 26410
    },
    {
      "epoch": 1.070480743907133,
      "grad_norm": 0.5791878700256348,
      "learning_rate": 0.0001435986571466266,
      "loss": 1.0878,
      "step": 26420
    },
    {
      "epoch": 1.0708859220842366,
      "grad_norm": 0.6056097149848938,
      "learning_rate": 0.00014356044148277575,
      "loss": 1.0868,
      "step": 26430
    },
    {
      "epoch": 1.07129110026134,
      "grad_norm": 0.614347517490387,
      "learning_rate": 0.00014352221796573757,
      "loss": 1.1176,
      "step": 26440
    },
    {
      "epoch": 1.0716962784384434,
      "grad_norm": 0.6573976874351501,
      "learning_rate": 0.0001434839866024031,
      "loss": 1.1246,
      "step": 26450
    },
    {
      "epoch": 1.0721014566155467,
      "grad_norm": 0.538991391658783,
      "learning_rate": 0.00014344574739966477,
      "loss": 1.1345,
      "step": 26460
    },
    {
      "epoch": 1.0725066347926502,
      "grad_norm": 0.592564582824707,
      "learning_rate": 0.00014340750036441648,
      "loss": 1.107,
      "step": 26470
    },
    {
      "epoch": 1.0729118129697535,
      "grad_norm": 0.550403356552124,
      "learning_rate": 0.00014336924550355348,
      "loss": 1.1258,
      "step": 26480
    },
    {
      "epoch": 1.0733169911468567,
      "grad_norm": 0.5860499739646912,
      "learning_rate": 0.0001433309828239725,
      "loss": 1.0609,
      "step": 26490
    },
    {
      "epoch": 1.0737221693239603,
      "grad_norm": 0.6311929821968079,
      "learning_rate": 0.00014329271233257148,
      "loss": 1.0726,
      "step": 26500
    },
    {
      "epoch": 1.0741273475010635,
      "grad_norm": 0.6649184226989746,
      "learning_rate": 0.0001432544340362501,
      "loss": 1.0955,
      "step": 26510
    },
    {
      "epoch": 1.074532525678167,
      "grad_norm": 0.5974748134613037,
      "learning_rate": 0.0001432161479419092,
      "loss": 1.0917,
      "step": 26520
    },
    {
      "epoch": 1.0749377038552703,
      "grad_norm": 0.6541274189949036,
      "learning_rate": 0.00014317785405645108,
      "loss": 1.1234,
      "step": 26530
    },
    {
      "epoch": 1.0753428820323738,
      "grad_norm": 0.6051272749900818,
      "learning_rate": 0.00014313955238677946,
      "loss": 1.073,
      "step": 26540
    },
    {
      "epoch": 1.075748060209477,
      "grad_norm": 0.5743160843849182,
      "learning_rate": 0.0001431012429397995,
      "loss": 1.0606,
      "step": 26550
    },
    {
      "epoch": 1.0761532383865804,
      "grad_norm": 0.6038636565208435,
      "learning_rate": 0.00014306292572241767,
      "loss": 1.0946,
      "step": 26560
    },
    {
      "epoch": 1.076558416563684,
      "grad_norm": 0.560884952545166,
      "learning_rate": 0.000143024600741542,
      "loss": 1.1243,
      "step": 26570
    },
    {
      "epoch": 1.0769635947407872,
      "grad_norm": 0.5876015424728394,
      "learning_rate": 0.00014298626800408166,
      "loss": 1.109,
      "step": 26580
    },
    {
      "epoch": 1.0773687729178907,
      "grad_norm": 0.5858681201934814,
      "learning_rate": 0.00014294792751694748,
      "loss": 1.1182,
      "step": 26590
    },
    {
      "epoch": 1.077773951094994,
      "grad_norm": 0.580522358417511,
      "learning_rate": 0.00014290957928705155,
      "loss": 1.1298,
      "step": 26600
    },
    {
      "epoch": 1.0781791292720975,
      "grad_norm": 0.6577509641647339,
      "learning_rate": 0.0001428712233213074,
      "loss": 1.1253,
      "step": 26610
    },
    {
      "epoch": 1.0785843074492008,
      "grad_norm": 0.6152412295341492,
      "learning_rate": 0.0001428328596266299,
      "loss": 1.1202,
      "step": 26620
    },
    {
      "epoch": 1.0789894856263043,
      "grad_norm": 0.603152871131897,
      "learning_rate": 0.0001427944882099354,
      "loss": 1.1051,
      "step": 26630
    },
    {
      "epoch": 1.0793946638034075,
      "grad_norm": 0.6269429326057434,
      "learning_rate": 0.0001427561090781416,
      "loss": 1.1391,
      "step": 26640
    },
    {
      "epoch": 1.079799841980511,
      "grad_norm": 0.5670099258422852,
      "learning_rate": 0.00014271772223816757,
      "loss": 1.1097,
      "step": 26650
    },
    {
      "epoch": 1.0802050201576143,
      "grad_norm": 0.5756505727767944,
      "learning_rate": 0.0001426793276969338,
      "loss": 1.1389,
      "step": 26660
    },
    {
      "epoch": 1.0806101983347176,
      "grad_norm": 0.7047620415687561,
      "learning_rate": 0.00014264092546136212,
      "loss": 1.152,
      "step": 26670
    },
    {
      "epoch": 1.0810153765118211,
      "grad_norm": 0.6967247128486633,
      "learning_rate": 0.0001426025155383758,
      "loss": 1.0788,
      "step": 26680
    },
    {
      "epoch": 1.0814205546889244,
      "grad_norm": 0.6133438944816589,
      "learning_rate": 0.00014256409793489953,
      "loss": 1.0933,
      "step": 26690
    },
    {
      "epoch": 1.081825732866028,
      "grad_norm": 0.5546432733535767,
      "learning_rate": 0.00014252567265785924,
      "loss": 1.0976,
      "step": 26700
    },
    {
      "epoch": 1.0822309110431312,
      "grad_norm": 0.6033117175102234,
      "learning_rate": 0.0001424872397141824,
      "loss": 1.1118,
      "step": 26710
    },
    {
      "epoch": 1.0826360892202347,
      "grad_norm": 0.6258071660995483,
      "learning_rate": 0.00014244879911079779,
      "loss": 1.1205,
      "step": 26720
    },
    {
      "epoch": 1.083041267397338,
      "grad_norm": 0.5777971744537354,
      "learning_rate": 0.00014241035085463556,
      "loss": 1.1004,
      "step": 26730
    },
    {
      "epoch": 1.0834464455744413,
      "grad_norm": 0.6235049962997437,
      "learning_rate": 0.00014237189495262727,
      "loss": 1.0661,
      "step": 26740
    },
    {
      "epoch": 1.0838516237515448,
      "grad_norm": 0.628309965133667,
      "learning_rate": 0.0001423334314117059,
      "loss": 1.1198,
      "step": 26750
    },
    {
      "epoch": 1.084256801928648,
      "grad_norm": 0.6381855010986328,
      "learning_rate": 0.00014229496023880566,
      "loss": 1.1128,
      "step": 26760
    },
    {
      "epoch": 1.0846619801057515,
      "grad_norm": 0.6697658896446228,
      "learning_rate": 0.00014225648144086228,
      "loss": 1.1374,
      "step": 26770
    },
    {
      "epoch": 1.0850671582828548,
      "grad_norm": 0.6239079833030701,
      "learning_rate": 0.00014221799502481283,
      "loss": 1.0885,
      "step": 26780
    },
    {
      "epoch": 1.0854723364599583,
      "grad_norm": 0.5700874924659729,
      "learning_rate": 0.00014217950099759569,
      "loss": 1.0649,
      "step": 26790
    },
    {
      "epoch": 1.0858775146370616,
      "grad_norm": 0.6274865865707397,
      "learning_rate": 0.00014214099936615075,
      "loss": 1.0923,
      "step": 26800
    },
    {
      "epoch": 1.0862826928141651,
      "grad_norm": 0.6046205759048462,
      "learning_rate": 0.00014210249013741908,
      "loss": 1.0866,
      "step": 26810
    },
    {
      "epoch": 1.0866878709912684,
      "grad_norm": 0.585839033126831,
      "learning_rate": 0.0001420639733183433,
      "loss": 1.0768,
      "step": 26820
    },
    {
      "epoch": 1.087093049168372,
      "grad_norm": 0.5761154294013977,
      "learning_rate": 0.00014202544891586726,
      "loss": 1.1231,
      "step": 26830
    },
    {
      "epoch": 1.0874982273454752,
      "grad_norm": 0.6005216836929321,
      "learning_rate": 0.0001419869169369363,
      "loss": 1.1113,
      "step": 26840
    },
    {
      "epoch": 1.0879034055225785,
      "grad_norm": 0.5862252116203308,
      "learning_rate": 0.000141948377388497,
      "loss": 1.1277,
      "step": 26850
    },
    {
      "epoch": 1.088308583699682,
      "grad_norm": 0.5907478928565979,
      "learning_rate": 0.0001419098302774974,
      "loss": 1.1246,
      "step": 26860
    },
    {
      "epoch": 1.0887137618767853,
      "grad_norm": 0.5985583066940308,
      "learning_rate": 0.00014187127561088688,
      "loss": 1.0858,
      "step": 26870
    },
    {
      "epoch": 1.0891189400538888,
      "grad_norm": 0.6343196034431458,
      "learning_rate": 0.00014183271339561615,
      "loss": 1.0732,
      "step": 26880
    },
    {
      "epoch": 1.089524118230992,
      "grad_norm": 0.5742111206054688,
      "learning_rate": 0.00014179414363863733,
      "loss": 1.1257,
      "step": 26890
    },
    {
      "epoch": 1.0899292964080955,
      "grad_norm": 0.6167534589767456,
      "learning_rate": 0.00014175556634690384,
      "loss": 1.134,
      "step": 26900
    },
    {
      "epoch": 1.0903344745851988,
      "grad_norm": 0.5855323672294617,
      "learning_rate": 0.00014171698152737054,
      "loss": 1.111,
      "step": 26910
    },
    {
      "epoch": 1.0907396527623021,
      "grad_norm": 0.6325700283050537,
      "learning_rate": 0.00014167838918699354,
      "loss": 1.0577,
      "step": 26920
    },
    {
      "epoch": 1.0911448309394056,
      "grad_norm": 0.6601378321647644,
      "learning_rate": 0.0001416397893327304,
      "loss": 1.1044,
      "step": 26930
    },
    {
      "epoch": 1.091550009116509,
      "grad_norm": 0.5777128338813782,
      "learning_rate": 0.00014160118197154,
      "loss": 1.074,
      "step": 26940
    },
    {
      "epoch": 1.0919551872936124,
      "grad_norm": 0.6005432605743408,
      "learning_rate": 0.0001415625671103826,
      "loss": 1.0887,
      "step": 26950
    },
    {
      "epoch": 1.0923603654707157,
      "grad_norm": 0.610356330871582,
      "learning_rate": 0.00014152394475621974,
      "loss": 1.0879,
      "step": 26960
    },
    {
      "epoch": 1.0927655436478192,
      "grad_norm": 0.689161479473114,
      "learning_rate": 0.00014148531491601439,
      "loss": 1.0857,
      "step": 26970
    },
    {
      "epoch": 1.0931707218249225,
      "grad_norm": 0.6296124458312988,
      "learning_rate": 0.00014144667759673077,
      "loss": 1.1235,
      "step": 26980
    },
    {
      "epoch": 1.093575900002026,
      "grad_norm": 0.7085165977478027,
      "learning_rate": 0.0001414080328053346,
      "loss": 1.0937,
      "step": 26990
    },
    {
      "epoch": 1.0939810781791293,
      "grad_norm": 0.6241776347160339,
      "learning_rate": 0.00014136938054879283,
      "loss": 1.0912,
      "step": 27000
    },
    {
      "epoch": 1.0943862563562328,
      "grad_norm": 0.59569251537323,
      "learning_rate": 0.00014133072083407375,
      "loss": 1.0797,
      "step": 27010
    },
    {
      "epoch": 1.094791434533336,
      "grad_norm": 0.6028964519500732,
      "learning_rate": 0.0001412920536681471,
      "loss": 1.097,
      "step": 27020
    },
    {
      "epoch": 1.0951966127104393,
      "grad_norm": 0.5880803465843201,
      "learning_rate": 0.00014125337905798382,
      "loss": 1.0628,
      "step": 27030
    },
    {
      "epoch": 1.0956017908875428,
      "grad_norm": 0.5959032773971558,
      "learning_rate": 0.00014121469701055633,
      "loss": 1.0713,
      "step": 27040
    },
    {
      "epoch": 1.0960069690646461,
      "grad_norm": 0.6123442649841309,
      "learning_rate": 0.0001411760075328383,
      "loss": 1.1131,
      "step": 27050
    },
    {
      "epoch": 1.0964121472417496,
      "grad_norm": 0.6159288883209229,
      "learning_rate": 0.00014113731063180476,
      "loss": 1.0967,
      "step": 27060
    },
    {
      "epoch": 1.096817325418853,
      "grad_norm": 0.6250906586647034,
      "learning_rate": 0.00014109860631443213,
      "loss": 1.0693,
      "step": 27070
    },
    {
      "epoch": 1.0972225035959564,
      "grad_norm": 0.5735248923301697,
      "learning_rate": 0.00014105989458769805,
      "loss": 1.1007,
      "step": 27080
    },
    {
      "epoch": 1.0976276817730597,
      "grad_norm": 0.5919581651687622,
      "learning_rate": 0.00014102117545858163,
      "loss": 1.1045,
      "step": 27090
    },
    {
      "epoch": 1.098032859950163,
      "grad_norm": 0.6055933833122253,
      "learning_rate": 0.0001409824489340632,
      "loss": 1.0935,
      "step": 27100
    },
    {
      "epoch": 1.0984380381272665,
      "grad_norm": 0.49889346957206726,
      "learning_rate": 0.00014094371502112457,
      "loss": 1.1245,
      "step": 27110
    },
    {
      "epoch": 1.0988432163043698,
      "grad_norm": 0.6483665704727173,
      "learning_rate": 0.00014090497372674866,
      "loss": 1.0985,
      "step": 27120
    },
    {
      "epoch": 1.0992483944814733,
      "grad_norm": 0.6391435265541077,
      "learning_rate": 0.00014086622505792,
      "loss": 1.0788,
      "step": 27130
    },
    {
      "epoch": 1.0996535726585766,
      "grad_norm": 0.6820603013038635,
      "learning_rate": 0.00014082746902162414,
      "loss": 1.1066,
      "step": 27140
    },
    {
      "epoch": 1.10005875083568,
      "grad_norm": 0.5707612633705139,
      "learning_rate": 0.0001407887056248482,
      "loss": 1.0893,
      "step": 27150
    },
    {
      "epoch": 1.1004639290127833,
      "grad_norm": 0.6191715598106384,
      "learning_rate": 0.00014074993487458057,
      "loss": 1.0838,
      "step": 27160
    },
    {
      "epoch": 1.1008691071898868,
      "grad_norm": 0.5927021503448486,
      "learning_rate": 0.00014071115677781086,
      "loss": 1.0903,
      "step": 27170
    },
    {
      "epoch": 1.1012742853669901,
      "grad_norm": 0.6548678874969482,
      "learning_rate": 0.00014067237134153017,
      "loss": 1.1071,
      "step": 27180
    },
    {
      "epoch": 1.1016794635440936,
      "grad_norm": 0.6651172041893005,
      "learning_rate": 0.00014063357857273077,
      "loss": 1.0896,
      "step": 27190
    },
    {
      "epoch": 1.102084641721197,
      "grad_norm": 0.6071325540542603,
      "learning_rate": 0.00014059477847840637,
      "loss": 1.1137,
      "step": 27200
    },
    {
      "epoch": 1.1024898198983002,
      "grad_norm": 0.5879667401313782,
      "learning_rate": 0.00014055597106555192,
      "loss": 1.0873,
      "step": 27210
    },
    {
      "epoch": 1.1028949980754037,
      "grad_norm": 0.5434188842773438,
      "learning_rate": 0.00014051715634116373,
      "loss": 1.0794,
      "step": 27220
    },
    {
      "epoch": 1.103300176252507,
      "grad_norm": 0.5912086963653564,
      "learning_rate": 0.00014047833431223938,
      "loss": 1.1102,
      "step": 27230
    },
    {
      "epoch": 1.1037053544296105,
      "grad_norm": 0.6053138971328735,
      "learning_rate": 0.00014043950498577787,
      "loss": 1.1099,
      "step": 27240
    },
    {
      "epoch": 1.1041105326067138,
      "grad_norm": 0.5635268092155457,
      "learning_rate": 0.0001404006683687794,
      "loss": 1.0892,
      "step": 27250
    },
    {
      "epoch": 1.1045157107838173,
      "grad_norm": 0.5757640600204468,
      "learning_rate": 0.00014036182446824555,
      "loss": 1.1057,
      "step": 27260
    },
    {
      "epoch": 1.1049208889609206,
      "grad_norm": 0.603508710861206,
      "learning_rate": 0.0001403229732911792,
      "loss": 1.0804,
      "step": 27270
    },
    {
      "epoch": 1.1053260671380238,
      "grad_norm": 0.600121259689331,
      "learning_rate": 0.00014028411484458454,
      "loss": 1.128,
      "step": 27280
    },
    {
      "epoch": 1.1057312453151273,
      "grad_norm": 0.6206448674201965,
      "learning_rate": 0.00014024524913546705,
      "loss": 1.1062,
      "step": 27290
    },
    {
      "epoch": 1.1061364234922306,
      "grad_norm": 0.6245536208152771,
      "learning_rate": 0.00014020637617083358,
      "loss": 1.0858,
      "step": 27300
    },
    {
      "epoch": 1.1065416016693341,
      "grad_norm": 0.5960824489593506,
      "learning_rate": 0.00014016749595769222,
      "loss": 1.1434,
      "step": 27310
    },
    {
      "epoch": 1.1069467798464374,
      "grad_norm": 0.640734076499939,
      "learning_rate": 0.0001401286085030524,
      "loss": 1.1156,
      "step": 27320
    },
    {
      "epoch": 1.107351958023541,
      "grad_norm": 0.6902362704277039,
      "learning_rate": 0.00014008971381392483,
      "loss": 1.0814,
      "step": 27330
    },
    {
      "epoch": 1.1077571362006442,
      "grad_norm": 0.6791260242462158,
      "learning_rate": 0.00014005081189732158,
      "loss": 1.0656,
      "step": 27340
    },
    {
      "epoch": 1.1081623143777477,
      "grad_norm": 0.6061962246894836,
      "learning_rate": 0.00014001190276025593,
      "loss": 1.1088,
      "step": 27350
    },
    {
      "epoch": 1.108567492554851,
      "grad_norm": 0.602505624294281,
      "learning_rate": 0.00013997298640974258,
      "loss": 1.063,
      "step": 27360
    },
    {
      "epoch": 1.1089726707319545,
      "grad_norm": 0.5898577570915222,
      "learning_rate": 0.0001399340628527974,
      "loss": 1.0797,
      "step": 27370
    },
    {
      "epoch": 1.1093778489090578,
      "grad_norm": 0.6196503043174744,
      "learning_rate": 0.00013989513209643775,
      "loss": 1.1,
      "step": 27380
    },
    {
      "epoch": 1.109783027086161,
      "grad_norm": 0.6348869800567627,
      "learning_rate": 0.00013985619414768206,
      "loss": 1.094,
      "step": 27390
    },
    {
      "epoch": 1.1101882052632646,
      "grad_norm": 0.5773240327835083,
      "learning_rate": 0.00013981724901355016,
      "loss": 1.1044,
      "step": 27400
    },
    {
      "epoch": 1.1105933834403678,
      "grad_norm": 0.5290539264678955,
      "learning_rate": 0.00013977829670106323,
      "loss": 1.1042,
      "step": 27410
    },
    {
      "epoch": 1.1109985616174713,
      "grad_norm": 0.6141327023506165,
      "learning_rate": 0.00013973933721724363,
      "loss": 1.0774,
      "step": 27420
    },
    {
      "epoch": 1.1114037397945746,
      "grad_norm": 0.6001359224319458,
      "learning_rate": 0.00013970037056911517,
      "loss": 1.111,
      "step": 27430
    },
    {
      "epoch": 1.1118089179716781,
      "grad_norm": 0.5852965712547302,
      "learning_rate": 0.00013966139676370275,
      "loss": 1.1186,
      "step": 27440
    },
    {
      "epoch": 1.1122140961487814,
      "grad_norm": 0.6617714762687683,
      "learning_rate": 0.00013962241580803278,
      "loss": 1.0866,
      "step": 27450
    },
    {
      "epoch": 1.1126192743258847,
      "grad_norm": 0.6397736668586731,
      "learning_rate": 0.00013958342770913274,
      "loss": 1.1108,
      "step": 27460
    },
    {
      "epoch": 1.1130244525029882,
      "grad_norm": 0.6202048659324646,
      "learning_rate": 0.00013954443247403154,
      "loss": 1.0859,
      "step": 27470
    },
    {
      "epoch": 1.1134296306800915,
      "grad_norm": 0.6720661520957947,
      "learning_rate": 0.00013950543010975936,
      "loss": 1.1045,
      "step": 27480
    },
    {
      "epoch": 1.113834808857195,
      "grad_norm": 0.6454536318778992,
      "learning_rate": 0.00013946642062334766,
      "loss": 1.082,
      "step": 27490
    },
    {
      "epoch": 1.1142399870342983,
      "grad_norm": 0.5782843232154846,
      "learning_rate": 0.0001394274040218291,
      "loss": 1.0683,
      "step": 27500
    },
    {
      "epoch": 1.1146451652114018,
      "grad_norm": 0.5623925924301147,
      "learning_rate": 0.00013938838031223778,
      "loss": 1.111,
      "step": 27510
    },
    {
      "epoch": 1.115050343388505,
      "grad_norm": 0.622316300868988,
      "learning_rate": 0.00013934934950160893,
      "loss": 1.0961,
      "step": 27520
    },
    {
      "epoch": 1.1154555215656086,
      "grad_norm": 0.6075736880302429,
      "learning_rate": 0.00013931031159697914,
      "loss": 1.1155,
      "step": 27530
    },
    {
      "epoch": 1.1158606997427118,
      "grad_norm": 0.6608447432518005,
      "learning_rate": 0.00013927126660538626,
      "loss": 1.1015,
      "step": 27540
    },
    {
      "epoch": 1.1162658779198154,
      "grad_norm": 0.5758466124534607,
      "learning_rate": 0.00013923221453386944,
      "loss": 1.1246,
      "step": 27550
    },
    {
      "epoch": 1.1166710560969186,
      "grad_norm": 0.632280170917511,
      "learning_rate": 0.00013919315538946905,
      "loss": 1.1298,
      "step": 27560
    },
    {
      "epoch": 1.117076234274022,
      "grad_norm": 0.672844648361206,
      "learning_rate": 0.0001391540891792268,
      "loss": 1.1215,
      "step": 27570
    },
    {
      "epoch": 1.1174814124511254,
      "grad_norm": 0.5779176354408264,
      "learning_rate": 0.00013911501591018567,
      "loss": 1.1052,
      "step": 27580
    },
    {
      "epoch": 1.1178865906282287,
      "grad_norm": 0.6281021237373352,
      "learning_rate": 0.00013907593558938985,
      "loss": 1.0976,
      "step": 27590
    },
    {
      "epoch": 1.1182917688053322,
      "grad_norm": 0.5592892169952393,
      "learning_rate": 0.00013903684822388484,
      "loss": 1.1449,
      "step": 27600
    },
    {
      "epoch": 1.1186969469824355,
      "grad_norm": 0.6328446269035339,
      "learning_rate": 0.00013899775382071743,
      "loss": 1.0843,
      "step": 27610
    },
    {
      "epoch": 1.119102125159539,
      "grad_norm": 0.5505235195159912,
      "learning_rate": 0.00013895865238693563,
      "loss": 1.1193,
      "step": 27620
    },
    {
      "epoch": 1.1195073033366423,
      "grad_norm": 0.5973195433616638,
      "learning_rate": 0.00013891954392958878,
      "loss": 1.1177,
      "step": 27630
    },
    {
      "epoch": 1.1199124815137456,
      "grad_norm": 0.6273384690284729,
      "learning_rate": 0.0001388804284557274,
      "loss": 1.088,
      "step": 27640
    },
    {
      "epoch": 1.120317659690849,
      "grad_norm": 0.6512383222579956,
      "learning_rate": 0.0001388413059724034,
      "loss": 1.0768,
      "step": 27650
    },
    {
      "epoch": 1.1207228378679523,
      "grad_norm": 0.6307324767112732,
      "learning_rate": 0.00013880217648666983,
      "loss": 1.1012,
      "step": 27660
    },
    {
      "epoch": 1.1211280160450559,
      "grad_norm": 0.6457685828208923,
      "learning_rate": 0.0001387630400055811,
      "loss": 1.1445,
      "step": 27670
    },
    {
      "epoch": 1.1215331942221591,
      "grad_norm": 0.6638851165771484,
      "learning_rate": 0.00013872389653619274,
      "loss": 1.1072,
      "step": 27680
    },
    {
      "epoch": 1.1219383723992626,
      "grad_norm": 0.5970733165740967,
      "learning_rate": 0.00013868474608556177,
      "loss": 1.1075,
      "step": 27690
    },
    {
      "epoch": 1.122343550576366,
      "grad_norm": 0.5962731242179871,
      "learning_rate": 0.00013864558866074622,
      "loss": 1.1066,
      "step": 27700
    },
    {
      "epoch": 1.1227487287534694,
      "grad_norm": 0.649998664855957,
      "learning_rate": 0.00013860642426880552,
      "loss": 1.1147,
      "step": 27710
    },
    {
      "epoch": 1.1231539069305727,
      "grad_norm": 0.5940589308738708,
      "learning_rate": 0.00013856725291680035,
      "loss": 1.0649,
      "step": 27720
    },
    {
      "epoch": 1.1235590851076762,
      "grad_norm": 0.5607200860977173,
      "learning_rate": 0.00013852807461179261,
      "loss": 1.0972,
      "step": 27730
    },
    {
      "epoch": 1.1239642632847795,
      "grad_norm": 1.200250506401062,
      "learning_rate": 0.00013848888936084545,
      "loss": 1.0636,
      "step": 27740
    },
    {
      "epoch": 1.1243694414618828,
      "grad_norm": 0.6288065910339355,
      "learning_rate": 0.0001384496971710233,
      "loss": 1.069,
      "step": 27750
    },
    {
      "epoch": 1.1247746196389863,
      "grad_norm": 0.6106061935424805,
      "learning_rate": 0.00013841049804939185,
      "loss": 1.0787,
      "step": 27760
    },
    {
      "epoch": 1.1251797978160896,
      "grad_norm": 0.5610116124153137,
      "learning_rate": 0.00013837129200301794,
      "loss": 1.087,
      "step": 27770
    },
    {
      "epoch": 1.125584975993193,
      "grad_norm": 0.6568472981452942,
      "learning_rate": 0.00013833207903896983,
      "loss": 1.0749,
      "step": 27780
    },
    {
      "epoch": 1.1259901541702964,
      "grad_norm": 0.5837673544883728,
      "learning_rate": 0.0001382928591643169,
      "loss": 1.1537,
      "step": 27790
    },
    {
      "epoch": 1.1263953323473999,
      "grad_norm": 0.5647041201591492,
      "learning_rate": 0.00013825363238612978,
      "loss": 1.0809,
      "step": 27800
    },
    {
      "epoch": 1.1268005105245031,
      "grad_norm": 0.6215106844902039,
      "learning_rate": 0.00013821439871148042,
      "loss": 1.1201,
      "step": 27810
    },
    {
      "epoch": 1.1272056887016064,
      "grad_norm": 0.5409258008003235,
      "learning_rate": 0.00013817515814744188,
      "loss": 1.104,
      "step": 27820
    },
    {
      "epoch": 1.12761086687871,
      "grad_norm": 0.5859090685844421,
      "learning_rate": 0.00013813591070108868,
      "loss": 1.0778,
      "step": 27830
    },
    {
      "epoch": 1.1280160450558132,
      "grad_norm": 0.630394458770752,
      "learning_rate": 0.00013809665637949637,
      "loss": 1.0933,
      "step": 27840
    },
    {
      "epoch": 1.1284212232329167,
      "grad_norm": 0.5592109560966492,
      "learning_rate": 0.0001380573951897418,
      "loss": 1.1414,
      "step": 27850
    },
    {
      "epoch": 1.12882640141002,
      "grad_norm": 0.6050084829330444,
      "learning_rate": 0.00013801812713890312,
      "loss": 1.1207,
      "step": 27860
    },
    {
      "epoch": 1.1292315795871235,
      "grad_norm": 0.5879210829734802,
      "learning_rate": 0.0001379788522340597,
      "loss": 1.0746,
      "step": 27870
    },
    {
      "epoch": 1.1296367577642268,
      "grad_norm": 0.6682519316673279,
      "learning_rate": 0.00013793957048229203,
      "loss": 1.0816,
      "step": 27880
    },
    {
      "epoch": 1.1300419359413303,
      "grad_norm": 0.5897209048271179,
      "learning_rate": 0.00013790028189068198,
      "loss": 1.0968,
      "step": 27890
    },
    {
      "epoch": 1.1304471141184336,
      "grad_norm": 0.6283105611801147,
      "learning_rate": 0.0001378609864663126,
      "loss": 1.1132,
      "step": 27900
    },
    {
      "epoch": 1.130852292295537,
      "grad_norm": 0.5982614755630493,
      "learning_rate": 0.00013782168421626816,
      "loss": 1.1186,
      "step": 27910
    },
    {
      "epoch": 1.1312574704726404,
      "grad_norm": 0.608857274055481,
      "learning_rate": 0.00013778237514763418,
      "loss": 1.0899,
      "step": 27920
    },
    {
      "epoch": 1.1316626486497436,
      "grad_norm": 0.5838295221328735,
      "learning_rate": 0.0001377430592674974,
      "loss": 1.0953,
      "step": 27930
    },
    {
      "epoch": 1.1320678268268471,
      "grad_norm": 0.5614155530929565,
      "learning_rate": 0.00013770373658294574,
      "loss": 1.0879,
      "step": 27940
    },
    {
      "epoch": 1.1324730050039504,
      "grad_norm": 0.6421583294868469,
      "learning_rate": 0.00013766440710106847,
      "loss": 1.1045,
      "step": 27950
    },
    {
      "epoch": 1.132878183181054,
      "grad_norm": 0.5798307061195374,
      "learning_rate": 0.00013762507082895593,
      "loss": 1.1231,
      "step": 27960
    },
    {
      "epoch": 1.1332833613581572,
      "grad_norm": 0.5741163492202759,
      "learning_rate": 0.00013758572777369983,
      "loss": 1.1402,
      "step": 27970
    },
    {
      "epoch": 1.1336885395352607,
      "grad_norm": 0.6244339346885681,
      "learning_rate": 0.000137546377942393,
      "loss": 1.0928,
      "step": 27980
    },
    {
      "epoch": 1.134093717712364,
      "grad_norm": 0.5632155537605286,
      "learning_rate": 0.00013750702134212954,
      "loss": 1.0694,
      "step": 27990
    },
    {
      "epoch": 1.1344988958894673,
      "grad_norm": 0.560606837272644,
      "learning_rate": 0.00013746765798000477,
      "loss": 1.1088,
      "step": 28000
    },
    {
      "epoch": 1.1349040740665708,
      "grad_norm": 0.5608622431755066,
      "learning_rate": 0.0001374282878631152,
      "loss": 1.1077,
      "step": 28010
    },
    {
      "epoch": 1.135309252243674,
      "grad_norm": 0.705204427242279,
      "learning_rate": 0.00013738891099855855,
      "loss": 1.1182,
      "step": 28020
    },
    {
      "epoch": 1.1357144304207776,
      "grad_norm": 0.5462162494659424,
      "learning_rate": 0.00013734952739343382,
      "loss": 1.1143,
      "step": 28030
    },
    {
      "epoch": 1.1361196085978809,
      "grad_norm": 0.6358756422996521,
      "learning_rate": 0.00013731013705484122,
      "loss": 1.0885,
      "step": 28040
    },
    {
      "epoch": 1.1365247867749844,
      "grad_norm": 0.5974259376525879,
      "learning_rate": 0.00013727073998988202,
      "loss": 1.1212,
      "step": 28050
    },
    {
      "epoch": 1.1369299649520876,
      "grad_norm": 0.534686267375946,
      "learning_rate": 0.00013723133620565896,
      "loss": 1.126,
      "step": 28060
    },
    {
      "epoch": 1.1373351431291912,
      "grad_norm": 0.6581855416297913,
      "learning_rate": 0.00013719192570927575,
      "loss": 1.0997,
      "step": 28070
    },
    {
      "epoch": 1.1377403213062944,
      "grad_norm": 0.5521950721740723,
      "learning_rate": 0.00013715250850783749,
      "loss": 1.064,
      "step": 28080
    },
    {
      "epoch": 1.138145499483398,
      "grad_norm": 0.5708963871002197,
      "learning_rate": 0.00013711308460845036,
      "loss": 1.0588,
      "step": 28090
    },
    {
      "epoch": 1.1385506776605012,
      "grad_norm": 0.6137231588363647,
      "learning_rate": 0.00013707365401822183,
      "loss": 1.114,
      "step": 28100
    },
    {
      "epoch": 1.1389558558376045,
      "grad_norm": 0.5939890742301941,
      "learning_rate": 0.00013703421674426052,
      "loss": 1.1129,
      "step": 28110
    },
    {
      "epoch": 1.139361034014708,
      "grad_norm": 0.5904026627540588,
      "learning_rate": 0.00013699477279367636,
      "loss": 1.1206,
      "step": 28120
    },
    {
      "epoch": 1.1397662121918113,
      "grad_norm": 0.6224991083145142,
      "learning_rate": 0.0001369553221735803,
      "loss": 1.0847,
      "step": 28130
    },
    {
      "epoch": 1.1401713903689148,
      "grad_norm": 0.5633615255355835,
      "learning_rate": 0.00013691586489108465,
      "loss": 1.1202,
      "step": 28140
    },
    {
      "epoch": 1.140576568546018,
      "grad_norm": 0.6243061423301697,
      "learning_rate": 0.0001368764009533029,
      "loss": 1.1179,
      "step": 28150
    },
    {
      "epoch": 1.1409817467231216,
      "grad_norm": 0.5712858438491821,
      "learning_rate": 0.00013683693036734964,
      "loss": 1.1065,
      "step": 28160
    },
    {
      "epoch": 1.1413869249002249,
      "grad_norm": 0.6267960071563721,
      "learning_rate": 0.0001367974531403408,
      "loss": 1.1104,
      "step": 28170
    },
    {
      "epoch": 1.1417921030773281,
      "grad_norm": 0.6307588815689087,
      "learning_rate": 0.00013675796927939337,
      "loss": 1.0838,
      "step": 28180
    },
    {
      "epoch": 1.1421972812544317,
      "grad_norm": 0.6346031427383423,
      "learning_rate": 0.00013671847879162562,
      "loss": 1.1164,
      "step": 28190
    },
    {
      "epoch": 1.142602459431535,
      "grad_norm": 0.6077895164489746,
      "learning_rate": 0.00013667898168415703,
      "loss": 1.0827,
      "step": 28200
    },
    {
      "epoch": 1.1430076376086384,
      "grad_norm": 0.5595155358314514,
      "learning_rate": 0.0001366394779641082,
      "loss": 1.0923,
      "step": 28210
    },
    {
      "epoch": 1.1434128157857417,
      "grad_norm": 0.6161838173866272,
      "learning_rate": 0.00013659996763860097,
      "loss": 1.093,
      "step": 28220
    },
    {
      "epoch": 1.1438179939628452,
      "grad_norm": 0.6321885585784912,
      "learning_rate": 0.00013656045071475838,
      "loss": 1.1284,
      "step": 28230
    },
    {
      "epoch": 1.1442231721399485,
      "grad_norm": 0.5782040357589722,
      "learning_rate": 0.00013652092719970463,
      "loss": 1.1192,
      "step": 28240
    },
    {
      "epoch": 1.144628350317052,
      "grad_norm": 0.6806547045707703,
      "learning_rate": 0.0001364813971005651,
      "loss": 1.1034,
      "step": 28250
    },
    {
      "epoch": 1.1450335284941553,
      "grad_norm": 0.6491135358810425,
      "learning_rate": 0.0001364418604244664,
      "loss": 1.0722,
      "step": 28260
    },
    {
      "epoch": 1.1454387066712588,
      "grad_norm": 0.6195778250694275,
      "learning_rate": 0.0001364023171785363,
      "loss": 1.0759,
      "step": 28270
    },
    {
      "epoch": 1.145843884848362,
      "grad_norm": 0.6177242994308472,
      "learning_rate": 0.00013636276736990377,
      "loss": 1.1055,
      "step": 28280
    },
    {
      "epoch": 1.1462490630254654,
      "grad_norm": 0.5624657273292542,
      "learning_rate": 0.0001363232110056989,
      "loss": 1.1129,
      "step": 28290
    },
    {
      "epoch": 1.1466542412025689,
      "grad_norm": 0.728869616985321,
      "learning_rate": 0.00013628364809305308,
      "loss": 1.0983,
      "step": 28300
    },
    {
      "epoch": 1.1470594193796722,
      "grad_norm": 0.5701168179512024,
      "learning_rate": 0.00013624407863909878,
      "loss": 1.0717,
      "step": 28310
    },
    {
      "epoch": 1.1474645975567757,
      "grad_norm": 0.6729862689971924,
      "learning_rate": 0.0001362045026509697,
      "loss": 1.1012,
      "step": 28320
    },
    {
      "epoch": 1.147869775733879,
      "grad_norm": 0.6254222393035889,
      "learning_rate": 0.00013616492013580062,
      "loss": 1.1259,
      "step": 28330
    },
    {
      "epoch": 1.1482749539109824,
      "grad_norm": 0.5708780288696289,
      "learning_rate": 0.0001361253311007277,
      "loss": 1.0973,
      "step": 28340
    },
    {
      "epoch": 1.1486801320880857,
      "grad_norm": 0.5746476650238037,
      "learning_rate": 0.00013608573555288806,
      "loss": 1.1076,
      "step": 28350
    },
    {
      "epoch": 1.149085310265189,
      "grad_norm": 0.5834535360336304,
      "learning_rate": 0.00013604613349942011,
      "loss": 1.0647,
      "step": 28360
    },
    {
      "epoch": 1.1494904884422925,
      "grad_norm": 0.5688482522964478,
      "learning_rate": 0.00013600652494746344,
      "loss": 1.0866,
      "step": 28370
    },
    {
      "epoch": 1.1498956666193958,
      "grad_norm": 0.6327924132347107,
      "learning_rate": 0.00013596690990415878,
      "loss": 1.135,
      "step": 28380
    },
    {
      "epoch": 1.1503008447964993,
      "grad_norm": 0.6476439237594604,
      "learning_rate": 0.00013592728837664793,
      "loss": 1.1328,
      "step": 28390
    },
    {
      "epoch": 1.1507060229736026,
      "grad_norm": 0.5936959981918335,
      "learning_rate": 0.0001358876603720741,
      "loss": 1.1026,
      "step": 28400
    },
    {
      "epoch": 1.151111201150706,
      "grad_norm": 0.6014825105667114,
      "learning_rate": 0.00013584802589758143,
      "loss": 1.0928,
      "step": 28410
    },
    {
      "epoch": 1.1515163793278094,
      "grad_norm": 0.5824862122535706,
      "learning_rate": 0.00013580838496031537,
      "loss": 1.0764,
      "step": 28420
    },
    {
      "epoch": 1.1519215575049129,
      "grad_norm": 0.6886267066001892,
      "learning_rate": 0.00013576873756742248,
      "loss": 1.0961,
      "step": 28430
    },
    {
      "epoch": 1.1523267356820162,
      "grad_norm": 0.6216195225715637,
      "learning_rate": 0.00013572908372605052,
      "loss": 1.1198,
      "step": 28440
    },
    {
      "epoch": 1.1527319138591197,
      "grad_norm": 0.5819383263587952,
      "learning_rate": 0.0001356894234433483,
      "loss": 1.0854,
      "step": 28450
    },
    {
      "epoch": 1.153137092036223,
      "grad_norm": 0.6358830332756042,
      "learning_rate": 0.00013564975672646597,
      "loss": 1.0772,
      "step": 28460
    },
    {
      "epoch": 1.1535422702133262,
      "grad_norm": 0.6158119440078735,
      "learning_rate": 0.00013561008358255468,
      "loss": 1.1305,
      "step": 28470
    },
    {
      "epoch": 1.1539474483904297,
      "grad_norm": 0.6063367128372192,
      "learning_rate": 0.00013557040401876688,
      "loss": 1.1292,
      "step": 28480
    },
    {
      "epoch": 1.154352626567533,
      "grad_norm": 0.6628134846687317,
      "learning_rate": 0.00013553071804225605,
      "loss": 1.0392,
      "step": 28490
    },
    {
      "epoch": 1.1547578047446365,
      "grad_norm": 0.5872376561164856,
      "learning_rate": 0.00013549102566017687,
      "loss": 1.0701,
      "step": 28500
    },
    {
      "epoch": 1.1551629829217398,
      "grad_norm": 0.6152101755142212,
      "learning_rate": 0.0001354513268796852,
      "loss": 1.1014,
      "step": 28510
    },
    {
      "epoch": 1.1555681610988433,
      "grad_norm": 0.6444042921066284,
      "learning_rate": 0.00013541162170793806,
      "loss": 1.0739,
      "step": 28520
    },
    {
      "epoch": 1.1559733392759466,
      "grad_norm": 0.6219022870063782,
      "learning_rate": 0.00013537191015209357,
      "loss": 1.1069,
      "step": 28530
    },
    {
      "epoch": 1.1563785174530499,
      "grad_norm": 0.7698732018470764,
      "learning_rate": 0.00013533219221931102,
      "loss": 1.0631,
      "step": 28540
    },
    {
      "epoch": 1.1567836956301534,
      "grad_norm": 0.6028838157653809,
      "learning_rate": 0.00013529246791675088,
      "loss": 1.0827,
      "step": 28550
    },
    {
      "epoch": 1.1571888738072567,
      "grad_norm": 0.619141697883606,
      "learning_rate": 0.00013525273725157474,
      "loss": 1.1144,
      "step": 28560
    },
    {
      "epoch": 1.1575940519843602,
      "grad_norm": 0.6247971653938293,
      "learning_rate": 0.00013521300023094538,
      "loss": 1.085,
      "step": 28570
    },
    {
      "epoch": 1.1579992301614634,
      "grad_norm": 0.6257079839706421,
      "learning_rate": 0.00013517325686202665,
      "loss": 1.1052,
      "step": 28580
    },
    {
      "epoch": 1.158404408338567,
      "grad_norm": 0.5270061492919922,
      "learning_rate": 0.0001351335071519836,
      "loss": 1.115,
      "step": 28590
    },
    {
      "epoch": 1.1588095865156702,
      "grad_norm": 0.6344119906425476,
      "learning_rate": 0.00013509375110798243,
      "loss": 1.0943,
      "step": 28600
    },
    {
      "epoch": 1.1592147646927737,
      "grad_norm": 0.6420412063598633,
      "learning_rate": 0.0001350539887371904,
      "loss": 1.0999,
      "step": 28610
    },
    {
      "epoch": 1.159619942869877,
      "grad_norm": 0.5767116546630859,
      "learning_rate": 0.00013501422004677605,
      "loss": 1.1111,
      "step": 28620
    },
    {
      "epoch": 1.1600251210469805,
      "grad_norm": 0.6061212420463562,
      "learning_rate": 0.0001349744450439089,
      "loss": 1.1089,
      "step": 28630
    },
    {
      "epoch": 1.1604302992240838,
      "grad_norm": 0.6034737229347229,
      "learning_rate": 0.00013493466373575976,
      "loss": 1.0541,
      "step": 28640
    },
    {
      "epoch": 1.160835477401187,
      "grad_norm": 0.6015185713768005,
      "learning_rate": 0.00013489487612950045,
      "loss": 1.1218,
      "step": 28650
    },
    {
      "epoch": 1.1612406555782906,
      "grad_norm": 0.6290345191955566,
      "learning_rate": 0.00013485508223230402,
      "loss": 1.1092,
      "step": 28660
    },
    {
      "epoch": 1.1616458337553939,
      "grad_norm": 0.5878185629844666,
      "learning_rate": 0.0001348152820513446,
      "loss": 1.1064,
      "step": 28670
    },
    {
      "epoch": 1.1620510119324974,
      "grad_norm": 0.5786883234977722,
      "learning_rate": 0.00013477547559379748,
      "loss": 1.0938,
      "step": 28680
    },
    {
      "epoch": 1.1624561901096007,
      "grad_norm": 0.6449988484382629,
      "learning_rate": 0.00013473566286683909,
      "loss": 1.0904,
      "step": 28690
    },
    {
      "epoch": 1.1628613682867042,
      "grad_norm": 0.6475859880447388,
      "learning_rate": 0.00013469584387764688,
      "loss": 1.1065,
      "step": 28700
    },
    {
      "epoch": 1.1632665464638074,
      "grad_norm": 0.6218219995498657,
      "learning_rate": 0.00013465601863339957,
      "loss": 1.1031,
      "step": 28710
    },
    {
      "epoch": 1.1636717246409107,
      "grad_norm": 0.5907488465309143,
      "learning_rate": 0.000134616187141277,
      "loss": 1.0765,
      "step": 28720
    },
    {
      "epoch": 1.1640769028180142,
      "grad_norm": 0.6078987121582031,
      "learning_rate": 0.00013457634940846006,
      "loss": 1.0711,
      "step": 28730
    },
    {
      "epoch": 1.1644820809951175,
      "grad_norm": 0.5874805450439453,
      "learning_rate": 0.00013453650544213076,
      "loss": 1.0834,
      "step": 28740
    },
    {
      "epoch": 1.164887259172221,
      "grad_norm": 0.6462891101837158,
      "learning_rate": 0.00013449665524947234,
      "loss": 1.1355,
      "step": 28750
    },
    {
      "epoch": 1.1652924373493243,
      "grad_norm": 0.6056782603263855,
      "learning_rate": 0.00013445679883766903,
      "loss": 1.1066,
      "step": 28760
    },
    {
      "epoch": 1.1656976155264278,
      "grad_norm": 0.5952337384223938,
      "learning_rate": 0.0001344169362139063,
      "loss": 1.0954,
      "step": 28770
    },
    {
      "epoch": 1.166102793703531,
      "grad_norm": 0.6616261601448059,
      "learning_rate": 0.00013437706738537067,
      "loss": 1.1047,
      "step": 28780
    },
    {
      "epoch": 1.1665079718806346,
      "grad_norm": 0.6586977243423462,
      "learning_rate": 0.0001343371923592498,
      "loss": 1.0761,
      "step": 28790
    },
    {
      "epoch": 1.1669131500577379,
      "grad_norm": 0.5687082409858704,
      "learning_rate": 0.00013429731114273242,
      "loss": 1.0914,
      "step": 28800
    },
    {
      "epoch": 1.1673183282348414,
      "grad_norm": 0.6246545910835266,
      "learning_rate": 0.00013425742374300841,
      "loss": 1.1122,
      "step": 28810
    },
    {
      "epoch": 1.1677235064119447,
      "grad_norm": 0.5557677149772644,
      "learning_rate": 0.00013421753016726887,
      "loss": 1.0803,
      "step": 28820
    },
    {
      "epoch": 1.168128684589048,
      "grad_norm": 0.6100905537605286,
      "learning_rate": 0.00013417763042270585,
      "loss": 1.0613,
      "step": 28830
    },
    {
      "epoch": 1.1685338627661515,
      "grad_norm": 0.7004369497299194,
      "learning_rate": 0.00013413772451651253,
      "loss": 1.1117,
      "step": 28840
    },
    {
      "epoch": 1.1689390409432547,
      "grad_norm": 0.5445922613143921,
      "learning_rate": 0.00013409781245588332,
      "loss": 1.1211,
      "step": 28850
    },
    {
      "epoch": 1.1693442191203582,
      "grad_norm": 0.5755471587181091,
      "learning_rate": 0.00013405789424801362,
      "loss": 1.1283,
      "step": 28860
    },
    {
      "epoch": 1.1697493972974615,
      "grad_norm": 0.5941116809844971,
      "learning_rate": 0.00013401796990010007,
      "loss": 1.1065,
      "step": 28870
    },
    {
      "epoch": 1.170154575474565,
      "grad_norm": 0.6187422871589661,
      "learning_rate": 0.00013397803941934025,
      "loss": 1.1032,
      "step": 28880
    },
    {
      "epoch": 1.1705597536516683,
      "grad_norm": 0.5959059596061707,
      "learning_rate": 0.00013393810281293292,
      "loss": 1.1098,
      "step": 28890
    },
    {
      "epoch": 1.1709649318287716,
      "grad_norm": 0.5601053237915039,
      "learning_rate": 0.00013389816008807802,
      "loss": 1.123,
      "step": 28900
    },
    {
      "epoch": 1.171370110005875,
      "grad_norm": 0.6526708006858826,
      "learning_rate": 0.00013385821125197652,
      "loss": 1.1015,
      "step": 28910
    },
    {
      "epoch": 1.1717752881829784,
      "grad_norm": 0.6006954908370972,
      "learning_rate": 0.00013381825631183041,
      "loss": 1.0801,
      "step": 28920
    },
    {
      "epoch": 1.1721804663600819,
      "grad_norm": 0.5745493769645691,
      "learning_rate": 0.00013377829527484297,
      "loss": 1.1337,
      "step": 28930
    },
    {
      "epoch": 1.1725856445371852,
      "grad_norm": 0.6620622873306274,
      "learning_rate": 0.00013373832814821842,
      "loss": 1.0475,
      "step": 28940
    },
    {
      "epoch": 1.1729908227142887,
      "grad_norm": 0.5627890229225159,
      "learning_rate": 0.00013369835493916215,
      "loss": 1.1114,
      "step": 28950
    },
    {
      "epoch": 1.173396000891392,
      "grad_norm": 0.6274038553237915,
      "learning_rate": 0.00013365837565488064,
      "loss": 1.0806,
      "step": 28960
    },
    {
      "epoch": 1.1738011790684955,
      "grad_norm": 0.5962109565734863,
      "learning_rate": 0.00013361839030258144,
      "loss": 1.0716,
      "step": 28970
    },
    {
      "epoch": 1.1742063572455987,
      "grad_norm": 0.5591891407966614,
      "learning_rate": 0.00013357839888947324,
      "loss": 1.0952,
      "step": 28980
    },
    {
      "epoch": 1.1746115354227022,
      "grad_norm": 0.5667641758918762,
      "learning_rate": 0.00013353840142276576,
      "loss": 1.0936,
      "step": 28990
    },
    {
      "epoch": 1.1750167135998055,
      "grad_norm": 0.5810002684593201,
      "learning_rate": 0.00013349839790966988,
      "loss": 1.0869,
      "step": 29000
    },
    {
      "epoch": 1.1754218917769088,
      "grad_norm": 0.6061210632324219,
      "learning_rate": 0.0001334583883573975,
      "loss": 1.1117,
      "step": 29010
    },
    {
      "epoch": 1.1758270699540123,
      "grad_norm": 0.5663205981254578,
      "learning_rate": 0.00013341837277316168,
      "loss": 1.0793,
      "step": 29020
    },
    {
      "epoch": 1.1762322481311156,
      "grad_norm": 0.6489531397819519,
      "learning_rate": 0.00013337835116417648,
      "loss": 1.0797,
      "step": 29030
    },
    {
      "epoch": 1.176637426308219,
      "grad_norm": 0.5888655781745911,
      "learning_rate": 0.00013333832353765713,
      "loss": 1.0827,
      "step": 29040
    },
    {
      "epoch": 1.1770426044853224,
      "grad_norm": 0.6974976658821106,
      "learning_rate": 0.00013329828990081994,
      "loss": 1.1307,
      "step": 29050
    },
    {
      "epoch": 1.177447782662426,
      "grad_norm": 0.5978584289550781,
      "learning_rate": 0.00013325825026088223,
      "loss": 1.1318,
      "step": 29060
    },
    {
      "epoch": 1.1778529608395292,
      "grad_norm": 0.6338521242141724,
      "learning_rate": 0.00013321820462506247,
      "loss": 1.0783,
      "step": 29070
    },
    {
      "epoch": 1.1782581390166325,
      "grad_norm": 0.5784775614738464,
      "learning_rate": 0.0001331781530005802,
      "loss": 1.0697,
      "step": 29080
    },
    {
      "epoch": 1.178663317193736,
      "grad_norm": 0.6207610964775085,
      "learning_rate": 0.000133138095394656,
      "loss": 1.1283,
      "step": 29090
    },
    {
      "epoch": 1.1790684953708392,
      "grad_norm": 0.6358835697174072,
      "learning_rate": 0.00013309803181451156,
      "loss": 1.0897,
      "step": 29100
    },
    {
      "epoch": 1.1794736735479427,
      "grad_norm": 0.5547202229499817,
      "learning_rate": 0.0001330579622673697,
      "loss": 1.097,
      "step": 29110
    },
    {
      "epoch": 1.179878851725046,
      "grad_norm": 0.6145870089530945,
      "learning_rate": 0.00013301788676045416,
      "loss": 1.0897,
      "step": 29120
    },
    {
      "epoch": 1.1802840299021495,
      "grad_norm": 0.6073143482208252,
      "learning_rate": 0.00013297780530098993,
      "loss": 1.0789,
      "step": 29130
    },
    {
      "epoch": 1.1806892080792528,
      "grad_norm": 0.6099177598953247,
      "learning_rate": 0.00013293771789620298,
      "loss": 1.0825,
      "step": 29140
    },
    {
      "epoch": 1.1810943862563563,
      "grad_norm": 0.5423462986946106,
      "learning_rate": 0.00013289762455332034,
      "loss": 1.0785,
      "step": 29150
    },
    {
      "epoch": 1.1814995644334596,
      "grad_norm": 0.6117521524429321,
      "learning_rate": 0.00013285752527957023,
      "loss": 1.1321,
      "step": 29160
    },
    {
      "epoch": 1.181904742610563,
      "grad_norm": 0.5945439338684082,
      "learning_rate": 0.0001328174200821817,
      "loss": 1.107,
      "step": 29170
    },
    {
      "epoch": 1.1823099207876664,
      "grad_norm": 0.5787346959114075,
      "learning_rate": 0.0001327773089683852,
      "loss": 1.1144,
      "step": 29180
    },
    {
      "epoch": 1.1827150989647697,
      "grad_norm": 0.5740688443183899,
      "learning_rate": 0.00013273719194541192,
      "loss": 1.115,
      "step": 29190
    },
    {
      "epoch": 1.1831202771418732,
      "grad_norm": 0.6640257239341736,
      "learning_rate": 0.0001326970690204943,
      "loss": 1.0975,
      "step": 29200
    },
    {
      "epoch": 1.1835254553189765,
      "grad_norm": 0.6296002864837646,
      "learning_rate": 0.00013265694020086582,
      "loss": 1.0939,
      "step": 29210
    },
    {
      "epoch": 1.18393063349608,
      "grad_norm": 0.5484721660614014,
      "learning_rate": 0.000132616805493761,
      "loss": 1.0681,
      "step": 29220
    },
    {
      "epoch": 1.1843358116731832,
      "grad_norm": 0.6254235506057739,
      "learning_rate": 0.0001325766649064154,
      "loss": 1.0921,
      "step": 29230
    },
    {
      "epoch": 1.1847409898502868,
      "grad_norm": 0.7158032059669495,
      "learning_rate": 0.00013253651844606572,
      "loss": 1.0813,
      "step": 29240
    },
    {
      "epoch": 1.18514616802739,
      "grad_norm": 0.5867267847061157,
      "learning_rate": 0.0001324963661199496,
      "loss": 1.1105,
      "step": 29250
    },
    {
      "epoch": 1.1855513462044933,
      "grad_norm": 0.5724896788597107,
      "learning_rate": 0.00013245620793530586,
      "loss": 1.092,
      "step": 29260
    },
    {
      "epoch": 1.1859565243815968,
      "grad_norm": 0.6192792057991028,
      "learning_rate": 0.00013241604389937432,
      "loss": 1.1088,
      "step": 29270
    },
    {
      "epoch": 1.1863617025587,
      "grad_norm": 0.6233285665512085,
      "learning_rate": 0.00013237587401939582,
      "loss": 1.0875,
      "step": 29280
    },
    {
      "epoch": 1.1867668807358036,
      "grad_norm": 0.590226411819458,
      "learning_rate": 0.00013233569830261227,
      "loss": 1.0822,
      "step": 29290
    },
    {
      "epoch": 1.187172058912907,
      "grad_norm": 0.5883432626724243,
      "learning_rate": 0.00013229551675626673,
      "loss": 1.0693,
      "step": 29300
    },
    {
      "epoch": 1.1875772370900104,
      "grad_norm": 0.6355063915252686,
      "learning_rate": 0.00013225532938760317,
      "loss": 1.1301,
      "step": 29310
    },
    {
      "epoch": 1.1879824152671137,
      "grad_norm": 0.6220588088035583,
      "learning_rate": 0.00013221513620386666,
      "loss": 1.1068,
      "step": 29320
    },
    {
      "epoch": 1.1883875934442172,
      "grad_norm": 0.6674821376800537,
      "learning_rate": 0.00013217493721230336,
      "loss": 1.0922,
      "step": 29330
    },
    {
      "epoch": 1.1887927716213205,
      "grad_norm": 0.5923930406570435,
      "learning_rate": 0.00013213473242016047,
      "loss": 1.0998,
      "step": 29340
    },
    {
      "epoch": 1.189197949798424,
      "grad_norm": 0.5728965997695923,
      "learning_rate": 0.00013209452183468617,
      "loss": 1.1074,
      "step": 29350
    },
    {
      "epoch": 1.1896031279755273,
      "grad_norm": 0.6942472457885742,
      "learning_rate": 0.00013205430546312972,
      "loss": 1.1093,
      "step": 29360
    },
    {
      "epoch": 1.1900083061526305,
      "grad_norm": 0.6321328282356262,
      "learning_rate": 0.00013201408331274146,
      "loss": 1.0792,
      "step": 29370
    },
    {
      "epoch": 1.190413484329734,
      "grad_norm": 0.6363593339920044,
      "learning_rate": 0.00013197385539077275,
      "loss": 1.106,
      "step": 29380
    },
    {
      "epoch": 1.1908186625068373,
      "grad_norm": 0.609278678894043,
      "learning_rate": 0.00013193362170447594,
      "loss": 1.0937,
      "step": 29390
    },
    {
      "epoch": 1.1912238406839408,
      "grad_norm": 0.505521297454834,
      "learning_rate": 0.0001318933822611045,
      "loss": 1.0866,
      "step": 29400
    },
    {
      "epoch": 1.191629018861044,
      "grad_norm": 0.615839421749115,
      "learning_rate": 0.0001318531370679129,
      "loss": 1.1109,
      "step": 29410
    },
    {
      "epoch": 1.1920341970381476,
      "grad_norm": 0.5531451106071472,
      "learning_rate": 0.0001318128861321566,
      "loss": 1.1116,
      "step": 29420
    },
    {
      "epoch": 1.192439375215251,
      "grad_norm": 0.6887074112892151,
      "learning_rate": 0.00013177262946109216,
      "loss": 1.1093,
      "step": 29430
    },
    {
      "epoch": 1.1928445533923542,
      "grad_norm": 0.5314295887947083,
      "learning_rate": 0.00013173236706197719,
      "loss": 1.1071,
      "step": 29440
    },
    {
      "epoch": 1.1932497315694577,
      "grad_norm": 0.6287745833396912,
      "learning_rate": 0.0001316920989420703,
      "loss": 1.1131,
      "step": 29450
    },
    {
      "epoch": 1.193654909746561,
      "grad_norm": 0.6150389313697815,
      "learning_rate": 0.00013165182510863107,
      "loss": 1.1261,
      "step": 29460
    },
    {
      "epoch": 1.1940600879236645,
      "grad_norm": 0.5983867049217224,
      "learning_rate": 0.00013161154556892018,
      "loss": 1.122,
      "step": 29470
    },
    {
      "epoch": 1.1944652661007678,
      "grad_norm": 0.5855502486228943,
      "learning_rate": 0.0001315712603301994,
      "loss": 1.0662,
      "step": 29480
    },
    {
      "epoch": 1.1948704442778713,
      "grad_norm": 0.5401041507720947,
      "learning_rate": 0.0001315309693997314,
      "loss": 1.1129,
      "step": 29490
    },
    {
      "epoch": 1.1952756224549745,
      "grad_norm": 0.5900583267211914,
      "learning_rate": 0.00013149067278477993,
      "loss": 1.0998,
      "step": 29500
    },
    {
      "epoch": 1.195680800632078,
      "grad_norm": 0.7097587585449219,
      "learning_rate": 0.0001314503704926098,
      "loss": 1.1,
      "step": 29510
    },
    {
      "epoch": 1.1960859788091813,
      "grad_norm": 0.5851007699966431,
      "learning_rate": 0.00013141006253048672,
      "loss": 1.0644,
      "step": 29520
    },
    {
      "epoch": 1.1964911569862848,
      "grad_norm": 0.5670154094696045,
      "learning_rate": 0.00013136974890567765,
      "loss": 1.0907,
      "step": 29530
    },
    {
      "epoch": 1.1968963351633881,
      "grad_norm": 0.6695858836174011,
      "learning_rate": 0.00013132942962545033,
      "loss": 1.1071,
      "step": 29540
    },
    {
      "epoch": 1.1973015133404914,
      "grad_norm": 0.6142239570617676,
      "learning_rate": 0.00013128910469707364,
      "loss": 1.0862,
      "step": 29550
    },
    {
      "epoch": 1.197706691517595,
      "grad_norm": 0.6221494674682617,
      "learning_rate": 0.00013124877412781747,
      "loss": 1.1214,
      "step": 29560
    },
    {
      "epoch": 1.1981118696946982,
      "grad_norm": 0.6239984631538391,
      "learning_rate": 0.00013120843792495274,
      "loss": 1.1128,
      "step": 29570
    },
    {
      "epoch": 1.1985170478718017,
      "grad_norm": 0.6665108799934387,
      "learning_rate": 0.00013116809609575134,
      "loss": 1.0654,
      "step": 29580
    },
    {
      "epoch": 1.198922226048905,
      "grad_norm": 0.6651536226272583,
      "learning_rate": 0.00013112774864748621,
      "loss": 1.1456,
      "step": 29590
    },
    {
      "epoch": 1.1993274042260085,
      "grad_norm": 0.5908018946647644,
      "learning_rate": 0.00013108739558743125,
      "loss": 1.103,
      "step": 29600
    },
    {
      "epoch": 1.1997325824031118,
      "grad_norm": 0.6155626773834229,
      "learning_rate": 0.0001310470369228615,
      "loss": 1.11,
      "step": 29610
    },
    {
      "epoch": 1.200137760580215,
      "grad_norm": 0.6242784857749939,
      "learning_rate": 0.00013100667266105283,
      "loss": 1.0841,
      "step": 29620
    },
    {
      "epoch": 1.2005429387573185,
      "grad_norm": 0.6119043827056885,
      "learning_rate": 0.00013096630280928223,
      "loss": 1.0816,
      "step": 29630
    },
    {
      "epoch": 1.2009481169344218,
      "grad_norm": 0.6744487285614014,
      "learning_rate": 0.00013092592737482775,
      "loss": 1.0819,
      "step": 29640
    },
    {
      "epoch": 1.2013532951115253,
      "grad_norm": 0.5834386944770813,
      "learning_rate": 0.00013088554636496826,
      "loss": 1.086,
      "step": 29650
    },
    {
      "epoch": 1.2017584732886286,
      "grad_norm": 0.6427997350692749,
      "learning_rate": 0.0001308451597869839,
      "loss": 1.1136,
      "step": 29660
    },
    {
      "epoch": 1.2021636514657321,
      "grad_norm": 0.5908557772636414,
      "learning_rate": 0.00013080476764815554,
      "loss": 1.0942,
      "step": 29670
    },
    {
      "epoch": 1.2025688296428354,
      "grad_norm": 0.6264598369598389,
      "learning_rate": 0.0001307643699557652,
      "loss": 1.1395,
      "step": 29680
    },
    {
      "epoch": 1.202974007819939,
      "grad_norm": 0.7836688756942749,
      "learning_rate": 0.0001307239667170959,
      "loss": 1.0847,
      "step": 29690
    },
    {
      "epoch": 1.2033791859970422,
      "grad_norm": 0.5812416076660156,
      "learning_rate": 0.00013068355793943167,
      "loss": 1.1037,
      "step": 29700
    },
    {
      "epoch": 1.2037843641741457,
      "grad_norm": 0.6012585759162903,
      "learning_rate": 0.00013064314363005745,
      "loss": 1.0973,
      "step": 29710
    },
    {
      "epoch": 1.204189542351249,
      "grad_norm": 0.6259954571723938,
      "learning_rate": 0.00013060272379625926,
      "loss": 1.0824,
      "step": 29720
    },
    {
      "epoch": 1.2045947205283523,
      "grad_norm": 0.6375868916511536,
      "learning_rate": 0.0001305622984453241,
      "loss": 1.0733,
      "step": 29730
    },
    {
      "epoch": 1.2049998987054558,
      "grad_norm": 0.6478286385536194,
      "learning_rate": 0.00013052186758453994,
      "loss": 1.1014,
      "step": 29740
    },
    {
      "epoch": 1.205405076882559,
      "grad_norm": 0.611178994178772,
      "learning_rate": 0.00013048143122119577,
      "loss": 1.0975,
      "step": 29750
    },
    {
      "epoch": 1.2058102550596626,
      "grad_norm": 0.6040990948677063,
      "learning_rate": 0.00013044098936258153,
      "loss": 1.0819,
      "step": 29760
    },
    {
      "epoch": 1.2062154332367658,
      "grad_norm": 0.6225639581680298,
      "learning_rate": 0.00013040054201598822,
      "loss": 1.0997,
      "step": 29770
    },
    {
      "epoch": 1.2066206114138693,
      "grad_norm": 0.6219396591186523,
      "learning_rate": 0.0001303600891887078,
      "loss": 1.1592,
      "step": 29780
    },
    {
      "epoch": 1.2070257895909726,
      "grad_norm": 0.6626801490783691,
      "learning_rate": 0.00013031963088803314,
      "loss": 1.0978,
      "step": 29790
    },
    {
      "epoch": 1.207430967768076,
      "grad_norm": 0.6044269800186157,
      "learning_rate": 0.00013027916712125826,
      "loss": 1.0894,
      "step": 29800
    },
    {
      "epoch": 1.2078361459451794,
      "grad_norm": 0.5654250383377075,
      "learning_rate": 0.00013023869789567798,
      "loss": 1.0719,
      "step": 29810
    },
    {
      "epoch": 1.2082413241222827,
      "grad_norm": 0.5815960168838501,
      "learning_rate": 0.00013019822321858824,
      "loss": 1.0681,
      "step": 29820
    },
    {
      "epoch": 1.2086465022993862,
      "grad_norm": 0.5715739130973816,
      "learning_rate": 0.00013015774309728596,
      "loss": 1.0799,
      "step": 29830
    },
    {
      "epoch": 1.2090516804764895,
      "grad_norm": 0.5822177529335022,
      "learning_rate": 0.00013011725753906892,
      "loss": 1.1434,
      "step": 29840
    },
    {
      "epoch": 1.209456858653593,
      "grad_norm": 0.5971710681915283,
      "learning_rate": 0.00013007676655123597,
      "loss": 1.0819,
      "step": 29850
    },
    {
      "epoch": 1.2098620368306963,
      "grad_norm": 0.5121126770973206,
      "learning_rate": 0.000130036270141087,
      "loss": 1.0992,
      "step": 29860
    },
    {
      "epoch": 1.2102672150077998,
      "grad_norm": 0.6323240995407104,
      "learning_rate": 0.00012999576831592273,
      "loss": 1.096,
      "step": 29870
    },
    {
      "epoch": 1.210672393184903,
      "grad_norm": 0.6306718587875366,
      "learning_rate": 0.00012995526108304495,
      "loss": 1.1176,
      "step": 29880
    },
    {
      "epoch": 1.2110775713620066,
      "grad_norm": 0.6558603048324585,
      "learning_rate": 0.00012991474844975645,
      "loss": 1.1148,
      "step": 29890
    },
    {
      "epoch": 1.2114827495391098,
      "grad_norm": 0.6298867464065552,
      "learning_rate": 0.00012987423042336088,
      "loss": 1.1013,
      "step": 29900
    },
    {
      "epoch": 1.2118879277162131,
      "grad_norm": 0.6055012345314026,
      "learning_rate": 0.00012983370701116298,
      "loss": 1.0947,
      "step": 29910
    },
    {
      "epoch": 1.2122931058933166,
      "grad_norm": 0.5551670789718628,
      "learning_rate": 0.00012979317822046842,
      "loss": 1.0586,
      "step": 29920
    },
    {
      "epoch": 1.21269828407042,
      "grad_norm": 0.6099047660827637,
      "learning_rate": 0.0001297526440585838,
      "loss": 1.0937,
      "step": 29930
    },
    {
      "epoch": 1.2131034622475234,
      "grad_norm": 0.5982128381729126,
      "learning_rate": 0.00012971210453281674,
      "loss": 1.0902,
      "step": 29940
    },
    {
      "epoch": 1.2135086404246267,
      "grad_norm": 0.5956372618675232,
      "learning_rate": 0.00012967155965047585,
      "loss": 1.0744,
      "step": 29950
    },
    {
      "epoch": 1.2139138186017302,
      "grad_norm": 0.5669245719909668,
      "learning_rate": 0.00012963100941887053,
      "loss": 1.0924,
      "step": 29960
    },
    {
      "epoch": 1.2143189967788335,
      "grad_norm": 0.6116385459899902,
      "learning_rate": 0.00012959045384531147,
      "loss": 1.0946,
      "step": 29970
    },
    {
      "epoch": 1.2147241749559368,
      "grad_norm": 0.5720478296279907,
      "learning_rate": 0.00012954989293710998,
      "loss": 1.1301,
      "step": 29980
    },
    {
      "epoch": 1.2151293531330403,
      "grad_norm": 0.590000331401825,
      "learning_rate": 0.00012950932670157856,
      "loss": 1.0983,
      "step": 29990
    },
    {
      "epoch": 1.2155345313101436,
      "grad_norm": 0.6344233155250549,
      "learning_rate": 0.00012946875514603057,
      "loss": 1.1299,
      "step": 30000
    },
    {
      "epoch": 1.215939709487247,
      "grad_norm": 0.6032997965812683,
      "learning_rate": 0.00012942817827778038,
      "loss": 1.1011,
      "step": 30010
    },
    {
      "epoch": 1.2163448876643503,
      "grad_norm": 0.5429093837738037,
      "learning_rate": 0.00012938759610414322,
      "loss": 1.1145,
      "step": 30020
    },
    {
      "epoch": 1.2167500658414538,
      "grad_norm": 0.6288044452667236,
      "learning_rate": 0.0001293470086324354,
      "loss": 1.115,
      "step": 30030
    },
    {
      "epoch": 1.2171552440185571,
      "grad_norm": 0.6403694152832031,
      "learning_rate": 0.00012930641586997416,
      "loss": 1.1001,
      "step": 30040
    },
    {
      "epoch": 1.2175604221956606,
      "grad_norm": 0.6623139381408691,
      "learning_rate": 0.0001292658178240776,
      "loss": 1.1194,
      "step": 30050
    },
    {
      "epoch": 1.217965600372764,
      "grad_norm": 0.5643832087516785,
      "learning_rate": 0.0001292252145020649,
      "loss": 1.0692,
      "step": 30060
    },
    {
      "epoch": 1.2183707785498674,
      "grad_norm": 0.5656167268753052,
      "learning_rate": 0.00012918460591125608,
      "loss": 1.1237,
      "step": 30070
    },
    {
      "epoch": 1.2187759567269707,
      "grad_norm": 0.5895429849624634,
      "learning_rate": 0.0001291439920589722,
      "loss": 1.0908,
      "step": 30080
    },
    {
      "epoch": 1.219181134904074,
      "grad_norm": 0.6425740718841553,
      "learning_rate": 0.0001291033729525352,
      "loss": 1.0932,
      "step": 30090
    },
    {
      "epoch": 1.2195863130811775,
      "grad_norm": 0.5590290427207947,
      "learning_rate": 0.00012906274859926802,
      "loss": 1.1098,
      "step": 30100
    },
    {
      "epoch": 1.2199914912582808,
      "grad_norm": 0.5793080925941467,
      "learning_rate": 0.0001290221190064945,
      "loss": 1.0783,
      "step": 30110
    },
    {
      "epoch": 1.2203966694353843,
      "grad_norm": 0.5939346551895142,
      "learning_rate": 0.00012898148418153947,
      "loss": 1.0935,
      "step": 30120
    },
    {
      "epoch": 1.2208018476124876,
      "grad_norm": 0.638044536113739,
      "learning_rate": 0.00012894084413172865,
      "loss": 1.1251,
      "step": 30130
    },
    {
      "epoch": 1.221207025789591,
      "grad_norm": 0.5803507566452026,
      "learning_rate": 0.00012890019886438876,
      "loss": 1.105,
      "step": 30140
    },
    {
      "epoch": 1.2216122039666943,
      "grad_norm": 0.5602997541427612,
      "learning_rate": 0.00012885954838684743,
      "loss": 1.0946,
      "step": 30150
    },
    {
      "epoch": 1.2220173821437976,
      "grad_norm": 0.6194604635238647,
      "learning_rate": 0.0001288188927064332,
      "loss": 1.0893,
      "step": 30160
    },
    {
      "epoch": 1.2224225603209011,
      "grad_norm": 0.568237841129303,
      "learning_rate": 0.0001287782318304756,
      "loss": 1.0843,
      "step": 30170
    },
    {
      "epoch": 1.2228277384980044,
      "grad_norm": 0.5815658569335938,
      "learning_rate": 0.00012873756576630509,
      "loss": 1.0714,
      "step": 30180
    },
    {
      "epoch": 1.223232916675108,
      "grad_norm": 0.5973024964332581,
      "learning_rate": 0.00012869689452125305,
      "loss": 1.0943,
      "step": 30190
    },
    {
      "epoch": 1.2236380948522112,
      "grad_norm": 0.6293757557868958,
      "learning_rate": 0.00012865621810265177,
      "loss": 1.0908,
      "step": 30200
    },
    {
      "epoch": 1.2240432730293147,
      "grad_norm": 0.5657309889793396,
      "learning_rate": 0.0001286155365178345,
      "loss": 1.0962,
      "step": 30210
    },
    {
      "epoch": 1.224448451206418,
      "grad_norm": 0.6488863229751587,
      "learning_rate": 0.00012857484977413545,
      "loss": 1.139,
      "step": 30220
    },
    {
      "epoch": 1.2248536293835215,
      "grad_norm": 0.5684194564819336,
      "learning_rate": 0.0001285341578788897,
      "loss": 1.0946,
      "step": 30230
    },
    {
      "epoch": 1.2252588075606248,
      "grad_norm": 0.5793749094009399,
      "learning_rate": 0.00012849346083943335,
      "loss": 1.0885,
      "step": 30240
    },
    {
      "epoch": 1.2256639857377283,
      "grad_norm": 0.5779211521148682,
      "learning_rate": 0.00012845275866310324,
      "loss": 1.087,
      "step": 30250
    },
    {
      "epoch": 1.2260691639148316,
      "grad_norm": 0.6455779671669006,
      "learning_rate": 0.0001284120513572374,
      "loss": 1.1047,
      "step": 30260
    },
    {
      "epoch": 1.2264743420919348,
      "grad_norm": 0.5469922423362732,
      "learning_rate": 0.00012837133892917456,
      "loss": 1.1247,
      "step": 30270
    },
    {
      "epoch": 1.2268795202690383,
      "grad_norm": 0.5907217264175415,
      "learning_rate": 0.0001283306213862545,
      "loss": 1.0679,
      "step": 30280
    },
    {
      "epoch": 1.2272846984461416,
      "grad_norm": 0.6749882698059082,
      "learning_rate": 0.00012828989873581785,
      "loss": 1.087,
      "step": 30290
    },
    {
      "epoch": 1.2276898766232451,
      "grad_norm": 0.5751379132270813,
      "learning_rate": 0.0001282491709852062,
      "loss": 1.1107,
      "step": 30300
    },
    {
      "epoch": 1.2280950548003484,
      "grad_norm": 0.6100271344184875,
      "learning_rate": 0.00012820843814176212,
      "loss": 1.0884,
      "step": 30310
    },
    {
      "epoch": 1.228500232977452,
      "grad_norm": 0.6675779819488525,
      "learning_rate": 0.00012816770021282893,
      "loss": 1.1047,
      "step": 30320
    },
    {
      "epoch": 1.2289054111545552,
      "grad_norm": 0.5774780511856079,
      "learning_rate": 0.000128126957205751,
      "loss": 1.084,
      "step": 30330
    },
    {
      "epoch": 1.2293105893316585,
      "grad_norm": 0.5787057280540466,
      "learning_rate": 0.00012808620912787365,
      "loss": 1.124,
      "step": 30340
    },
    {
      "epoch": 1.229715767508762,
      "grad_norm": 0.6577019691467285,
      "learning_rate": 0.00012804545598654296,
      "loss": 1.109,
      "step": 30350
    },
    {
      "epoch": 1.2301209456858653,
      "grad_norm": 0.5465518832206726,
      "learning_rate": 0.00012800469778910601,
      "loss": 1.1039,
      "step": 30360
    },
    {
      "epoch": 1.2305261238629688,
      "grad_norm": 0.773047924041748,
      "learning_rate": 0.0001279639345429109,
      "loss": 1.0232,
      "step": 30370
    },
    {
      "epoch": 1.230931302040072,
      "grad_norm": 0.6134364008903503,
      "learning_rate": 0.00012792316625530637,
      "loss": 1.095,
      "step": 30380
    },
    {
      "epoch": 1.2313364802171756,
      "grad_norm": 0.5867806673049927,
      "learning_rate": 0.0001278823929336424,
      "loss": 1.0793,
      "step": 30390
    },
    {
      "epoch": 1.2317416583942788,
      "grad_norm": 0.65009605884552,
      "learning_rate": 0.00012784161458526954,
      "loss": 1.1129,
      "step": 30400
    },
    {
      "epoch": 1.2321468365713824,
      "grad_norm": 0.594313383102417,
      "learning_rate": 0.00012780083121753956,
      "loss": 1.106,
      "step": 30410
    },
    {
      "epoch": 1.2325520147484856,
      "grad_norm": 0.580213725566864,
      "learning_rate": 0.00012776004283780493,
      "loss": 1.0783,
      "step": 30420
    },
    {
      "epoch": 1.2329571929255891,
      "grad_norm": 0.5446280837059021,
      "learning_rate": 0.00012771924945341906,
      "loss": 1.1156,
      "step": 30430
    },
    {
      "epoch": 1.2333623711026924,
      "grad_norm": 0.6494799256324768,
      "learning_rate": 0.0001276784510717363,
      "loss": 1.0636,
      "step": 30440
    },
    {
      "epoch": 1.2337675492797957,
      "grad_norm": 0.603540301322937,
      "learning_rate": 0.00012763764770011194,
      "loss": 1.0924,
      "step": 30450
    },
    {
      "epoch": 1.2341727274568992,
      "grad_norm": 0.6213786602020264,
      "learning_rate": 0.00012759683934590203,
      "loss": 1.1272,
      "step": 30460
    },
    {
      "epoch": 1.2345779056340025,
      "grad_norm": 0.5461667776107788,
      "learning_rate": 0.00012755602601646364,
      "loss": 1.0414,
      "step": 30470
    },
    {
      "epoch": 1.234983083811106,
      "grad_norm": 0.5774096846580505,
      "learning_rate": 0.00012751520771915473,
      "loss": 1.1117,
      "step": 30480
    },
    {
      "epoch": 1.2353882619882093,
      "grad_norm": 0.6750566959381104,
      "learning_rate": 0.00012747438446133409,
      "loss": 1.1038,
      "step": 30490
    },
    {
      "epoch": 1.2357934401653128,
      "grad_norm": 0.6023582816123962,
      "learning_rate": 0.00012743355625036143,
      "loss": 1.064,
      "step": 30500
    },
    {
      "epoch": 1.236198618342416,
      "grad_norm": 0.5705142617225647,
      "learning_rate": 0.0001273927230935974,
      "loss": 1.065,
      "step": 30510
    },
    {
      "epoch": 1.2366037965195193,
      "grad_norm": 0.5703936219215393,
      "learning_rate": 0.00012735188499840355,
      "loss": 1.1301,
      "step": 30520
    },
    {
      "epoch": 1.2370089746966229,
      "grad_norm": 0.5608184337615967,
      "learning_rate": 0.0001273110419721421,
      "loss": 1.1306,
      "step": 30530
    },
    {
      "epoch": 1.2374141528737261,
      "grad_norm": 0.5858995914459229,
      "learning_rate": 0.00012727019402217655,
      "loss": 1.0857,
      "step": 30540
    },
    {
      "epoch": 1.2378193310508296,
      "grad_norm": 0.6597129106521606,
      "learning_rate": 0.00012722934115587094,
      "loss": 1.1236,
      "step": 30550
    },
    {
      "epoch": 1.238224509227933,
      "grad_norm": 0.6412172317504883,
      "learning_rate": 0.00012718848338059036,
      "loss": 1.1193,
      "step": 30560
    },
    {
      "epoch": 1.2386296874050364,
      "grad_norm": 0.6100867986679077,
      "learning_rate": 0.00012714762070370077,
      "loss": 1.0871,
      "step": 30570
    },
    {
      "epoch": 1.2390348655821397,
      "grad_norm": 0.5826025605201721,
      "learning_rate": 0.00012710675313256896,
      "loss": 1.1014,
      "step": 30580
    },
    {
      "epoch": 1.2394400437592432,
      "grad_norm": 0.6084961295127869,
      "learning_rate": 0.0001270658806745627,
      "loss": 1.0542,
      "step": 30590
    },
    {
      "epoch": 1.2398452219363465,
      "grad_norm": 0.5925716161727905,
      "learning_rate": 0.0001270250033370505,
      "loss": 1.0915,
      "step": 30600
    },
    {
      "epoch": 1.24025040011345,
      "grad_norm": 0.588134229183197,
      "learning_rate": 0.0001269841211274019,
      "loss": 1.0922,
      "step": 30610
    },
    {
      "epoch": 1.2406555782905533,
      "grad_norm": 0.5945121645927429,
      "learning_rate": 0.00012694323405298722,
      "loss": 1.117,
      "step": 30620
    },
    {
      "epoch": 1.2410607564676566,
      "grad_norm": 0.6090722680091858,
      "learning_rate": 0.00012690234212117771,
      "loss": 1.0713,
      "step": 30630
    },
    {
      "epoch": 1.24146593464476,
      "grad_norm": 0.5968718528747559,
      "learning_rate": 0.0001268614453393454,
      "loss": 1.1518,
      "step": 30640
    },
    {
      "epoch": 1.2418711128218634,
      "grad_norm": 0.5719189643859863,
      "learning_rate": 0.00012682054371486337,
      "loss": 1.0915,
      "step": 30650
    },
    {
      "epoch": 1.2422762909989669,
      "grad_norm": 0.5613201260566711,
      "learning_rate": 0.00012677963725510542,
      "loss": 1.1074,
      "step": 30660
    },
    {
      "epoch": 1.2426814691760701,
      "grad_norm": 0.6786828637123108,
      "learning_rate": 0.00012673872596744622,
      "loss": 1.1198,
      "step": 30670
    },
    {
      "epoch": 1.2430866473531736,
      "grad_norm": 0.705791711807251,
      "learning_rate": 0.00012669780985926146,
      "loss": 1.1181,
      "step": 30680
    },
    {
      "epoch": 1.243491825530277,
      "grad_norm": 0.5801306962966919,
      "learning_rate": 0.0001266568889379275,
      "loss": 1.1201,
      "step": 30690
    },
    {
      "epoch": 1.2438970037073802,
      "grad_norm": 0.5297249555587769,
      "learning_rate": 0.00012661596321082174,
      "loss": 1.1367,
      "step": 30700
    },
    {
      "epoch": 1.2443021818844837,
      "grad_norm": 0.5929053425788879,
      "learning_rate": 0.00012657503268532236,
      "loss": 1.1032,
      "step": 30710
    },
    {
      "epoch": 1.244707360061587,
      "grad_norm": 0.6028868556022644,
      "learning_rate": 0.0001265340973688084,
      "loss": 1.0817,
      "step": 30720
    },
    {
      "epoch": 1.2451125382386905,
      "grad_norm": 0.5754979252815247,
      "learning_rate": 0.0001264931572686598,
      "loss": 1.06,
      "step": 30730
    },
    {
      "epoch": 1.2455177164157938,
      "grad_norm": 0.6811994910240173,
      "learning_rate": 0.0001264522123922573,
      "loss": 1.1295,
      "step": 30740
    },
    {
      "epoch": 1.2459228945928973,
      "grad_norm": 0.5736775398254395,
      "learning_rate": 0.00012641126274698264,
      "loss": 1.1097,
      "step": 30750
    },
    {
      "epoch": 1.2463280727700006,
      "grad_norm": 0.6453863978385925,
      "learning_rate": 0.00012637030834021824,
      "loss": 1.1013,
      "step": 30760
    },
    {
      "epoch": 1.246733250947104,
      "grad_norm": 0.5426545143127441,
      "learning_rate": 0.00012632934917934752,
      "loss": 1.0831,
      "step": 30770
    },
    {
      "epoch": 1.2471384291242074,
      "grad_norm": 0.6085041761398315,
      "learning_rate": 0.00012628838527175464,
      "loss": 1.0816,
      "step": 30780
    },
    {
      "epoch": 1.2475436073013109,
      "grad_norm": 0.719394862651825,
      "learning_rate": 0.00012624741662482477,
      "loss": 1.0895,
      "step": 30790
    },
    {
      "epoch": 1.2479487854784141,
      "grad_norm": 0.5360556244850159,
      "learning_rate": 0.00012620644324594374,
      "loss": 1.0991,
      "step": 30800
    },
    {
      "epoch": 1.2483539636555174,
      "grad_norm": 0.6382176876068115,
      "learning_rate": 0.0001261654651424984,
      "loss": 1.117,
      "step": 30810
    },
    {
      "epoch": 1.248759141832621,
      "grad_norm": 0.630162239074707,
      "learning_rate": 0.0001261244823218764,
      "loss": 1.1089,
      "step": 30820
    },
    {
      "epoch": 1.2491643200097242,
      "grad_norm": 0.6838631629943848,
      "learning_rate": 0.00012608349479146616,
      "loss": 1.1143,
      "step": 30830
    },
    {
      "epoch": 1.2495694981868277,
      "grad_norm": 0.6155714392662048,
      "learning_rate": 0.00012604250255865703,
      "loss": 1.098,
      "step": 30840
    },
    {
      "epoch": 1.249974676363931,
      "grad_norm": 0.6404242515563965,
      "learning_rate": 0.00012600150563083927,
      "loss": 1.1033,
      "step": 30850
    },
    {
      "epoch": 1.2503798545410345,
      "grad_norm": 0.5593951940536499,
      "learning_rate": 0.00012596050401540384,
      "loss": 1.1348,
      "step": 30860
    },
    {
      "epoch": 1.2507850327181378,
      "grad_norm": 0.6172027587890625,
      "learning_rate": 0.00012591949771974264,
      "loss": 1.0745,
      "step": 30870
    },
    {
      "epoch": 1.251190210895241,
      "grad_norm": 0.6594313979148865,
      "learning_rate": 0.0001258784867512484,
      "loss": 1.1134,
      "step": 30880
    },
    {
      "epoch": 1.2515953890723446,
      "grad_norm": 0.6468364596366882,
      "learning_rate": 0.0001258374711173146,
      "loss": 1.097,
      "step": 30890
    },
    {
      "epoch": 1.252000567249448,
      "grad_norm": 0.6392737030982971,
      "learning_rate": 0.0001257964508253358,
      "loss": 1.1377,
      "step": 30900
    },
    {
      "epoch": 1.2524057454265514,
      "grad_norm": 0.585642397403717,
      "learning_rate": 0.00012575542588270708,
      "loss": 1.0884,
      "step": 30910
    },
    {
      "epoch": 1.2528109236036546,
      "grad_norm": 0.5789276361465454,
      "learning_rate": 0.0001257143962968246,
      "loss": 1.0707,
      "step": 30920
    },
    {
      "epoch": 1.2532161017807582,
      "grad_norm": 0.6581063866615295,
      "learning_rate": 0.0001256733620750853,
      "loss": 1.0864,
      "step": 30930
    },
    {
      "epoch": 1.2536212799578614,
      "grad_norm": 0.7161687016487122,
      "learning_rate": 0.00012563232322488688,
      "loss": 1.0503,
      "step": 30940
    },
    {
      "epoch": 1.2540264581349647,
      "grad_norm": 0.6011103391647339,
      "learning_rate": 0.00012559127975362796,
      "loss": 1.0777,
      "step": 30950
    },
    {
      "epoch": 1.2544316363120682,
      "grad_norm": 0.7221575975418091,
      "learning_rate": 0.00012555023166870797,
      "loss": 1.0937,
      "step": 30960
    },
    {
      "epoch": 1.2548368144891717,
      "grad_norm": 0.6487327218055725,
      "learning_rate": 0.00012550917897752713,
      "loss": 1.0882,
      "step": 30970
    },
    {
      "epoch": 1.255241992666275,
      "grad_norm": 0.5947148203849792,
      "learning_rate": 0.00012546812168748653,
      "loss": 1.1327,
      "step": 30980
    },
    {
      "epoch": 1.2556471708433783,
      "grad_norm": 0.6562842130661011,
      "learning_rate": 0.00012542705980598813,
      "loss": 1.0788,
      "step": 30990
    },
    {
      "epoch": 1.2560523490204818,
      "grad_norm": 0.6345022916793823,
      "learning_rate": 0.0001253859933404346,
      "loss": 1.0968,
      "step": 31000
    },
    {
      "epoch": 1.256457527197585,
      "grad_norm": 0.6572739481925964,
      "learning_rate": 0.00012534492229822954,
      "loss": 1.1151,
      "step": 31010
    },
    {
      "epoch": 1.2568627053746886,
      "grad_norm": 0.6440476179122925,
      "learning_rate": 0.00012530384668677736,
      "loss": 1.1382,
      "step": 31020
    },
    {
      "epoch": 1.2572678835517919,
      "grad_norm": 0.5958414673805237,
      "learning_rate": 0.00012526276651348325,
      "loss": 1.1356,
      "step": 31030
    },
    {
      "epoch": 1.2576730617288954,
      "grad_norm": 0.5482903122901917,
      "learning_rate": 0.00012522168178575327,
      "loss": 1.0801,
      "step": 31040
    },
    {
      "epoch": 1.2580782399059987,
      "grad_norm": 0.6429648399353027,
      "learning_rate": 0.00012518059251099428,
      "loss": 1.1273,
      "step": 31050
    },
    {
      "epoch": 1.258483418083102,
      "grad_norm": 0.6014103889465332,
      "learning_rate": 0.0001251394986966139,
      "loss": 1.1049,
      "step": 31060
    },
    {
      "epoch": 1.2588885962602054,
      "grad_norm": 0.620859682559967,
      "learning_rate": 0.00012509840035002071,
      "loss": 1.062,
      "step": 31070
    },
    {
      "epoch": 1.259293774437309,
      "grad_norm": 0.6533938646316528,
      "learning_rate": 0.000125057297478624,
      "loss": 1.1295,
      "step": 31080
    },
    {
      "epoch": 1.2596989526144122,
      "grad_norm": 0.6291989684104919,
      "learning_rate": 0.00012501619008983386,
      "loss": 1.1038,
      "step": 31090
    },
    {
      "epoch": 1.2601041307915155,
      "grad_norm": 0.5508863925933838,
      "learning_rate": 0.00012497507819106125,
      "loss": 1.1038,
      "step": 31100
    },
    {
      "epoch": 1.260509308968619,
      "grad_norm": 0.5651665329933167,
      "learning_rate": 0.00012493396178971797,
      "loss": 1.1056,
      "step": 31110
    },
    {
      "epoch": 1.2609144871457223,
      "grad_norm": 0.6471253037452698,
      "learning_rate": 0.00012489284089321653,
      "loss": 1.1177,
      "step": 31120
    },
    {
      "epoch": 1.2613196653228256,
      "grad_norm": 0.5450230836868286,
      "learning_rate": 0.00012485171550897037,
      "loss": 1.0804,
      "step": 31130
    },
    {
      "epoch": 1.261724843499929,
      "grad_norm": 0.6272131204605103,
      "learning_rate": 0.00012481058564439365,
      "loss": 1.1195,
      "step": 31140
    },
    {
      "epoch": 1.2621300216770326,
      "grad_norm": 0.6076406836509705,
      "learning_rate": 0.00012476945130690135,
      "loss": 1.0928,
      "step": 31150
    },
    {
      "epoch": 1.2625351998541359,
      "grad_norm": 0.6179577112197876,
      "learning_rate": 0.0001247283125039093,
      "loss": 1.109,
      "step": 31160
    },
    {
      "epoch": 1.2629403780312392,
      "grad_norm": 0.6074665188789368,
      "learning_rate": 0.0001246871692428341,
      "loss": 1.1323,
      "step": 31170
    },
    {
      "epoch": 1.2633455562083427,
      "grad_norm": 0.6576475501060486,
      "learning_rate": 0.00012464602153109314,
      "loss": 1.131,
      "step": 31180
    },
    {
      "epoch": 1.263750734385446,
      "grad_norm": 0.5725626349449158,
      "learning_rate": 0.0001246048693761047,
      "loss": 1.0908,
      "step": 31190
    },
    {
      "epoch": 1.2641559125625494,
      "grad_norm": 0.6094080209732056,
      "learning_rate": 0.0001245637127852877,
      "loss": 1.0917,
      "step": 31200
    },
    {
      "epoch": 1.2645610907396527,
      "grad_norm": 0.5648010969161987,
      "learning_rate": 0.00012452255176606202,
      "loss": 1.0939,
      "step": 31210
    },
    {
      "epoch": 1.2649662689167562,
      "grad_norm": 0.6869997382164001,
      "learning_rate": 0.00012448138632584828,
      "loss": 1.1298,
      "step": 31220
    },
    {
      "epoch": 1.2653714470938595,
      "grad_norm": 0.5864867568016052,
      "learning_rate": 0.00012444021647206783,
      "loss": 1.0645,
      "step": 31230
    },
    {
      "epoch": 1.2657766252709628,
      "grad_norm": 0.661052405834198,
      "learning_rate": 0.00012439904221214293,
      "loss": 1.0913,
      "step": 31240
    },
    {
      "epoch": 1.2661818034480663,
      "grad_norm": 0.6200491786003113,
      "learning_rate": 0.00012435786355349658,
      "loss": 1.0931,
      "step": 31250
    },
    {
      "epoch": 1.2665869816251698,
      "grad_norm": 0.556620180606842,
      "learning_rate": 0.0001243166805035525,
      "loss": 1.1181,
      "step": 31260
    },
    {
      "epoch": 1.266992159802273,
      "grad_norm": 0.6608259081840515,
      "learning_rate": 0.0001242754930697354,
      "loss": 1.1201,
      "step": 31270
    },
    {
      "epoch": 1.2673973379793764,
      "grad_norm": 0.6055263876914978,
      "learning_rate": 0.00012423430125947058,
      "loss": 1.1075,
      "step": 31280
    },
    {
      "epoch": 1.2678025161564799,
      "grad_norm": 0.6220522522926331,
      "learning_rate": 0.00012419310508018417,
      "loss": 1.0914,
      "step": 31290
    },
    {
      "epoch": 1.2682076943335832,
      "grad_norm": 0.549385666847229,
      "learning_rate": 0.0001241519045393032,
      "loss": 1.1236,
      "step": 31300
    },
    {
      "epoch": 1.2686128725106864,
      "grad_norm": 0.6267966628074646,
      "learning_rate": 0.0001241106996442554,
      "loss": 1.123,
      "step": 31310
    },
    {
      "epoch": 1.26901805068779,
      "grad_norm": 0.6122318506240845,
      "learning_rate": 0.0001240694904024692,
      "loss": 1.0955,
      "step": 31320
    },
    {
      "epoch": 1.2694232288648934,
      "grad_norm": 0.6156759858131409,
      "learning_rate": 0.00012402827682137402,
      "loss": 1.0768,
      "step": 31330
    },
    {
      "epoch": 1.2698284070419967,
      "grad_norm": 0.5812082290649414,
      "learning_rate": 0.00012398705890839988,
      "loss": 1.0726,
      "step": 31340
    },
    {
      "epoch": 1.2702335852191,
      "grad_norm": 0.5958369374275208,
      "learning_rate": 0.00012394583667097768,
      "loss": 1.1342,
      "step": 31350
    },
    {
      "epoch": 1.2706387633962035,
      "grad_norm": 0.540890097618103,
      "learning_rate": 0.0001239046101165391,
      "loss": 1.075,
      "step": 31360
    },
    {
      "epoch": 1.2710439415733068,
      "grad_norm": 0.6281302571296692,
      "learning_rate": 0.00012386337925251648,
      "loss": 1.1011,
      "step": 31370
    },
    {
      "epoch": 1.2714491197504103,
      "grad_norm": 0.6646365523338318,
      "learning_rate": 0.00012382214408634308,
      "loss": 1.1055,
      "step": 31380
    },
    {
      "epoch": 1.2718542979275136,
      "grad_norm": 0.6200403571128845,
      "learning_rate": 0.0001237809046254529,
      "loss": 1.1256,
      "step": 31390
    },
    {
      "epoch": 1.272259476104617,
      "grad_norm": 0.7050949335098267,
      "learning_rate": 0.00012373966087728065,
      "loss": 1.0905,
      "step": 31400
    },
    {
      "epoch": 1.2726646542817204,
      "grad_norm": 0.5680478811264038,
      "learning_rate": 0.00012369841284926188,
      "loss": 1.0913,
      "step": 31410
    },
    {
      "epoch": 1.2730698324588237,
      "grad_norm": 0.5862569808959961,
      "learning_rate": 0.0001236571605488329,
      "loss": 1.0675,
      "step": 31420
    },
    {
      "epoch": 1.2734750106359272,
      "grad_norm": 0.5319828987121582,
      "learning_rate": 0.0001236159039834308,
      "loss": 1.1234,
      "step": 31430
    },
    {
      "epoch": 1.2738801888130307,
      "grad_norm": 0.6113647222518921,
      "learning_rate": 0.00012357464316049336,
      "loss": 1.0948,
      "step": 31440
    },
    {
      "epoch": 1.274285366990134,
      "grad_norm": 0.5930939316749573,
      "learning_rate": 0.00012353337808745923,
      "loss": 1.0788,
      "step": 31450
    },
    {
      "epoch": 1.2746905451672372,
      "grad_norm": 0.6083719730377197,
      "learning_rate": 0.00012349210877176775,
      "loss": 1.1103,
      "step": 31460
    },
    {
      "epoch": 1.2750957233443407,
      "grad_norm": 0.6151151061058044,
      "learning_rate": 0.00012345083522085907,
      "loss": 1.0985,
      "step": 31470
    },
    {
      "epoch": 1.275500901521444,
      "grad_norm": 0.5977274775505066,
      "learning_rate": 0.00012340955744217412,
      "loss": 1.0941,
      "step": 31480
    },
    {
      "epoch": 1.2759060796985473,
      "grad_norm": 0.5845439434051514,
      "learning_rate": 0.00012336827544315453,
      "loss": 1.0776,
      "step": 31490
    },
    {
      "epoch": 1.2763112578756508,
      "grad_norm": 0.6126797795295715,
      "learning_rate": 0.00012332698923124273,
      "loss": 1.0926,
      "step": 31500
    },
    {
      "epoch": 1.2767164360527543,
      "grad_norm": 0.641761839389801,
      "learning_rate": 0.00012328569881388187,
      "loss": 1.1172,
      "step": 31510
    },
    {
      "epoch": 1.2771216142298576,
      "grad_norm": 0.6006807088851929,
      "learning_rate": 0.00012324440419851597,
      "loss": 1.1067,
      "step": 31520
    },
    {
      "epoch": 1.2775267924069609,
      "grad_norm": 0.5942564606666565,
      "learning_rate": 0.0001232031053925897,
      "loss": 1.1055,
      "step": 31530
    },
    {
      "epoch": 1.2779319705840644,
      "grad_norm": 0.6797270774841309,
      "learning_rate": 0.00012316180240354844,
      "loss": 1.109,
      "step": 31540
    },
    {
      "epoch": 1.2783371487611677,
      "grad_norm": 0.5869057774543762,
      "learning_rate": 0.00012312049523883852,
      "loss": 1.0747,
      "step": 31550
    },
    {
      "epoch": 1.2787423269382712,
      "grad_norm": 0.6110022664070129,
      "learning_rate": 0.00012307918390590684,
      "loss": 1.1205,
      "step": 31560
    },
    {
      "epoch": 1.2791475051153745,
      "grad_norm": 0.6326236724853516,
      "learning_rate": 0.00012303786841220103,
      "loss": 1.0664,
      "step": 31570
    },
    {
      "epoch": 1.279552683292478,
      "grad_norm": 0.5968607068061829,
      "learning_rate": 0.0001229965487651697,
      "loss": 1.0765,
      "step": 31580
    },
    {
      "epoch": 1.2799578614695812,
      "grad_norm": 0.5889349579811096,
      "learning_rate": 0.000122955224972262,
      "loss": 1.1042,
      "step": 31590
    },
    {
      "epoch": 1.2803630396466845,
      "grad_norm": 0.5765251517295837,
      "learning_rate": 0.00012291389704092786,
      "loss": 1.0843,
      "step": 31600
    },
    {
      "epoch": 1.280768217823788,
      "grad_norm": 0.6267028450965881,
      "learning_rate": 0.000122872564978618,
      "loss": 1.1338,
      "step": 31610
    },
    {
      "epoch": 1.2811733960008915,
      "grad_norm": 0.588252604007721,
      "learning_rate": 0.00012283122879278393,
      "loss": 1.0992,
      "step": 31620
    },
    {
      "epoch": 1.2815785741779948,
      "grad_norm": 0.5703197121620178,
      "learning_rate": 0.0001227898884908777,
      "loss": 1.0689,
      "step": 31630
    },
    {
      "epoch": 1.281983752355098,
      "grad_norm": 0.6067803502082825,
      "learning_rate": 0.0001227485440803524,
      "loss": 1.1276,
      "step": 31640
    },
    {
      "epoch": 1.2823889305322016,
      "grad_norm": 0.5919371843338013,
      "learning_rate": 0.0001227071955686616,
      "loss": 1.1073,
      "step": 31650
    },
    {
      "epoch": 1.2827941087093049,
      "grad_norm": 0.5512567162513733,
      "learning_rate": 0.00012266584296325982,
      "loss": 1.1101,
      "step": 31660
    },
    {
      "epoch": 1.2831992868864082,
      "grad_norm": 0.5777183175086975,
      "learning_rate": 0.0001226244862716021,
      "loss": 1.0625,
      "step": 31670
    },
    {
      "epoch": 1.2836044650635117,
      "grad_norm": 0.5966898798942566,
      "learning_rate": 0.0001225831255011444,
      "loss": 1.0987,
      "step": 31680
    },
    {
      "epoch": 1.2840096432406152,
      "grad_norm": 0.6025921106338501,
      "learning_rate": 0.0001225417606593433,
      "loss": 1.1275,
      "step": 31690
    },
    {
      "epoch": 1.2844148214177185,
      "grad_norm": 0.6448807716369629,
      "learning_rate": 0.00012250039175365623,
      "loss": 1.087,
      "step": 31700
    },
    {
      "epoch": 1.2848199995948217,
      "grad_norm": 0.6376199126243591,
      "learning_rate": 0.0001224590187915412,
      "loss": 1.0814,
      "step": 31710
    },
    {
      "epoch": 1.2852251777719252,
      "grad_norm": 0.6112326383590698,
      "learning_rate": 0.00012241764178045706,
      "loss": 1.0943,
      "step": 31720
    },
    {
      "epoch": 1.2856303559490285,
      "grad_norm": 0.6274551153182983,
      "learning_rate": 0.00012237626072786334,
      "loss": 1.1095,
      "step": 31730
    },
    {
      "epoch": 1.286035534126132,
      "grad_norm": 0.5976235866546631,
      "learning_rate": 0.0001223348756412204,
      "loss": 1.0997,
      "step": 31740
    },
    {
      "epoch": 1.2864407123032353,
      "grad_norm": 0.6112478971481323,
      "learning_rate": 0.00012229348652798916,
      "loss": 1.1057,
      "step": 31750
    },
    {
      "epoch": 1.2868458904803388,
      "grad_norm": 0.6166186332702637,
      "learning_rate": 0.00012225209339563145,
      "loss": 1.1099,
      "step": 31760
    },
    {
      "epoch": 1.287251068657442,
      "grad_norm": 0.5692600607872009,
      "learning_rate": 0.00012221069625160963,
      "loss": 1.1011,
      "step": 31770
    },
    {
      "epoch": 1.2876562468345454,
      "grad_norm": 0.5917956233024597,
      "learning_rate": 0.00012216929510338695,
      "loss": 1.1304,
      "step": 31780
    },
    {
      "epoch": 1.2880614250116489,
      "grad_norm": 0.5659837126731873,
      "learning_rate": 0.0001221278899584273,
      "loss": 1.13,
      "step": 31790
    },
    {
      "epoch": 1.2884666031887524,
      "grad_norm": 0.6251693964004517,
      "learning_rate": 0.00012208648082419527,
      "loss": 1.1024,
      "step": 31800
    },
    {
      "epoch": 1.2888717813658557,
      "grad_norm": 0.5650818943977356,
      "learning_rate": 0.00012204506770815625,
      "loss": 1.0684,
      "step": 31810
    },
    {
      "epoch": 1.289276959542959,
      "grad_norm": 0.60479336977005,
      "learning_rate": 0.00012200365061777628,
      "loss": 1.1226,
      "step": 31820
    },
    {
      "epoch": 1.2896821377200625,
      "grad_norm": 0.6291558742523193,
      "learning_rate": 0.00012196222956052214,
      "loss": 1.0994,
      "step": 31830
    },
    {
      "epoch": 1.2900873158971657,
      "grad_norm": 0.5651136636734009,
      "learning_rate": 0.00012192080454386133,
      "loss": 1.0967,
      "step": 31840
    },
    {
      "epoch": 1.290492494074269,
      "grad_norm": 0.5306501984596252,
      "learning_rate": 0.00012187937557526203,
      "loss": 1.0788,
      "step": 31850
    },
    {
      "epoch": 1.2908976722513725,
      "grad_norm": 0.5674635171890259,
      "learning_rate": 0.0001218379426621932,
      "loss": 1.112,
      "step": 31860
    },
    {
      "epoch": 1.291302850428476,
      "grad_norm": 0.5890016555786133,
      "learning_rate": 0.00012179650581212446,
      "loss": 1.1027,
      "step": 31870
    },
    {
      "epoch": 1.2917080286055793,
      "grad_norm": 0.6116552352905273,
      "learning_rate": 0.00012175506503252612,
      "loss": 1.0968,
      "step": 31880
    },
    {
      "epoch": 1.2921132067826826,
      "grad_norm": 0.5268003940582275,
      "learning_rate": 0.00012171362033086928,
      "loss": 1.0589,
      "step": 31890
    },
    {
      "epoch": 1.292518384959786,
      "grad_norm": 0.578721284866333,
      "learning_rate": 0.00012167217171462566,
      "loss": 1.0982,
      "step": 31900
    },
    {
      "epoch": 1.2929235631368894,
      "grad_norm": 0.6328438520431519,
      "learning_rate": 0.00012163071919126772,
      "loss": 1.0814,
      "step": 31910
    },
    {
      "epoch": 1.293328741313993,
      "grad_norm": 0.6472244262695312,
      "learning_rate": 0.00012158926276826866,
      "loss": 1.134,
      "step": 31920
    },
    {
      "epoch": 1.2937339194910962,
      "grad_norm": 0.629347562789917,
      "learning_rate": 0.0001215478024531023,
      "loss": 1.09,
      "step": 31930
    },
    {
      "epoch": 1.2941390976681997,
      "grad_norm": 0.5896032452583313,
      "learning_rate": 0.00012150633825324327,
      "loss": 1.1195,
      "step": 31940
    },
    {
      "epoch": 1.294544275845303,
      "grad_norm": 0.6048210859298706,
      "learning_rate": 0.00012146487017616679,
      "loss": 1.1023,
      "step": 31950
    },
    {
      "epoch": 1.2949494540224062,
      "grad_norm": 0.6039345264434814,
      "learning_rate": 0.00012142339822934885,
      "loss": 1.0411,
      "step": 31960
    },
    {
      "epoch": 1.2953546321995097,
      "grad_norm": 0.5771040916442871,
      "learning_rate": 0.00012138192242026614,
      "loss": 1.0882,
      "step": 31970
    },
    {
      "epoch": 1.2957598103766133,
      "grad_norm": 0.6548736095428467,
      "learning_rate": 0.000121340442756396,
      "loss": 1.0641,
      "step": 31980
    },
    {
      "epoch": 1.2961649885537165,
      "grad_norm": 0.6059533357620239,
      "learning_rate": 0.00012129895924521646,
      "loss": 1.133,
      "step": 31990
    },
    {
      "epoch": 1.2965701667308198,
      "grad_norm": 0.6514201760292053,
      "learning_rate": 0.00012125747189420635,
      "loss": 1.1135,
      "step": 32000
    },
    {
      "epoch": 1.2969753449079233,
      "grad_norm": 0.6272081732749939,
      "learning_rate": 0.00012121598071084505,
      "loss": 1.0999,
      "step": 32010
    },
    {
      "epoch": 1.2973805230850266,
      "grad_norm": 0.6543036103248596,
      "learning_rate": 0.00012117448570261272,
      "loss": 1.1123,
      "step": 32020
    },
    {
      "epoch": 1.2977857012621299,
      "grad_norm": 0.5404976606369019,
      "learning_rate": 0.0001211329868769902,
      "loss": 1.0999,
      "step": 32030
    },
    {
      "epoch": 1.2981908794392334,
      "grad_norm": 0.6430730819702148,
      "learning_rate": 0.00012109148424145898,
      "loss": 1.0679,
      "step": 32040
    },
    {
      "epoch": 1.298596057616337,
      "grad_norm": 0.5755747556686401,
      "learning_rate": 0.00012104997780350123,
      "loss": 1.1325,
      "step": 32050
    },
    {
      "epoch": 1.2990012357934402,
      "grad_norm": 0.6228376030921936,
      "learning_rate": 0.00012100846757059992,
      "loss": 1.0734,
      "step": 32060
    },
    {
      "epoch": 1.2994064139705435,
      "grad_norm": 0.5868873000144958,
      "learning_rate": 0.00012096695355023855,
      "loss": 1.1003,
      "step": 32070
    },
    {
      "epoch": 1.299811592147647,
      "grad_norm": 0.6566130518913269,
      "learning_rate": 0.00012092543574990141,
      "loss": 1.1085,
      "step": 32080
    },
    {
      "epoch": 1.3002167703247502,
      "grad_norm": 0.5663368701934814,
      "learning_rate": 0.00012088391417707341,
      "loss": 1.1186,
      "step": 32090
    },
    {
      "epoch": 1.3006219485018538,
      "grad_norm": 0.6412637233734131,
      "learning_rate": 0.00012084238883924019,
      "loss": 1.0717,
      "step": 32100
    },
    {
      "epoch": 1.301027126678957,
      "grad_norm": 0.6479397416114807,
      "learning_rate": 0.00012080085974388802,
      "loss": 1.1351,
      "step": 32110
    },
    {
      "epoch": 1.3014323048560605,
      "grad_norm": 0.6237925291061401,
      "learning_rate": 0.00012075932689850391,
      "loss": 1.0983,
      "step": 32120
    },
    {
      "epoch": 1.3018374830331638,
      "grad_norm": 0.588298499584198,
      "learning_rate": 0.00012071779031057544,
      "loss": 1.0919,
      "step": 32130
    },
    {
      "epoch": 1.302242661210267,
      "grad_norm": 0.5671430230140686,
      "learning_rate": 0.00012067624998759102,
      "loss": 1.0664,
      "step": 32140
    },
    {
      "epoch": 1.3026478393873706,
      "grad_norm": 0.5968964099884033,
      "learning_rate": 0.00012063470593703959,
      "loss": 1.0967,
      "step": 32150
    },
    {
      "epoch": 1.3030530175644741,
      "grad_norm": 0.541799783706665,
      "learning_rate": 0.0001205931581664108,
      "loss": 1.1135,
      "step": 32160
    },
    {
      "epoch": 1.3034581957415774,
      "grad_norm": 0.6273707747459412,
      "learning_rate": 0.000120551606683195,
      "loss": 1.1046,
      "step": 32170
    },
    {
      "epoch": 1.3038633739186807,
      "grad_norm": 0.6137861609458923,
      "learning_rate": 0.00012051005149488326,
      "loss": 1.0925,
      "step": 32180
    },
    {
      "epoch": 1.3042685520957842,
      "grad_norm": 0.5989969968795776,
      "learning_rate": 0.00012046849260896715,
      "loss": 1.0835,
      "step": 32190
    },
    {
      "epoch": 1.3046737302728875,
      "grad_norm": 0.6230854988098145,
      "learning_rate": 0.0001204269300329391,
      "loss": 1.0811,
      "step": 32200
    },
    {
      "epoch": 1.3050789084499907,
      "grad_norm": 0.592089056968689,
      "learning_rate": 0.00012038536377429208,
      "loss": 1.1069,
      "step": 32210
    },
    {
      "epoch": 1.3054840866270943,
      "grad_norm": 0.6033912301063538,
      "learning_rate": 0.00012034379384051975,
      "loss": 1.1005,
      "step": 32220
    },
    {
      "epoch": 1.3058892648041978,
      "grad_norm": 0.5514757633209229,
      "learning_rate": 0.00012030222023911648,
      "loss": 1.0911,
      "step": 32230
    },
    {
      "epoch": 1.306294442981301,
      "grad_norm": 0.588191568851471,
      "learning_rate": 0.00012026064297757725,
      "loss": 1.117,
      "step": 32240
    },
    {
      "epoch": 1.3066996211584043,
      "grad_norm": 0.589688777923584,
      "learning_rate": 0.00012021906206339766,
      "loss": 1.0715,
      "step": 32250
    },
    {
      "epoch": 1.3071047993355078,
      "grad_norm": 0.5652269721031189,
      "learning_rate": 0.00012017747750407411,
      "loss": 1.1102,
      "step": 32260
    },
    {
      "epoch": 1.307509977512611,
      "grad_norm": 0.6184343695640564,
      "learning_rate": 0.00012013588930710352,
      "loss": 1.1005,
      "step": 32270
    },
    {
      "epoch": 1.3079151556897146,
      "grad_norm": 0.5361155867576599,
      "learning_rate": 0.00012009429747998354,
      "loss": 1.0957,
      "step": 32280
    },
    {
      "epoch": 1.308320333866818,
      "grad_norm": 0.6164462566375732,
      "learning_rate": 0.00012005270203021241,
      "loss": 1.1169,
      "step": 32290
    },
    {
      "epoch": 1.3087255120439214,
      "grad_norm": 0.6401699185371399,
      "learning_rate": 0.0001200111029652891,
      "loss": 1.1016,
      "step": 32300
    },
    {
      "epoch": 1.3091306902210247,
      "grad_norm": 0.5596681833267212,
      "learning_rate": 0.00011996950029271318,
      "loss": 1.0838,
      "step": 32310
    },
    {
      "epoch": 1.309535868398128,
      "grad_norm": 0.5743493437767029,
      "learning_rate": 0.00011992789401998492,
      "loss": 1.1014,
      "step": 32320
    },
    {
      "epoch": 1.3099410465752315,
      "grad_norm": 0.5848032236099243,
      "learning_rate": 0.00011988628415460514,
      "loss": 1.0935,
      "step": 32330
    },
    {
      "epoch": 1.310346224752335,
      "grad_norm": 0.5730831027030945,
      "learning_rate": 0.00011984467070407546,
      "loss": 1.0955,
      "step": 32340
    },
    {
      "epoch": 1.3107514029294383,
      "grad_norm": 0.7438188791275024,
      "learning_rate": 0.00011980305367589797,
      "loss": 1.0974,
      "step": 32350
    },
    {
      "epoch": 1.3111565811065415,
      "grad_norm": 0.6036993861198425,
      "learning_rate": 0.00011976143307757552,
      "loss": 1.0778,
      "step": 32360
    },
    {
      "epoch": 1.311561759283645,
      "grad_norm": 0.6623371243476868,
      "learning_rate": 0.00011971980891661163,
      "loss": 1.0922,
      "step": 32370
    },
    {
      "epoch": 1.3119669374607483,
      "grad_norm": 0.5985866189002991,
      "learning_rate": 0.00011967818120051033,
      "loss": 1.0695,
      "step": 32380
    },
    {
      "epoch": 1.3123721156378516,
      "grad_norm": 0.5753145217895508,
      "learning_rate": 0.00011963654993677645,
      "loss": 1.1142,
      "step": 32390
    },
    {
      "epoch": 1.3127772938149551,
      "grad_norm": 0.6155942678451538,
      "learning_rate": 0.0001195949151329153,
      "loss": 1.0966,
      "step": 32400
    },
    {
      "epoch": 1.3131824719920586,
      "grad_norm": 0.5941824316978455,
      "learning_rate": 0.00011955327679643295,
      "loss": 1.0987,
      "step": 32410
    },
    {
      "epoch": 1.313587650169162,
      "grad_norm": 0.5448503494262695,
      "learning_rate": 0.00011951163493483609,
      "loss": 1.0697,
      "step": 32420
    },
    {
      "epoch": 1.3139928283462652,
      "grad_norm": 0.6044782400131226,
      "learning_rate": 0.00011946998955563198,
      "loss": 1.1185,
      "step": 32430
    },
    {
      "epoch": 1.3143980065233687,
      "grad_norm": 0.6538883447647095,
      "learning_rate": 0.00011942834066632857,
      "loss": 1.0934,
      "step": 32440
    },
    {
      "epoch": 1.314803184700472,
      "grad_norm": 0.6431025862693787,
      "learning_rate": 0.00011938668827443445,
      "loss": 1.0754,
      "step": 32450
    },
    {
      "epoch": 1.3152083628775755,
      "grad_norm": 0.6125227212905884,
      "learning_rate": 0.00011934503238745878,
      "loss": 1.0685,
      "step": 32460
    },
    {
      "epoch": 1.3156135410546788,
      "grad_norm": 0.5892810821533203,
      "learning_rate": 0.00011930337301291139,
      "loss": 1.0896,
      "step": 32470
    },
    {
      "epoch": 1.3160187192317823,
      "grad_norm": 0.6171096563339233,
      "learning_rate": 0.00011926171015830276,
      "loss": 1.1102,
      "step": 32480
    },
    {
      "epoch": 1.3164238974088855,
      "grad_norm": 0.7251404523849487,
      "learning_rate": 0.000119220043831144,
      "loss": 1.1194,
      "step": 32490
    },
    {
      "epoch": 1.3168290755859888,
      "grad_norm": 0.5359848737716675,
      "learning_rate": 0.00011917837403894676,
      "loss": 1.0528,
      "step": 32500
    },
    {
      "epoch": 1.3172342537630923,
      "grad_norm": 0.6592816710472107,
      "learning_rate": 0.00011913670078922344,
      "loss": 1.0936,
      "step": 32510
    },
    {
      "epoch": 1.3176394319401958,
      "grad_norm": 0.6321480870246887,
      "learning_rate": 0.00011909502408948694,
      "loss": 1.1119,
      "step": 32520
    },
    {
      "epoch": 1.3180446101172991,
      "grad_norm": 0.6014195084571838,
      "learning_rate": 0.00011905334394725085,
      "loss": 1.1221,
      "step": 32530
    },
    {
      "epoch": 1.3184497882944024,
      "grad_norm": 0.6689180135726929,
      "learning_rate": 0.00011901166037002944,
      "loss": 1.1156,
      "step": 32540
    },
    {
      "epoch": 1.318854966471506,
      "grad_norm": 0.5454925298690796,
      "learning_rate": 0.00011896997336533746,
      "loss": 1.049,
      "step": 32550
    },
    {
      "epoch": 1.3192601446486092,
      "grad_norm": 0.5246903300285339,
      "learning_rate": 0.00011892828294069036,
      "loss": 1.0831,
      "step": 32560
    },
    {
      "epoch": 1.3196653228257125,
      "grad_norm": 0.5734118819236755,
      "learning_rate": 0.00011888658910360422,
      "loss": 1.1033,
      "step": 32570
    },
    {
      "epoch": 1.320070501002816,
      "grad_norm": 0.6502008438110352,
      "learning_rate": 0.00011884489186159569,
      "loss": 1.0837,
      "step": 32580
    },
    {
      "epoch": 1.3204756791799195,
      "grad_norm": 0.6185864210128784,
      "learning_rate": 0.00011880319122218209,
      "loss": 1.0685,
      "step": 32590
    },
    {
      "epoch": 1.3208808573570228,
      "grad_norm": 0.6125285625457764,
      "learning_rate": 0.00011876148719288128,
      "loss": 1.0915,
      "step": 32600
    },
    {
      "epoch": 1.321286035534126,
      "grad_norm": 0.597996711730957,
      "learning_rate": 0.00011871977978121176,
      "loss": 1.088,
      "step": 32610
    },
    {
      "epoch": 1.3216912137112296,
      "grad_norm": 0.6233706474304199,
      "learning_rate": 0.0001186780689946927,
      "loss": 1.0764,
      "step": 32620
    },
    {
      "epoch": 1.3220963918883328,
      "grad_norm": 0.7090185880661011,
      "learning_rate": 0.00011863635484084377,
      "loss": 1.0891,
      "step": 32630
    },
    {
      "epoch": 1.3225015700654363,
      "grad_norm": 0.5980803370475769,
      "learning_rate": 0.00011859463732718533,
      "loss": 1.0772,
      "step": 32640
    },
    {
      "epoch": 1.3229067482425396,
      "grad_norm": 0.7087922692298889,
      "learning_rate": 0.00011855291646123832,
      "loss": 1.1338,
      "step": 32650
    },
    {
      "epoch": 1.3233119264196431,
      "grad_norm": 0.556007444858551,
      "learning_rate": 0.00011851119225052429,
      "loss": 1.1374,
      "step": 32660
    },
    {
      "epoch": 1.3237171045967464,
      "grad_norm": 0.5996232628822327,
      "learning_rate": 0.00011846946470256538,
      "loss": 1.1193,
      "step": 32670
    },
    {
      "epoch": 1.3241222827738497,
      "grad_norm": 0.6400900483131409,
      "learning_rate": 0.00011842773382488434,
      "loss": 1.0654,
      "step": 32680
    },
    {
      "epoch": 1.3245274609509532,
      "grad_norm": 0.5320407152175903,
      "learning_rate": 0.00011838599962500452,
      "loss": 1.0975,
      "step": 32690
    },
    {
      "epoch": 1.3249326391280567,
      "grad_norm": 0.574620246887207,
      "learning_rate": 0.00011834426211044984,
      "loss": 1.1247,
      "step": 32700
    },
    {
      "epoch": 1.32533781730516,
      "grad_norm": 0.6139172911643982,
      "learning_rate": 0.00011830252128874491,
      "loss": 1.1023,
      "step": 32710
    },
    {
      "epoch": 1.3257429954822633,
      "grad_norm": 0.5568187832832336,
      "learning_rate": 0.00011826077716741477,
      "loss": 1.0693,
      "step": 32720
    },
    {
      "epoch": 1.3261481736593668,
      "grad_norm": 0.6437784433364868,
      "learning_rate": 0.00011821902975398529,
      "loss": 1.1154,
      "step": 32730
    },
    {
      "epoch": 1.32655335183647,
      "grad_norm": 0.6623374819755554,
      "learning_rate": 0.00011817727905598268,
      "loss": 1.1374,
      "step": 32740
    },
    {
      "epoch": 1.3269585300135733,
      "grad_norm": 0.6313092112541199,
      "learning_rate": 0.00011813552508093393,
      "loss": 1.0914,
      "step": 32750
    },
    {
      "epoch": 1.3273637081906768,
      "grad_norm": 0.638087809085846,
      "learning_rate": 0.00011809376783636652,
      "loss": 1.0776,
      "step": 32760
    },
    {
      "epoch": 1.3277688863677803,
      "grad_norm": 0.7047549486160278,
      "learning_rate": 0.0001180520073298086,
      "loss": 1.1221,
      "step": 32770
    },
    {
      "epoch": 1.3281740645448836,
      "grad_norm": 0.5717036724090576,
      "learning_rate": 0.00011801024356878879,
      "loss": 1.072,
      "step": 32780
    },
    {
      "epoch": 1.328579242721987,
      "grad_norm": 0.6844361424446106,
      "learning_rate": 0.0001179684765608364,
      "loss": 1.1401,
      "step": 32790
    },
    {
      "epoch": 1.3289844208990904,
      "grad_norm": 0.5937425494194031,
      "learning_rate": 0.00011792670631348133,
      "loss": 1.0877,
      "step": 32800
    },
    {
      "epoch": 1.3293895990761937,
      "grad_norm": 0.6107401847839355,
      "learning_rate": 0.00011788493283425397,
      "loss": 1.1034,
      "step": 32810
    },
    {
      "epoch": 1.3297947772532972,
      "grad_norm": 0.6256533265113831,
      "learning_rate": 0.00011784315613068537,
      "loss": 1.1158,
      "step": 32820
    },
    {
      "epoch": 1.3301999554304005,
      "grad_norm": 0.5586012005805969,
      "learning_rate": 0.00011780137621030718,
      "loss": 1.1058,
      "step": 32830
    },
    {
      "epoch": 1.330605133607504,
      "grad_norm": 0.7330683469772339,
      "learning_rate": 0.00011775959308065149,
      "loss": 1.0861,
      "step": 32840
    },
    {
      "epoch": 1.3310103117846073,
      "grad_norm": 0.6015393137931824,
      "learning_rate": 0.0001177178067492512,
      "loss": 1.0924,
      "step": 32850
    },
    {
      "epoch": 1.3314154899617106,
      "grad_norm": 0.6007891893386841,
      "learning_rate": 0.00011767601722363955,
      "loss": 1.1402,
      "step": 32860
    },
    {
      "epoch": 1.331820668138814,
      "grad_norm": 0.5249758362770081,
      "learning_rate": 0.00011763422451135049,
      "loss": 1.1355,
      "step": 32870
    },
    {
      "epoch": 1.3322258463159176,
      "grad_norm": 0.6658341288566589,
      "learning_rate": 0.00011759242861991855,
      "loss": 1.1311,
      "step": 32880
    },
    {
      "epoch": 1.3326310244930208,
      "grad_norm": 0.6428666114807129,
      "learning_rate": 0.00011755062955687877,
      "loss": 1.1209,
      "step": 32890
    },
    {
      "epoch": 1.3330362026701241,
      "grad_norm": 0.6379505395889282,
      "learning_rate": 0.0001175088273297668,
      "loss": 1.1088,
      "step": 32900
    },
    {
      "epoch": 1.3334413808472276,
      "grad_norm": 0.5763934850692749,
      "learning_rate": 0.00011746702194611885,
      "loss": 1.0801,
      "step": 32910
    },
    {
      "epoch": 1.333846559024331,
      "grad_norm": 0.6469150185585022,
      "learning_rate": 0.00011742521341347168,
      "loss": 1.0757,
      "step": 32920
    },
    {
      "epoch": 1.3342517372014342,
      "grad_norm": 0.5449137091636658,
      "learning_rate": 0.00011738340173936268,
      "loss": 1.0844,
      "step": 32930
    },
    {
      "epoch": 1.3346569153785377,
      "grad_norm": 0.6381601691246033,
      "learning_rate": 0.00011734158693132978,
      "loss": 1.0981,
      "step": 32940
    },
    {
      "epoch": 1.3350620935556412,
      "grad_norm": 0.617039680480957,
      "learning_rate": 0.00011729976899691137,
      "loss": 1.143,
      "step": 32950
    },
    {
      "epoch": 1.3354672717327445,
      "grad_norm": 0.5878420472145081,
      "learning_rate": 0.00011725794794364657,
      "loss": 1.0923,
      "step": 32960
    },
    {
      "epoch": 1.3358724499098478,
      "grad_norm": 0.5912373065948486,
      "learning_rate": 0.00011721612377907499,
      "loss": 1.0752,
      "step": 32970
    },
    {
      "epoch": 1.3362776280869513,
      "grad_norm": 0.6183993220329285,
      "learning_rate": 0.0001171742965107367,
      "loss": 1.0846,
      "step": 32980
    },
    {
      "epoch": 1.3366828062640546,
      "grad_norm": 0.571692943572998,
      "learning_rate": 0.00011713246614617257,
      "loss": 1.1034,
      "step": 32990
    },
    {
      "epoch": 1.337087984441158,
      "grad_norm": 0.610343337059021,
      "learning_rate": 0.0001170906326929238,
      "loss": 1.0463,
      "step": 33000
    },
    {
      "epoch": 1.3374931626182613,
      "grad_norm": 0.6335545778274536,
      "learning_rate": 0.00011704879615853222,
      "loss": 1.0884,
      "step": 33010
    },
    {
      "epoch": 1.3378983407953648,
      "grad_norm": 0.5846521854400635,
      "learning_rate": 0.00011700695655054026,
      "loss": 1.0693,
      "step": 33020
    },
    {
      "epoch": 1.3383035189724681,
      "grad_norm": 0.5807381868362427,
      "learning_rate": 0.00011696511387649085,
      "loss": 1.0769,
      "step": 33030
    },
    {
      "epoch": 1.3387086971495714,
      "grad_norm": 0.6067166924476624,
      "learning_rate": 0.00011692326814392753,
      "loss": 1.0439,
      "step": 33040
    },
    {
      "epoch": 1.339113875326675,
      "grad_norm": 0.5825990438461304,
      "learning_rate": 0.00011688141936039429,
      "loss": 1.0963,
      "step": 33050
    },
    {
      "epoch": 1.3395190535037784,
      "grad_norm": 0.6306445598602295,
      "learning_rate": 0.00011683956753343584,
      "loss": 1.0951,
      "step": 33060
    },
    {
      "epoch": 1.3399242316808817,
      "grad_norm": 0.5955783128738403,
      "learning_rate": 0.00011679771267059723,
      "loss": 1.0415,
      "step": 33070
    },
    {
      "epoch": 1.340329409857985,
      "grad_norm": 0.6374233365058899,
      "learning_rate": 0.0001167558547794242,
      "loss": 1.094,
      "step": 33080
    },
    {
      "epoch": 1.3407345880350885,
      "grad_norm": 0.6579474806785583,
      "learning_rate": 0.000116713993867463,
      "loss": 1.0547,
      "step": 33090
    },
    {
      "epoch": 1.3411397662121918,
      "grad_norm": 0.6093845367431641,
      "learning_rate": 0.00011667212994226044,
      "loss": 1.0877,
      "step": 33100
    },
    {
      "epoch": 1.341544944389295,
      "grad_norm": 0.6339181661605835,
      "learning_rate": 0.00011663026301136384,
      "loss": 1.0949,
      "step": 33110
    },
    {
      "epoch": 1.3419501225663986,
      "grad_norm": 0.5468798875808716,
      "learning_rate": 0.00011658839308232106,
      "loss": 1.0867,
      "step": 33120
    },
    {
      "epoch": 1.342355300743502,
      "grad_norm": 0.609060525894165,
      "learning_rate": 0.00011654652016268053,
      "loss": 1.0994,
      "step": 33130
    },
    {
      "epoch": 1.3427604789206053,
      "grad_norm": 0.5889800786972046,
      "learning_rate": 0.00011650464425999122,
      "loss": 1.1025,
      "step": 33140
    },
    {
      "epoch": 1.3431656570977086,
      "grad_norm": 0.5994822978973389,
      "learning_rate": 0.00011646276538180258,
      "loss": 1.0906,
      "step": 33150
    },
    {
      "epoch": 1.3435708352748121,
      "grad_norm": 0.6405680179595947,
      "learning_rate": 0.00011642088353566469,
      "loss": 1.095,
      "step": 33160
    },
    {
      "epoch": 1.3439760134519154,
      "grad_norm": 0.6586445569992065,
      "learning_rate": 0.00011637899872912812,
      "loss": 1.0981,
      "step": 33170
    },
    {
      "epoch": 1.344381191629019,
      "grad_norm": 0.6227049231529236,
      "learning_rate": 0.00011633711096974391,
      "loss": 1.1045,
      "step": 33180
    },
    {
      "epoch": 1.3447863698061222,
      "grad_norm": 0.5784358382225037,
      "learning_rate": 0.00011629522026506374,
      "loss": 1.0798,
      "step": 33190
    },
    {
      "epoch": 1.3451915479832257,
      "grad_norm": 0.6213807463645935,
      "learning_rate": 0.00011625332662263977,
      "loss": 1.0674,
      "step": 33200
    },
    {
      "epoch": 1.345596726160329,
      "grad_norm": 0.6590396761894226,
      "learning_rate": 0.00011621143005002469,
      "loss": 1.0985,
      "step": 33210
    },
    {
      "epoch": 1.3460019043374323,
      "grad_norm": 0.6481648683547974,
      "learning_rate": 0.00011616953055477171,
      "loss": 1.0928,
      "step": 33220
    },
    {
      "epoch": 1.3464070825145358,
      "grad_norm": 0.6376748085021973,
      "learning_rate": 0.00011612762814443459,
      "loss": 1.0896,
      "step": 33230
    },
    {
      "epoch": 1.3468122606916393,
      "grad_norm": 0.8491140604019165,
      "learning_rate": 0.00011608572282656757,
      "loss": 1.1154,
      "step": 33240
    },
    {
      "epoch": 1.3472174388687426,
      "grad_norm": 0.6276373267173767,
      "learning_rate": 0.00011604381460872551,
      "loss": 1.0702,
      "step": 33250
    },
    {
      "epoch": 1.3476226170458458,
      "grad_norm": 0.5831151008605957,
      "learning_rate": 0.00011600190349846367,
      "loss": 1.1055,
      "step": 33260
    },
    {
      "epoch": 1.3480277952229494,
      "grad_norm": 0.6645970940589905,
      "learning_rate": 0.00011595998950333793,
      "loss": 1.0848,
      "step": 33270
    },
    {
      "epoch": 1.3484329734000526,
      "grad_norm": 0.5124262571334839,
      "learning_rate": 0.00011591807263090464,
      "loss": 1.1269,
      "step": 33280
    },
    {
      "epoch": 1.348838151577156,
      "grad_norm": 0.5940738916397095,
      "learning_rate": 0.00011587615288872067,
      "loss": 1.1113,
      "step": 33290
    },
    {
      "epoch": 1.3492433297542594,
      "grad_norm": 0.64183509349823,
      "learning_rate": 0.00011583423028434344,
      "loss": 1.0767,
      "step": 33300
    },
    {
      "epoch": 1.349648507931363,
      "grad_norm": 0.5902032852172852,
      "learning_rate": 0.00011579230482533084,
      "loss": 1.0719,
      "step": 33310
    },
    {
      "epoch": 1.3500536861084662,
      "grad_norm": 0.6327356100082397,
      "learning_rate": 0.0001157503765192413,
      "loss": 1.1255,
      "step": 33320
    },
    {
      "epoch": 1.3504588642855695,
      "grad_norm": 0.6248056292533875,
      "learning_rate": 0.00011570844537363376,
      "loss": 1.1015,
      "step": 33330
    },
    {
      "epoch": 1.350864042462673,
      "grad_norm": 0.7117113471031189,
      "learning_rate": 0.0001156665113960677,
      "loss": 1.1052,
      "step": 33340
    },
    {
      "epoch": 1.3512692206397763,
      "grad_norm": 0.650829553604126,
      "learning_rate": 0.00011562457459410307,
      "loss": 1.1057,
      "step": 33350
    },
    {
      "epoch": 1.3516743988168798,
      "grad_norm": 0.6297826170921326,
      "learning_rate": 0.00011558263497530035,
      "loss": 1.1186,
      "step": 33360
    },
    {
      "epoch": 1.352079576993983,
      "grad_norm": 0.6252959966659546,
      "learning_rate": 0.00011554069254722051,
      "loss": 1.0772,
      "step": 33370
    },
    {
      "epoch": 1.3524847551710866,
      "grad_norm": 0.5776532888412476,
      "learning_rate": 0.00011549874731742502,
      "loss": 1.1155,
      "step": 33380
    },
    {
      "epoch": 1.3528899333481899,
      "grad_norm": 0.5789064168930054,
      "learning_rate": 0.0001154567992934759,
      "loss": 1.0918,
      "step": 33390
    },
    {
      "epoch": 1.3532951115252931,
      "grad_norm": 0.6423953771591187,
      "learning_rate": 0.00011541484848293566,
      "loss": 1.064,
      "step": 33400
    },
    {
      "epoch": 1.3537002897023966,
      "grad_norm": 0.5084463953971863,
      "learning_rate": 0.00011537289489336727,
      "loss": 1.0443,
      "step": 33410
    },
    {
      "epoch": 1.3541054678795001,
      "grad_norm": 0.588420569896698,
      "learning_rate": 0.00011533093853233426,
      "loss": 1.1366,
      "step": 33420
    },
    {
      "epoch": 1.3545106460566034,
      "grad_norm": 0.6043353080749512,
      "learning_rate": 0.00011528897940740056,
      "loss": 1.1099,
      "step": 33430
    },
    {
      "epoch": 1.3549158242337067,
      "grad_norm": 0.5571659207344055,
      "learning_rate": 0.00011524701752613074,
      "loss": 1.1164,
      "step": 33440
    },
    {
      "epoch": 1.3553210024108102,
      "grad_norm": 0.6523189544677734,
      "learning_rate": 0.0001152050528960898,
      "loss": 1.1361,
      "step": 33450
    },
    {
      "epoch": 1.3557261805879135,
      "grad_norm": 0.5536956787109375,
      "learning_rate": 0.00011516308552484316,
      "loss": 1.0906,
      "step": 33460
    },
    {
      "epoch": 1.3561313587650168,
      "grad_norm": 0.6081388592720032,
      "learning_rate": 0.00011512111541995687,
      "loss": 1.0985,
      "step": 33470
    },
    {
      "epoch": 1.3565365369421203,
      "grad_norm": 0.5815418362617493,
      "learning_rate": 0.00011507914258899739,
      "loss": 1.0866,
      "step": 33480
    },
    {
      "epoch": 1.3569417151192238,
      "grad_norm": 0.5959539413452148,
      "learning_rate": 0.00011503716703953166,
      "loss": 1.0603,
      "step": 33490
    },
    {
      "epoch": 1.357346893296327,
      "grad_norm": 0.545052170753479,
      "learning_rate": 0.00011499518877912718,
      "loss": 1.0971,
      "step": 33500
    },
    {
      "epoch": 1.3577520714734304,
      "grad_norm": 0.5912349820137024,
      "learning_rate": 0.00011495320781535186,
      "loss": 1.0928,
      "step": 33510
    },
    {
      "epoch": 1.3581572496505339,
      "grad_norm": 0.5830090641975403,
      "learning_rate": 0.00011491122415577418,
      "loss": 1.0978,
      "step": 33520
    },
    {
      "epoch": 1.3585624278276371,
      "grad_norm": 0.6447737216949463,
      "learning_rate": 0.00011486923780796303,
      "loss": 1.1112,
      "step": 33530
    },
    {
      "epoch": 1.3589676060047406,
      "grad_norm": 0.6198268532752991,
      "learning_rate": 0.0001148272487794878,
      "loss": 1.0971,
      "step": 33540
    },
    {
      "epoch": 1.359372784181844,
      "grad_norm": 0.6065701842308044,
      "learning_rate": 0.00011478525707791843,
      "loss": 1.1097,
      "step": 33550
    },
    {
      "epoch": 1.3597779623589474,
      "grad_norm": 0.5649124383926392,
      "learning_rate": 0.00011474326271082524,
      "loss": 1.115,
      "step": 33560
    },
    {
      "epoch": 1.3601831405360507,
      "grad_norm": 0.5999446511268616,
      "learning_rate": 0.00011470126568577911,
      "loss": 1.0696,
      "step": 33570
    },
    {
      "epoch": 1.360588318713154,
      "grad_norm": 0.6025034785270691,
      "learning_rate": 0.00011465926601035137,
      "loss": 1.0497,
      "step": 33580
    },
    {
      "epoch": 1.3609934968902575,
      "grad_norm": 0.5645689368247986,
      "learning_rate": 0.0001146172636921138,
      "loss": 1.1106,
      "step": 33590
    },
    {
      "epoch": 1.361398675067361,
      "grad_norm": 0.574652910232544,
      "learning_rate": 0.00011457525873863873,
      "loss": 1.087,
      "step": 33600
    },
    {
      "epoch": 1.3618038532444643,
      "grad_norm": 0.6573282480239868,
      "learning_rate": 0.00011453325115749889,
      "loss": 1.1071,
      "step": 33610
    },
    {
      "epoch": 1.3622090314215676,
      "grad_norm": 0.6010294556617737,
      "learning_rate": 0.00011449124095626754,
      "loss": 1.1043,
      "step": 33620
    },
    {
      "epoch": 1.362614209598671,
      "grad_norm": 0.6514992117881775,
      "learning_rate": 0.0001144492281425183,
      "loss": 1.1024,
      "step": 33630
    },
    {
      "epoch": 1.3630193877757744,
      "grad_norm": 0.6338147521018982,
      "learning_rate": 0.00011440721272382548,
      "loss": 1.0977,
      "step": 33640
    },
    {
      "epoch": 1.3634245659528776,
      "grad_norm": 0.6433400511741638,
      "learning_rate": 0.00011436519470776362,
      "loss": 1.071,
      "step": 33650
    },
    {
      "epoch": 1.3638297441299811,
      "grad_norm": 0.6253982782363892,
      "learning_rate": 0.00011432317410190791,
      "loss": 1.1214,
      "step": 33660
    },
    {
      "epoch": 1.3642349223070847,
      "grad_norm": 0.6144083738327026,
      "learning_rate": 0.00011428115091383389,
      "loss": 1.1356,
      "step": 33670
    },
    {
      "epoch": 1.364640100484188,
      "grad_norm": 0.5572004914283752,
      "learning_rate": 0.0001142391251511176,
      "loss": 1.1095,
      "step": 33680
    },
    {
      "epoch": 1.3650452786612912,
      "grad_norm": 0.6156103014945984,
      "learning_rate": 0.00011419709682133555,
      "loss": 1.0777,
      "step": 33690
    },
    {
      "epoch": 1.3654504568383947,
      "grad_norm": 0.6177895665168762,
      "learning_rate": 0.00011415506593206477,
      "loss": 1.0981,
      "step": 33700
    },
    {
      "epoch": 1.365855635015498,
      "grad_norm": 0.6344954967498779,
      "learning_rate": 0.00011411303249088261,
      "loss": 1.0771,
      "step": 33710
    },
    {
      "epoch": 1.3662608131926015,
      "grad_norm": 0.587524950504303,
      "learning_rate": 0.00011407099650536706,
      "loss": 1.0916,
      "step": 33720
    },
    {
      "epoch": 1.3666659913697048,
      "grad_norm": 0.564702570438385,
      "learning_rate": 0.00011402895798309639,
      "loss": 1.0951,
      "step": 33730
    },
    {
      "epoch": 1.3670711695468083,
      "grad_norm": 0.6318004131317139,
      "learning_rate": 0.00011398691693164947,
      "loss": 1.077,
      "step": 33740
    },
    {
      "epoch": 1.3674763477239116,
      "grad_norm": 0.6013349890708923,
      "learning_rate": 0.0001139448733586055,
      "loss": 1.1008,
      "step": 33750
    },
    {
      "epoch": 1.3678815259010149,
      "grad_norm": 0.5918259620666504,
      "learning_rate": 0.00011390282727154431,
      "loss": 1.0681,
      "step": 33760
    },
    {
      "epoch": 1.3682867040781184,
      "grad_norm": 0.5643131732940674,
      "learning_rate": 0.00011386077867804599,
      "loss": 1.1105,
      "step": 33770
    },
    {
      "epoch": 1.3686918822552219,
      "grad_norm": 0.6691088676452637,
      "learning_rate": 0.00011381872758569119,
      "loss": 1.0748,
      "step": 33780
    },
    {
      "epoch": 1.3690970604323252,
      "grad_norm": 0.5862001776695251,
      "learning_rate": 0.00011377667400206101,
      "loss": 1.1185,
      "step": 33790
    },
    {
      "epoch": 1.3695022386094284,
      "grad_norm": 0.5818652510643005,
      "learning_rate": 0.0001137346179347369,
      "loss": 1.1183,
      "step": 33800
    },
    {
      "epoch": 1.369907416786532,
      "grad_norm": 0.6411664485931396,
      "learning_rate": 0.00011369255939130095,
      "loss": 1.122,
      "step": 33810
    },
    {
      "epoch": 1.3703125949636352,
      "grad_norm": 0.5764086842536926,
      "learning_rate": 0.00011365049837933548,
      "loss": 1.1181,
      "step": 33820
    },
    {
      "epoch": 1.3707177731407385,
      "grad_norm": 0.5892274379730225,
      "learning_rate": 0.00011360843490642343,
      "loss": 1.079,
      "step": 33830
    },
    {
      "epoch": 1.371122951317842,
      "grad_norm": 0.652621865272522,
      "learning_rate": 0.00011356636898014807,
      "loss": 1.1052,
      "step": 33840
    },
    {
      "epoch": 1.3715281294949455,
      "grad_norm": 0.5842803120613098,
      "learning_rate": 0.00011352430060809317,
      "loss": 1.0865,
      "step": 33850
    },
    {
      "epoch": 1.3719333076720488,
      "grad_norm": 0.5919479131698608,
      "learning_rate": 0.00011348222979784289,
      "loss": 1.0832,
      "step": 33860
    },
    {
      "epoch": 1.372338485849152,
      "grad_norm": 0.5847943425178528,
      "learning_rate": 0.00011344015655698192,
      "loss": 1.0949,
      "step": 33870
    },
    {
      "epoch": 1.3727436640262556,
      "grad_norm": 0.5319411754608154,
      "learning_rate": 0.00011339808089309527,
      "loss": 1.083,
      "step": 33880
    },
    {
      "epoch": 1.3731488422033589,
      "grad_norm": 0.5951094031333923,
      "learning_rate": 0.00011335600281376851,
      "loss": 1.1087,
      "step": 33890
    },
    {
      "epoch": 1.3735540203804624,
      "grad_norm": 0.6213427186012268,
      "learning_rate": 0.00011331392232658753,
      "loss": 1.0761,
      "step": 33900
    },
    {
      "epoch": 1.3739591985575657,
      "grad_norm": 0.6758957505226135,
      "learning_rate": 0.0001132718394391387,
      "loss": 1.0883,
      "step": 33910
    },
    {
      "epoch": 1.3743643767346692,
      "grad_norm": 0.5517923831939697,
      "learning_rate": 0.00011322975415900889,
      "loss": 1.0775,
      "step": 33920
    },
    {
      "epoch": 1.3747695549117724,
      "grad_norm": 0.5795263648033142,
      "learning_rate": 0.00011318766649378532,
      "loss": 1.1091,
      "step": 33930
    },
    {
      "epoch": 1.3751747330888757,
      "grad_norm": 0.5972916483879089,
      "learning_rate": 0.0001131455764510556,
      "loss": 1.1163,
      "step": 33940
    },
    {
      "epoch": 1.3755799112659792,
      "grad_norm": 0.6252707242965698,
      "learning_rate": 0.00011310348403840794,
      "loss": 1.1278,
      "step": 33950
    },
    {
      "epoch": 1.3759850894430827,
      "grad_norm": 0.5667466521263123,
      "learning_rate": 0.00011306138926343075,
      "loss": 1.0657,
      "step": 33960
    },
    {
      "epoch": 1.376390267620186,
      "grad_norm": 0.601696252822876,
      "learning_rate": 0.00011301929213371309,
      "loss": 1.111,
      "step": 33970
    },
    {
      "epoch": 1.3767954457972893,
      "grad_norm": 0.6093372106552124,
      "learning_rate": 0.00011297719265684428,
      "loss": 1.1131,
      "step": 33980
    },
    {
      "epoch": 1.3772006239743928,
      "grad_norm": 0.626377522945404,
      "learning_rate": 0.00011293509084041409,
      "loss": 1.1135,
      "step": 33990
    },
    {
      "epoch": 1.377605802151496,
      "grad_norm": 0.6509422063827515,
      "learning_rate": 0.00011289298669201282,
      "loss": 1.0999,
      "step": 34000
    },
    {
      "epoch": 1.3780109803285994,
      "grad_norm": 0.6934104561805725,
      "learning_rate": 0.00011285088021923109,
      "loss": 1.0961,
      "step": 34010
    },
    {
      "epoch": 1.3784161585057029,
      "grad_norm": 0.6033422350883484,
      "learning_rate": 0.0001128087714296599,
      "loss": 1.0961,
      "step": 34020
    },
    {
      "epoch": 1.3788213366828064,
      "grad_norm": 0.7178053855895996,
      "learning_rate": 0.00011276666033089081,
      "loss": 1.1309,
      "step": 34030
    },
    {
      "epoch": 1.3792265148599097,
      "grad_norm": 0.595296323299408,
      "learning_rate": 0.00011272454693051566,
      "loss": 1.0868,
      "step": 34040
    },
    {
      "epoch": 1.379631693037013,
      "grad_norm": 0.6362353563308716,
      "learning_rate": 0.00011268243123612681,
      "loss": 1.0615,
      "step": 34050
    },
    {
      "epoch": 1.3800368712141164,
      "grad_norm": 0.6701503396034241,
      "learning_rate": 0.00011264031325531694,
      "loss": 1.0977,
      "step": 34060
    },
    {
      "epoch": 1.3804420493912197,
      "grad_norm": 0.7223421931266785,
      "learning_rate": 0.00011259819299567922,
      "loss": 1.1218,
      "step": 34070
    },
    {
      "epoch": 1.3808472275683232,
      "grad_norm": 0.698536217212677,
      "learning_rate": 0.00011255607046480716,
      "loss": 1.1036,
      "step": 34080
    },
    {
      "epoch": 1.3812524057454265,
      "grad_norm": 0.609889566898346,
      "learning_rate": 0.00011251394567029476,
      "loss": 1.0853,
      "step": 34090
    },
    {
      "epoch": 1.38165758392253,
      "grad_norm": 0.5927088856697083,
      "learning_rate": 0.00011247181861973638,
      "loss": 1.1375,
      "step": 34100
    },
    {
      "epoch": 1.3820627620996333,
      "grad_norm": 0.7491838932037354,
      "learning_rate": 0.00011242968932072675,
      "loss": 1.0628,
      "step": 34110
    },
    {
      "epoch": 1.3824679402767366,
      "grad_norm": 0.5972330570220947,
      "learning_rate": 0.0001123875577808611,
      "loss": 1.1358,
      "step": 34120
    },
    {
      "epoch": 1.38287311845384,
      "grad_norm": 0.6070208549499512,
      "learning_rate": 0.00011234542400773495,
      "loss": 1.0509,
      "step": 34130
    },
    {
      "epoch": 1.3832782966309436,
      "grad_norm": 0.5872363448143005,
      "learning_rate": 0.00011230328800894437,
      "loss": 1.1057,
      "step": 34140
    },
    {
      "epoch": 1.3836834748080469,
      "grad_norm": 0.5086923241615295,
      "learning_rate": 0.00011226114979208566,
      "loss": 1.0535,
      "step": 34150
    },
    {
      "epoch": 1.3840886529851502,
      "grad_norm": 0.6294886469841003,
      "learning_rate": 0.00011221900936475564,
      "loss": 1.121,
      "step": 34160
    },
    {
      "epoch": 1.3844938311622537,
      "grad_norm": 0.604988694190979,
      "learning_rate": 0.00011217686673455153,
      "loss": 1.1182,
      "step": 34170
    },
    {
      "epoch": 1.384899009339357,
      "grad_norm": 0.5684612989425659,
      "learning_rate": 0.00011213472190907083,
      "loss": 1.1058,
      "step": 34180
    },
    {
      "epoch": 1.3853041875164602,
      "grad_norm": 0.6145623326301575,
      "learning_rate": 0.00011209257489591158,
      "loss": 1.1113,
      "step": 34190
    },
    {
      "epoch": 1.3857093656935637,
      "grad_norm": 0.6430658102035522,
      "learning_rate": 0.00011205042570267214,
      "loss": 1.1321,
      "step": 34200
    },
    {
      "epoch": 1.3861145438706672,
      "grad_norm": 0.6531445980072021,
      "learning_rate": 0.00011200827433695127,
      "loss": 1.0586,
      "step": 34210
    },
    {
      "epoch": 1.3865197220477705,
      "grad_norm": 0.5561890602111816,
      "learning_rate": 0.0001119661208063481,
      "loss": 1.1075,
      "step": 34220
    },
    {
      "epoch": 1.3869249002248738,
      "grad_norm": 0.651442289352417,
      "learning_rate": 0.00011192396511846222,
      "loss": 1.1155,
      "step": 34230
    },
    {
      "epoch": 1.3873300784019773,
      "grad_norm": 0.5534287691116333,
      "learning_rate": 0.00011188180728089355,
      "loss": 1.0779,
      "step": 34240
    },
    {
      "epoch": 1.3877352565790806,
      "grad_norm": 0.6057088971138,
      "learning_rate": 0.00011183964730124236,
      "loss": 1.1126,
      "step": 34250
    },
    {
      "epoch": 1.388140434756184,
      "grad_norm": 0.6098588109016418,
      "learning_rate": 0.00011179748518710944,
      "loss": 1.1321,
      "step": 34260
    },
    {
      "epoch": 1.3885456129332874,
      "grad_norm": 0.6665880084037781,
      "learning_rate": 0.00011175532094609582,
      "loss": 1.0907,
      "step": 34270
    },
    {
      "epoch": 1.3889507911103909,
      "grad_norm": 0.5857251882553101,
      "learning_rate": 0.00011171315458580303,
      "loss": 1.0833,
      "step": 34280
    },
    {
      "epoch": 1.3893559692874942,
      "grad_norm": 0.609535813331604,
      "learning_rate": 0.0001116709861138329,
      "loss": 1.0928,
      "step": 34290
    },
    {
      "epoch": 1.3897611474645974,
      "grad_norm": 0.6104686856269836,
      "learning_rate": 0.00011162881553778766,
      "loss": 1.1012,
      "step": 34300
    },
    {
      "epoch": 1.390166325641701,
      "grad_norm": 0.7035071849822998,
      "learning_rate": 0.00011158664286526994,
      "loss": 1.1008,
      "step": 34310
    },
    {
      "epoch": 1.3905715038188045,
      "grad_norm": 0.6324309706687927,
      "learning_rate": 0.00011154446810388275,
      "loss": 1.066,
      "step": 34320
    },
    {
      "epoch": 1.3909766819959077,
      "grad_norm": 0.5786617994308472,
      "learning_rate": 0.00011150229126122941,
      "loss": 1.0624,
      "step": 34330
    },
    {
      "epoch": 1.391381860173011,
      "grad_norm": 0.5530617833137512,
      "learning_rate": 0.00011146011234491376,
      "loss": 1.1083,
      "step": 34340
    },
    {
      "epoch": 1.3917870383501145,
      "grad_norm": 0.6571694016456604,
      "learning_rate": 0.00011141793136253986,
      "loss": 1.1186,
      "step": 34350
    },
    {
      "epoch": 1.3921922165272178,
      "grad_norm": 0.5784533619880676,
      "learning_rate": 0.00011137574832171221,
      "loss": 1.1273,
      "step": 34360
    },
    {
      "epoch": 1.392597394704321,
      "grad_norm": 0.576474130153656,
      "learning_rate": 0.00011133356323003568,
      "loss": 1.0751,
      "step": 34370
    },
    {
      "epoch": 1.3930025728814246,
      "grad_norm": 0.5723327994346619,
      "learning_rate": 0.0001112913760951155,
      "loss": 1.0642,
      "step": 34380
    },
    {
      "epoch": 1.393407751058528,
      "grad_norm": 0.5872519016265869,
      "learning_rate": 0.0001112491869245573,
      "loss": 1.0895,
      "step": 34390
    },
    {
      "epoch": 1.3938129292356314,
      "grad_norm": 0.6492037773132324,
      "learning_rate": 0.000111206995725967,
      "loss": 1.0974,
      "step": 34400
    },
    {
      "epoch": 1.3942181074127347,
      "grad_norm": 0.614399790763855,
      "learning_rate": 0.00011116480250695101,
      "loss": 1.0831,
      "step": 34410
    },
    {
      "epoch": 1.3946232855898382,
      "grad_norm": 0.6250647902488708,
      "learning_rate": 0.00011112260727511596,
      "loss": 1.0798,
      "step": 34420
    },
    {
      "epoch": 1.3950284637669415,
      "grad_norm": 0.5748769640922546,
      "learning_rate": 0.00011108041003806897,
      "loss": 1.0838,
      "step": 34430
    },
    {
      "epoch": 1.395433641944045,
      "grad_norm": 0.5800958275794983,
      "learning_rate": 0.00011103821080341744,
      "loss": 1.0708,
      "step": 34440
    },
    {
      "epoch": 1.3958388201211482,
      "grad_norm": 0.5615698099136353,
      "learning_rate": 0.00011099600957876914,
      "loss": 1.0577,
      "step": 34450
    },
    {
      "epoch": 1.3962439982982517,
      "grad_norm": 0.5954749584197998,
      "learning_rate": 0.00011095380637173225,
      "loss": 1.1006,
      "step": 34460
    },
    {
      "epoch": 1.396649176475355,
      "grad_norm": 0.6228195428848267,
      "learning_rate": 0.00011091160118991522,
      "loss": 1.1224,
      "step": 34470
    },
    {
      "epoch": 1.3970543546524583,
      "grad_norm": 0.5889665484428406,
      "learning_rate": 0.00011086939404092697,
      "loss": 1.102,
      "step": 34480
    },
    {
      "epoch": 1.3974595328295618,
      "grad_norm": 0.6514177918434143,
      "learning_rate": 0.00011082718493237669,
      "loss": 1.1171,
      "step": 34490
    },
    {
      "epoch": 1.3978647110066653,
      "grad_norm": 0.6469988226890564,
      "learning_rate": 0.00011078497387187389,
      "loss": 1.0972,
      "step": 34500
    },
    {
      "epoch": 1.3982698891837686,
      "grad_norm": 0.6163913607597351,
      "learning_rate": 0.00011074276086702857,
      "loss": 1.057,
      "step": 34510
    },
    {
      "epoch": 1.3986750673608719,
      "grad_norm": 0.6225057244300842,
      "learning_rate": 0.00011070054592545098,
      "loss": 1.0806,
      "step": 34520
    },
    {
      "epoch": 1.3990802455379754,
      "grad_norm": 0.5911839604377747,
      "learning_rate": 0.00011065832905475169,
      "loss": 1.0805,
      "step": 34530
    },
    {
      "epoch": 1.3994854237150787,
      "grad_norm": 0.6328493356704712,
      "learning_rate": 0.0001106161102625417,
      "loss": 1.0968,
      "step": 34540
    },
    {
      "epoch": 1.399890601892182,
      "grad_norm": 0.6108220219612122,
      "learning_rate": 0.00011057388955643233,
      "loss": 1.0505,
      "step": 34550
    },
    {
      "epoch": 1.4002957800692855,
      "grad_norm": 0.5974302887916565,
      "learning_rate": 0.00011053166694403521,
      "loss": 1.1069,
      "step": 34560
    },
    {
      "epoch": 1.400700958246389,
      "grad_norm": 0.6139350533485413,
      "learning_rate": 0.00011048944243296236,
      "loss": 1.0978,
      "step": 34570
    },
    {
      "epoch": 1.4011061364234922,
      "grad_norm": 0.5869536995887756,
      "learning_rate": 0.00011044721603082609,
      "loss": 1.0964,
      "step": 34580
    },
    {
      "epoch": 1.4015113146005955,
      "grad_norm": 0.6408414244651794,
      "learning_rate": 0.00011040498774523912,
      "loss": 1.1095,
      "step": 34590
    },
    {
      "epoch": 1.401916492777699,
      "grad_norm": 0.6449887752532959,
      "learning_rate": 0.00011036275758381449,
      "loss": 1.0521,
      "step": 34600
    },
    {
      "epoch": 1.4023216709548023,
      "grad_norm": 0.5978503823280334,
      "learning_rate": 0.0001103205255541655,
      "loss": 1.0853,
      "step": 34610
    },
    {
      "epoch": 1.4027268491319058,
      "grad_norm": 0.5689315795898438,
      "learning_rate": 0.00011027829166390591,
      "loss": 1.1018,
      "step": 34620
    },
    {
      "epoch": 1.403132027309009,
      "grad_norm": 0.5704008340835571,
      "learning_rate": 0.0001102360559206497,
      "loss": 1.0729,
      "step": 34630
    },
    {
      "epoch": 1.4035372054861126,
      "grad_norm": 0.6645287871360779,
      "learning_rate": 0.00011019381833201125,
      "loss": 1.1041,
      "step": 34640
    },
    {
      "epoch": 1.4039423836632159,
      "grad_norm": 0.5907942652702332,
      "learning_rate": 0.0001101515789056053,
      "loss": 1.0741,
      "step": 34650
    },
    {
      "epoch": 1.4043475618403192,
      "grad_norm": 0.626250684261322,
      "learning_rate": 0.00011010933764904685,
      "loss": 1.0744,
      "step": 34660
    },
    {
      "epoch": 1.4047527400174227,
      "grad_norm": 0.6373708844184875,
      "learning_rate": 0.00011006709456995125,
      "loss": 1.0862,
      "step": 34670
    },
    {
      "epoch": 1.4051579181945262,
      "grad_norm": 0.5807908177375793,
      "learning_rate": 0.0001100248496759342,
      "loss": 1.0956,
      "step": 34680
    },
    {
      "epoch": 1.4055630963716295,
      "grad_norm": 0.6874992251396179,
      "learning_rate": 0.00010998260297461176,
      "loss": 1.1208,
      "step": 34690
    },
    {
      "epoch": 1.4059682745487327,
      "grad_norm": 0.6513946056365967,
      "learning_rate": 0.00010994035447360018,
      "loss": 1.1141,
      "step": 34700
    },
    {
      "epoch": 1.4063734527258362,
      "grad_norm": 0.6299512982368469,
      "learning_rate": 0.00010989810418051621,
      "loss": 1.0721,
      "step": 34710
    },
    {
      "epoch": 1.4067786309029395,
      "grad_norm": 0.6249724626541138,
      "learning_rate": 0.0001098558521029768,
      "loss": 1.0923,
      "step": 34720
    },
    {
      "epoch": 1.4071838090800428,
      "grad_norm": 0.6147335767745972,
      "learning_rate": 0.00010981359824859927,
      "loss": 1.0796,
      "step": 34730
    },
    {
      "epoch": 1.4075889872571463,
      "grad_norm": 0.564756453037262,
      "learning_rate": 0.00010977134262500126,
      "loss": 1.0969,
      "step": 34740
    },
    {
      "epoch": 1.4079941654342498,
      "grad_norm": 0.7307913899421692,
      "learning_rate": 0.00010972908523980072,
      "loss": 1.1132,
      "step": 34750
    },
    {
      "epoch": 1.408399343611353,
      "grad_norm": 0.623412013053894,
      "learning_rate": 0.00010968682610061591,
      "loss": 1.067,
      "step": 34760
    },
    {
      "epoch": 1.4088045217884564,
      "grad_norm": 0.6111620664596558,
      "learning_rate": 0.00010964456521506545,
      "loss": 1.1408,
      "step": 34770
    },
    {
      "epoch": 1.40920969996556,
      "grad_norm": 0.6528574228286743,
      "learning_rate": 0.00010960230259076818,
      "loss": 1.0616,
      "step": 34780
    },
    {
      "epoch": 1.4096148781426632,
      "grad_norm": 0.5995625257492065,
      "learning_rate": 0.00010956003823534336,
      "loss": 1.1002,
      "step": 34790
    },
    {
      "epoch": 1.4100200563197667,
      "grad_norm": 0.563672661781311,
      "learning_rate": 0.00010951777215641051,
      "loss": 1.1025,
      "step": 34800
    },
    {
      "epoch": 1.41042523449687,
      "grad_norm": 0.5445964336395264,
      "learning_rate": 0.00010947550436158942,
      "loss": 1.1106,
      "step": 34810
    },
    {
      "epoch": 1.4108304126739735,
      "grad_norm": 0.5608357191085815,
      "learning_rate": 0.00010943323485850032,
      "loss": 1.0875,
      "step": 34820
    },
    {
      "epoch": 1.4112355908510767,
      "grad_norm": 0.5783724784851074,
      "learning_rate": 0.0001093909636547636,
      "loss": 1.0901,
      "step": 34830
    },
    {
      "epoch": 1.41164076902818,
      "grad_norm": 0.6912353038787842,
      "learning_rate": 0.000109348690758,
      "loss": 1.0869,
      "step": 34840
    },
    {
      "epoch": 1.4120459472052835,
      "grad_norm": 0.6503822803497314,
      "learning_rate": 0.00010930641617583068,
      "loss": 1.1027,
      "step": 34850
    },
    {
      "epoch": 1.412451125382387,
      "grad_norm": 0.5995206236839294,
      "learning_rate": 0.00010926413991587694,
      "loss": 1.1353,
      "step": 34860
    },
    {
      "epoch": 1.4128563035594903,
      "grad_norm": 0.632533848285675,
      "learning_rate": 0.00010922186198576043,
      "loss": 1.0747,
      "step": 34870
    },
    {
      "epoch": 1.4132614817365936,
      "grad_norm": 0.5474535822868347,
      "learning_rate": 0.00010917958239310318,
      "loss": 1.0789,
      "step": 34880
    },
    {
      "epoch": 1.413666659913697,
      "grad_norm": 0.646422266960144,
      "learning_rate": 0.00010913730114552743,
      "loss": 1.0918,
      "step": 34890
    },
    {
      "epoch": 1.4140718380908004,
      "grad_norm": 0.624173104763031,
      "learning_rate": 0.00010909501825065576,
      "loss": 1.1055,
      "step": 34900
    },
    {
      "epoch": 1.4144770162679037,
      "grad_norm": 0.5669375658035278,
      "learning_rate": 0.00010905273371611105,
      "loss": 1.0866,
      "step": 34910
    },
    {
      "epoch": 1.4148821944450072,
      "grad_norm": 0.6007163524627686,
      "learning_rate": 0.00010901044754951644,
      "loss": 1.0872,
      "step": 34920
    },
    {
      "epoch": 1.4152873726221107,
      "grad_norm": 0.6030934453010559,
      "learning_rate": 0.00010896815975849542,
      "loss": 1.0974,
      "step": 34930
    },
    {
      "epoch": 1.415692550799214,
      "grad_norm": 0.627704918384552,
      "learning_rate": 0.0001089258703506717,
      "loss": 1.0854,
      "step": 34940
    },
    {
      "epoch": 1.4160977289763172,
      "grad_norm": 0.6791233420372009,
      "learning_rate": 0.00010888357933366933,
      "loss": 1.093,
      "step": 34950
    },
    {
      "epoch": 1.4165029071534208,
      "grad_norm": 0.6611270308494568,
      "learning_rate": 0.0001088412867151127,
      "loss": 1.0878,
      "step": 34960
    },
    {
      "epoch": 1.416908085330524,
      "grad_norm": 0.6001415848731995,
      "learning_rate": 0.00010879899250262635,
      "loss": 1.1268,
      "step": 34970
    },
    {
      "epoch": 1.4173132635076275,
      "grad_norm": 0.6399940848350525,
      "learning_rate": 0.00010875669670383521,
      "loss": 1.1165,
      "step": 34980
    },
    {
      "epoch": 1.4177184416847308,
      "grad_norm": 0.5713139772415161,
      "learning_rate": 0.00010871439932636451,
      "loss": 1.1129,
      "step": 34990
    },
    {
      "epoch": 1.4181236198618343,
      "grad_norm": 0.6013307571411133,
      "learning_rate": 0.00010867210037783973,
      "loss": 1.0841,
      "step": 35000
    },
    {
      "epoch": 1.4185287980389376,
      "grad_norm": 0.5728122591972351,
      "learning_rate": 0.00010862979986588654,
      "loss": 1.1132,
      "step": 35010
    },
    {
      "epoch": 1.418933976216041,
      "grad_norm": 0.5671045780181885,
      "learning_rate": 0.00010858749779813114,
      "loss": 1.1282,
      "step": 35020
    },
    {
      "epoch": 1.4193391543931444,
      "grad_norm": 0.5227568745613098,
      "learning_rate": 0.00010854519418219971,
      "loss": 1.0478,
      "step": 35030
    },
    {
      "epoch": 1.419744332570248,
      "grad_norm": 0.5959317684173584,
      "learning_rate": 0.00010850288902571893,
      "loss": 1.0687,
      "step": 35040
    },
    {
      "epoch": 1.4201495107473512,
      "grad_norm": 0.6348966956138611,
      "learning_rate": 0.00010846058233631565,
      "loss": 1.0928,
      "step": 35050
    },
    {
      "epoch": 1.4205546889244545,
      "grad_norm": 0.6467334032058716,
      "learning_rate": 0.00010841827412161705,
      "loss": 1.1023,
      "step": 35060
    },
    {
      "epoch": 1.420959867101558,
      "grad_norm": 0.6253318190574646,
      "learning_rate": 0.00010837596438925052,
      "loss": 1.1293,
      "step": 35070
    },
    {
      "epoch": 1.4213650452786613,
      "grad_norm": 0.5922788381576538,
      "learning_rate": 0.00010833365314684386,
      "loss": 1.0534,
      "step": 35080
    },
    {
      "epoch": 1.4217702234557645,
      "grad_norm": 0.6065093278884888,
      "learning_rate": 0.00010829134040202494,
      "loss": 1.124,
      "step": 35090
    },
    {
      "epoch": 1.422175401632868,
      "grad_norm": 0.5988242030143738,
      "learning_rate": 0.00010824902616242207,
      "loss": 1.1308,
      "step": 35100
    },
    {
      "epoch": 1.4225805798099715,
      "grad_norm": 0.606312096118927,
      "learning_rate": 0.00010820671043566378,
      "loss": 1.0533,
      "step": 35110
    },
    {
      "epoch": 1.4229857579870748,
      "grad_norm": 0.5791960954666138,
      "learning_rate": 0.00010816439322937879,
      "loss": 1.1135,
      "step": 35120
    },
    {
      "epoch": 1.4233909361641781,
      "grad_norm": 0.6141365170478821,
      "learning_rate": 0.00010812207455119625,
      "loss": 1.1045,
      "step": 35130
    },
    {
      "epoch": 1.4237961143412816,
      "grad_norm": 0.5796669125556946,
      "learning_rate": 0.0001080797544087454,
      "loss": 1.0947,
      "step": 35140
    },
    {
      "epoch": 1.424201292518385,
      "grad_norm": 0.5605058670043945,
      "learning_rate": 0.00010803743280965584,
      "loss": 1.1053,
      "step": 35150
    },
    {
      "epoch": 1.4246064706954884,
      "grad_norm": 0.574664831161499,
      "learning_rate": 0.00010799510976155743,
      "loss": 1.0791,
      "step": 35160
    },
    {
      "epoch": 1.4250116488725917,
      "grad_norm": 0.6042913794517517,
      "learning_rate": 0.00010795278527208028,
      "loss": 1.12,
      "step": 35170
    },
    {
      "epoch": 1.4254168270496952,
      "grad_norm": 0.5997887253761292,
      "learning_rate": 0.00010791045934885473,
      "loss": 1.1289,
      "step": 35180
    },
    {
      "epoch": 1.4258220052267985,
      "grad_norm": 0.5986152291297913,
      "learning_rate": 0.00010786813199951145,
      "loss": 1.0835,
      "step": 35190
    },
    {
      "epoch": 1.4262271834039018,
      "grad_norm": 0.597609281539917,
      "learning_rate": 0.00010782580323168126,
      "loss": 1.1073,
      "step": 35200
    },
    {
      "epoch": 1.4266323615810053,
      "grad_norm": 0.6875548958778381,
      "learning_rate": 0.00010778347305299537,
      "loss": 1.1024,
      "step": 35210
    },
    {
      "epoch": 1.4270375397581088,
      "grad_norm": 0.6283231973648071,
      "learning_rate": 0.00010774114147108512,
      "loss": 1.0991,
      "step": 35220
    },
    {
      "epoch": 1.427442717935212,
      "grad_norm": 0.623218834400177,
      "learning_rate": 0.00010769880849358215,
      "loss": 1.1413,
      "step": 35230
    },
    {
      "epoch": 1.4278478961123153,
      "grad_norm": 0.5575385093688965,
      "learning_rate": 0.00010765647412811841,
      "loss": 1.1033,
      "step": 35240
    },
    {
      "epoch": 1.4282530742894188,
      "grad_norm": 0.6549649834632874,
      "learning_rate": 0.00010761413838232601,
      "loss": 1.0673,
      "step": 35250
    },
    {
      "epoch": 1.4286582524665221,
      "grad_norm": 0.6167589426040649,
      "learning_rate": 0.00010757180126383735,
      "loss": 1.0816,
      "step": 35260
    },
    {
      "epoch": 1.4290634306436254,
      "grad_norm": 0.549662709236145,
      "learning_rate": 0.00010752946278028509,
      "loss": 1.1027,
      "step": 35270
    },
    {
      "epoch": 1.429468608820729,
      "grad_norm": 0.5736987590789795,
      "learning_rate": 0.0001074871229393021,
      "loss": 1.0997,
      "step": 35280
    },
    {
      "epoch": 1.4298737869978324,
      "grad_norm": 0.6662293076515198,
      "learning_rate": 0.00010744478174852148,
      "loss": 1.1299,
      "step": 35290
    },
    {
      "epoch": 1.4302789651749357,
      "grad_norm": 0.5724717378616333,
      "learning_rate": 0.0001074024392155767,
      "loss": 1.1279,
      "step": 35300
    },
    {
      "epoch": 1.430684143352039,
      "grad_norm": 0.6482726335525513,
      "learning_rate": 0.0001073600953481013,
      "loss": 1.1,
      "step": 35310
    },
    {
      "epoch": 1.4310893215291425,
      "grad_norm": 0.6774756908416748,
      "learning_rate": 0.00010731775015372916,
      "loss": 1.1159,
      "step": 35320
    },
    {
      "epoch": 1.4314944997062458,
      "grad_norm": 0.6305453181266785,
      "learning_rate": 0.0001072754036400944,
      "loss": 1.1266,
      "step": 35330
    },
    {
      "epoch": 1.4318996778833493,
      "grad_norm": 0.6353712677955627,
      "learning_rate": 0.00010723305581483133,
      "loss": 1.0791,
      "step": 35340
    },
    {
      "epoch": 1.4323048560604525,
      "grad_norm": 0.6062259078025818,
      "learning_rate": 0.00010719070668557456,
      "loss": 1.1544,
      "step": 35350
    },
    {
      "epoch": 1.432710034237556,
      "grad_norm": 0.5690946578979492,
      "learning_rate": 0.00010714835625995887,
      "loss": 1.108,
      "step": 35360
    },
    {
      "epoch": 1.4331152124146593,
      "grad_norm": 0.7693724632263184,
      "learning_rate": 0.00010710600454561928,
      "loss": 1.0973,
      "step": 35370
    },
    {
      "epoch": 1.4335203905917626,
      "grad_norm": 0.5999078154563904,
      "learning_rate": 0.00010706365155019113,
      "loss": 1.0724,
      "step": 35380
    },
    {
      "epoch": 1.4339255687688661,
      "grad_norm": 0.6761453151702881,
      "learning_rate": 0.00010702129728130987,
      "loss": 1.0918,
      "step": 35390
    },
    {
      "epoch": 1.4343307469459696,
      "grad_norm": 0.7078096270561218,
      "learning_rate": 0.00010697894174661127,
      "loss": 1.1088,
      "step": 35400
    },
    {
      "epoch": 1.434735925123073,
      "grad_norm": 0.6111046075820923,
      "learning_rate": 0.00010693658495373124,
      "loss": 1.1118,
      "step": 35410
    },
    {
      "epoch": 1.4351411033001762,
      "grad_norm": 0.6181390881538391,
      "learning_rate": 0.00010689422691030607,
      "loss": 1.1184,
      "step": 35420
    },
    {
      "epoch": 1.4355462814772797,
      "grad_norm": 0.6105769872665405,
      "learning_rate": 0.00010685186762397205,
      "loss": 1.1158,
      "step": 35430
    },
    {
      "epoch": 1.435951459654383,
      "grad_norm": 0.6124796867370605,
      "learning_rate": 0.00010680950710236594,
      "loss": 1.0973,
      "step": 35440
    },
    {
      "epoch": 1.4363566378314863,
      "grad_norm": 0.6446694135665894,
      "learning_rate": 0.00010676714535312456,
      "loss": 1.0814,
      "step": 35450
    },
    {
      "epoch": 1.4367618160085898,
      "grad_norm": 0.5901733636856079,
      "learning_rate": 0.00010672478238388492,
      "loss": 1.0896,
      "step": 35460
    },
    {
      "epoch": 1.4371669941856933,
      "grad_norm": 0.6569098234176636,
      "learning_rate": 0.00010668241820228444,
      "loss": 1.073,
      "step": 35470
    },
    {
      "epoch": 1.4375721723627966,
      "grad_norm": 0.6640233397483826,
      "learning_rate": 0.00010664005281596057,
      "loss": 1.0946,
      "step": 35480
    },
    {
      "epoch": 1.4379773505398998,
      "grad_norm": 0.6342954635620117,
      "learning_rate": 0.00010659768623255106,
      "loss": 1.0686,
      "step": 35490
    },
    {
      "epoch": 1.4383825287170033,
      "grad_norm": 0.6378554105758667,
      "learning_rate": 0.00010655531845969389,
      "loss": 1.0707,
      "step": 35500
    },
    {
      "epoch": 1.4387877068941066,
      "grad_norm": 0.6042903661727905,
      "learning_rate": 0.00010651294950502719,
      "loss": 1.0962,
      "step": 35510
    },
    {
      "epoch": 1.4391928850712101,
      "grad_norm": 0.5959990620613098,
      "learning_rate": 0.00010647057937618939,
      "loss": 1.0546,
      "step": 35520
    },
    {
      "epoch": 1.4395980632483134,
      "grad_norm": 0.6276496052742004,
      "learning_rate": 0.00010642820808081906,
      "loss": 1.0586,
      "step": 35530
    },
    {
      "epoch": 1.440003241425417,
      "grad_norm": 0.652094304561615,
      "learning_rate": 0.00010638583562655498,
      "loss": 1.0993,
      "step": 35540
    },
    {
      "epoch": 1.4404084196025202,
      "grad_norm": 0.6549781560897827,
      "learning_rate": 0.00010634346202103622,
      "loss": 1.0707,
      "step": 35550
    },
    {
      "epoch": 1.4408135977796235,
      "grad_norm": 0.6453515887260437,
      "learning_rate": 0.00010630108727190198,
      "loss": 1.1222,
      "step": 35560
    },
    {
      "epoch": 1.441218775956727,
      "grad_norm": 0.5899420380592346,
      "learning_rate": 0.00010625871138679165,
      "loss": 1.1275,
      "step": 35570
    },
    {
      "epoch": 1.4416239541338305,
      "grad_norm": 0.5604794025421143,
      "learning_rate": 0.0001062163343733449,
      "loss": 1.0909,
      "step": 35580
    },
    {
      "epoch": 1.4420291323109338,
      "grad_norm": 0.6089913845062256,
      "learning_rate": 0.00010617395623920159,
      "loss": 1.108,
      "step": 35590
    },
    {
      "epoch": 1.442434310488037,
      "grad_norm": 0.5732777714729309,
      "learning_rate": 0.00010613157699200167,
      "loss": 1.1132,
      "step": 35600
    },
    {
      "epoch": 1.4428394886651406,
      "grad_norm": 0.5890772342681885,
      "learning_rate": 0.00010608919663938549,
      "loss": 1.1314,
      "step": 35610
    },
    {
      "epoch": 1.4432446668422438,
      "grad_norm": 0.6442965865135193,
      "learning_rate": 0.0001060468151889934,
      "loss": 1.067,
      "step": 35620
    },
    {
      "epoch": 1.4436498450193471,
      "grad_norm": 0.6969403028488159,
      "learning_rate": 0.00010600443264846607,
      "loss": 1.1188,
      "step": 35630
    },
    {
      "epoch": 1.4440550231964506,
      "grad_norm": 0.5440079569816589,
      "learning_rate": 0.00010596204902544435,
      "loss": 1.1242,
      "step": 35640
    },
    {
      "epoch": 1.4444602013735541,
      "grad_norm": 0.655592143535614,
      "learning_rate": 0.00010591966432756923,
      "loss": 1.0806,
      "step": 35650
    },
    {
      "epoch": 1.4448653795506574,
      "grad_norm": 0.5828132629394531,
      "learning_rate": 0.00010587727856248195,
      "loss": 1.0869,
      "step": 35660
    },
    {
      "epoch": 1.4452705577277607,
      "grad_norm": 0.703987181186676,
      "learning_rate": 0.00010583489173782395,
      "loss": 1.1068,
      "step": 35670
    },
    {
      "epoch": 1.4456757359048642,
      "grad_norm": 0.6716835498809814,
      "learning_rate": 0.00010579250386123676,
      "loss": 1.0809,
      "step": 35680
    },
    {
      "epoch": 1.4460809140819675,
      "grad_norm": 0.669461190700531,
      "learning_rate": 0.00010575011494036226,
      "loss": 1.1031,
      "step": 35690
    },
    {
      "epoch": 1.446486092259071,
      "grad_norm": 0.5596643686294556,
      "learning_rate": 0.00010570772498284238,
      "loss": 1.0943,
      "step": 35700
    },
    {
      "epoch": 1.4468912704361743,
      "grad_norm": 0.6462826132774353,
      "learning_rate": 0.0001056653339963193,
      "loss": 1.0557,
      "step": 35710
    },
    {
      "epoch": 1.4472964486132778,
      "grad_norm": 0.563184916973114,
      "learning_rate": 0.00010562294198843539,
      "loss": 1.0557,
      "step": 35720
    },
    {
      "epoch": 1.447701626790381,
      "grad_norm": 0.5846025347709656,
      "learning_rate": 0.00010558054896683318,
      "loss": 1.0899,
      "step": 35730
    },
    {
      "epoch": 1.4481068049674843,
      "grad_norm": 0.6523349285125732,
      "learning_rate": 0.0001055381549391554,
      "loss": 1.1116,
      "step": 35740
    },
    {
      "epoch": 1.4485119831445878,
      "grad_norm": 0.5572874546051025,
      "learning_rate": 0.00010549575991304492,
      "loss": 1.118,
      "step": 35750
    },
    {
      "epoch": 1.4489171613216913,
      "grad_norm": 0.7136712670326233,
      "learning_rate": 0.00010545336389614486,
      "loss": 1.0806,
      "step": 35760
    },
    {
      "epoch": 1.4493223394987946,
      "grad_norm": 0.6438375115394592,
      "learning_rate": 0.00010541096689609846,
      "loss": 1.1267,
      "step": 35770
    },
    {
      "epoch": 1.449727517675898,
      "grad_norm": 0.5775963068008423,
      "learning_rate": 0.00010536856892054918,
      "loss": 1.1012,
      "step": 35780
    },
    {
      "epoch": 1.4501326958530014,
      "grad_norm": 0.6352435350418091,
      "learning_rate": 0.00010532616997714064,
      "loss": 1.0787,
      "step": 35790
    },
    {
      "epoch": 1.4505378740301047,
      "grad_norm": 0.7155622839927673,
      "learning_rate": 0.00010528377007351658,
      "loss": 1.0531,
      "step": 35800
    },
    {
      "epoch": 1.450943052207208,
      "grad_norm": 0.6627151966094971,
      "learning_rate": 0.00010524136921732101,
      "loss": 1.0752,
      "step": 35810
    },
    {
      "epoch": 1.4513482303843115,
      "grad_norm": 0.6571602821350098,
      "learning_rate": 0.00010519896741619803,
      "loss": 1.067,
      "step": 35820
    },
    {
      "epoch": 1.451753408561415,
      "grad_norm": 0.6245205998420715,
      "learning_rate": 0.000105156564677792,
      "loss": 1.0901,
      "step": 35830
    },
    {
      "epoch": 1.4521585867385183,
      "grad_norm": 0.6179918050765991,
      "learning_rate": 0.00010511416100974733,
      "loss": 1.0781,
      "step": 35840
    },
    {
      "epoch": 1.4525637649156216,
      "grad_norm": 0.6468571424484253,
      "learning_rate": 0.0001050717564197087,
      "loss": 1.1178,
      "step": 35850
    },
    {
      "epoch": 1.452968943092725,
      "grad_norm": 0.6318559646606445,
      "learning_rate": 0.00010502935091532092,
      "loss": 1.0976,
      "step": 35860
    },
    {
      "epoch": 1.4533741212698283,
      "grad_norm": 0.6228542327880859,
      "learning_rate": 0.00010498694450422895,
      "loss": 1.0835,
      "step": 35870
    },
    {
      "epoch": 1.4537792994469318,
      "grad_norm": 0.6514025330543518,
      "learning_rate": 0.00010494453719407788,
      "loss": 1.0852,
      "step": 35880
    },
    {
      "epoch": 1.4541844776240351,
      "grad_norm": 0.5839757323265076,
      "learning_rate": 0.00010490212899251309,
      "loss": 1.0795,
      "step": 35890
    },
    {
      "epoch": 1.4545896558011386,
      "grad_norm": 0.5875738859176636,
      "learning_rate": 0.00010485971990718,
      "loss": 1.1033,
      "step": 35900
    },
    {
      "epoch": 1.454994833978242,
      "grad_norm": 0.5906392931938171,
      "learning_rate": 0.0001048173099457242,
      "loss": 1.1107,
      "step": 35910
    },
    {
      "epoch": 1.4554000121553452,
      "grad_norm": 0.5959334373474121,
      "learning_rate": 0.00010477489911579154,
      "loss": 1.0654,
      "step": 35920
    },
    {
      "epoch": 1.4558051903324487,
      "grad_norm": 0.6081581711769104,
      "learning_rate": 0.00010473248742502788,
      "loss": 1.1102,
      "step": 35930
    },
    {
      "epoch": 1.4562103685095522,
      "grad_norm": 0.5640047788619995,
      "learning_rate": 0.00010469007488107934,
      "loss": 1.1198,
      "step": 35940
    },
    {
      "epoch": 1.4566155466866555,
      "grad_norm": 0.6615546345710754,
      "learning_rate": 0.00010464766149159215,
      "loss": 1.1255,
      "step": 35950
    },
    {
      "epoch": 1.4570207248637588,
      "grad_norm": 0.6099618077278137,
      "learning_rate": 0.00010460524726421275,
      "loss": 1.0571,
      "step": 35960
    },
    {
      "epoch": 1.4574259030408623,
      "grad_norm": 0.593173623085022,
      "learning_rate": 0.00010456283220658758,
      "loss": 1.1242,
      "step": 35970
    },
    {
      "epoch": 1.4578310812179656,
      "grad_norm": 1.2745981216430664,
      "learning_rate": 0.00010452041632636345,
      "loss": 1.0996,
      "step": 35980
    },
    {
      "epoch": 1.4582362593950688,
      "grad_norm": 0.613457977771759,
      "learning_rate": 0.00010447799963118712,
      "loss": 1.0441,
      "step": 35990
    },
    {
      "epoch": 1.4586414375721724,
      "grad_norm": 0.6279243230819702,
      "learning_rate": 0.00010443558212870563,
      "loss": 1.0998,
      "step": 36000
    },
    {
      "epoch": 1.4590466157492759,
      "grad_norm": 0.5532867312431335,
      "learning_rate": 0.0001043931638265661,
      "loss": 1.0956,
      "step": 36010
    },
    {
      "epoch": 1.4594517939263791,
      "grad_norm": 0.6463826298713684,
      "learning_rate": 0.0001043507447324158,
      "loss": 1.109,
      "step": 36020
    },
    {
      "epoch": 1.4598569721034824,
      "grad_norm": 0.6307222247123718,
      "learning_rate": 0.00010430832485390217,
      "loss": 1.0702,
      "step": 36030
    },
    {
      "epoch": 1.460262150280586,
      "grad_norm": 0.6102431416511536,
      "learning_rate": 0.00010426590419867274,
      "loss": 1.1055,
      "step": 36040
    },
    {
      "epoch": 1.4606673284576892,
      "grad_norm": 0.601505696773529,
      "learning_rate": 0.00010422348277437525,
      "loss": 1.1362,
      "step": 36050
    },
    {
      "epoch": 1.4610725066347927,
      "grad_norm": 0.5668410062789917,
      "learning_rate": 0.00010418106058865753,
      "loss": 1.0712,
      "step": 36060
    },
    {
      "epoch": 1.461477684811896,
      "grad_norm": 0.6057844758033752,
      "learning_rate": 0.00010413863764916758,
      "loss": 1.0985,
      "step": 36070
    },
    {
      "epoch": 1.4618828629889995,
      "grad_norm": 0.5929434299468994,
      "learning_rate": 0.00010409621396355347,
      "loss": 1.1131,
      "step": 36080
    },
    {
      "epoch": 1.4622880411661028,
      "grad_norm": 0.6920481324195862,
      "learning_rate": 0.00010405378953946347,
      "loss": 1.099,
      "step": 36090
    },
    {
      "epoch": 1.462693219343206,
      "grad_norm": 0.592388927936554,
      "learning_rate": 0.00010401136438454599,
      "loss": 1.1142,
      "step": 36100
    },
    {
      "epoch": 1.4630983975203096,
      "grad_norm": 0.5957005620002747,
      "learning_rate": 0.00010396893850644951,
      "loss": 1.0883,
      "step": 36110
    },
    {
      "epoch": 1.463503575697413,
      "grad_norm": 0.6022582054138184,
      "learning_rate": 0.0001039265119128227,
      "loss": 1.0909,
      "step": 36120
    },
    {
      "epoch": 1.4639087538745164,
      "grad_norm": 0.5572591423988342,
      "learning_rate": 0.00010388408461131429,
      "loss": 1.1039,
      "step": 36130
    },
    {
      "epoch": 1.4643139320516196,
      "grad_norm": 0.6394677758216858,
      "learning_rate": 0.00010384165660957326,
      "loss": 1.088,
      "step": 36140
    },
    {
      "epoch": 1.4647191102287231,
      "grad_norm": 0.6048550009727478,
      "learning_rate": 0.00010379922791524859,
      "loss": 1.0874,
      "step": 36150
    },
    {
      "epoch": 1.4651242884058264,
      "grad_norm": 0.666379988193512,
      "learning_rate": 0.00010375679853598941,
      "loss": 1.0577,
      "step": 36160
    },
    {
      "epoch": 1.4655294665829297,
      "grad_norm": 0.6133357882499695,
      "learning_rate": 0.00010371436847944503,
      "loss": 1.1128,
      "step": 36170
    },
    {
      "epoch": 1.4659346447600332,
      "grad_norm": 0.5964455604553223,
      "learning_rate": 0.00010367193775326484,
      "loss": 1.1,
      "step": 36180
    },
    {
      "epoch": 1.4663398229371367,
      "grad_norm": 0.5696868896484375,
      "learning_rate": 0.00010362950636509836,
      "loss": 1.0655,
      "step": 36190
    },
    {
      "epoch": 1.46674500111424,
      "grad_norm": 0.5981956720352173,
      "learning_rate": 0.00010358707432259525,
      "loss": 1.118,
      "step": 36200
    },
    {
      "epoch": 1.4671501792913433,
      "grad_norm": 0.6283653974533081,
      "learning_rate": 0.00010354464163340523,
      "loss": 1.1114,
      "step": 36210
    },
    {
      "epoch": 1.4675553574684468,
      "grad_norm": 0.6082949042320251,
      "learning_rate": 0.00010350220830517819,
      "loss": 1.0615,
      "step": 36220
    },
    {
      "epoch": 1.46796053564555,
      "grad_norm": 0.5993151664733887,
      "learning_rate": 0.00010345977434556411,
      "loss": 1.0717,
      "step": 36230
    },
    {
      "epoch": 1.4683657138226536,
      "grad_norm": 0.5489810705184937,
      "learning_rate": 0.00010341733976221313,
      "loss": 1.028,
      "step": 36240
    },
    {
      "epoch": 1.4687708919997569,
      "grad_norm": 0.6038268208503723,
      "learning_rate": 0.00010337490456277538,
      "loss": 1.0525,
      "step": 36250
    },
    {
      "epoch": 1.4691760701768604,
      "grad_norm": 0.5743525624275208,
      "learning_rate": 0.0001033324687549013,
      "loss": 1.0904,
      "step": 36260
    },
    {
      "epoch": 1.4695812483539636,
      "grad_norm": 0.6313695907592773,
      "learning_rate": 0.00010329003234624124,
      "loss": 1.1033,
      "step": 36270
    },
    {
      "epoch": 1.469986426531067,
      "grad_norm": 0.6309458017349243,
      "learning_rate": 0.00010324759534444578,
      "loss": 1.0767,
      "step": 36280
    },
    {
      "epoch": 1.4703916047081704,
      "grad_norm": 0.6023590564727783,
      "learning_rate": 0.00010320515775716555,
      "loss": 1.0616,
      "step": 36290
    },
    {
      "epoch": 1.470796782885274,
      "grad_norm": 0.6120538711547852,
      "learning_rate": 0.00010316271959205128,
      "loss": 1.0875,
      "step": 36300
    },
    {
      "epoch": 1.4712019610623772,
      "grad_norm": 0.6118815541267395,
      "learning_rate": 0.00010312028085675391,
      "loss": 1.127,
      "step": 36310
    },
    {
      "epoch": 1.4716071392394805,
      "grad_norm": 0.612797200679779,
      "learning_rate": 0.00010307784155892437,
      "loss": 1.0602,
      "step": 36320
    },
    {
      "epoch": 1.472012317416584,
      "grad_norm": 0.6097478270530701,
      "learning_rate": 0.00010303540170621371,
      "loss": 1.0757,
      "step": 36330
    },
    {
      "epoch": 1.4724174955936873,
      "grad_norm": 0.6252867579460144,
      "learning_rate": 0.00010299296130627309,
      "loss": 1.0702,
      "step": 36340
    },
    {
      "epoch": 1.4728226737707906,
      "grad_norm": 0.6276163458824158,
      "learning_rate": 0.00010295052036675379,
      "loss": 1.094,
      "step": 36350
    },
    {
      "epoch": 1.473227851947894,
      "grad_norm": 0.604167103767395,
      "learning_rate": 0.00010290807889530716,
      "loss": 1.0886,
      "step": 36360
    },
    {
      "epoch": 1.4736330301249976,
      "grad_norm": 0.606410562992096,
      "learning_rate": 0.00010286563689958467,
      "loss": 1.1429,
      "step": 36370
    },
    {
      "epoch": 1.4740382083021009,
      "grad_norm": 0.5931729674339294,
      "learning_rate": 0.00010282319438723782,
      "loss": 1.0911,
      "step": 36380
    },
    {
      "epoch": 1.4744433864792041,
      "grad_norm": 0.6571753025054932,
      "learning_rate": 0.00010278075136591834,
      "loss": 1.0902,
      "step": 36390
    },
    {
      "epoch": 1.4748485646563076,
      "grad_norm": 0.5926054120063782,
      "learning_rate": 0.00010273830784327792,
      "loss": 1.1144,
      "step": 36400
    },
    {
      "epoch": 1.475253742833411,
      "grad_norm": 0.5869762897491455,
      "learning_rate": 0.00010269586382696836,
      "loss": 1.0935,
      "step": 36410
    },
    {
      "epoch": 1.4756589210105144,
      "grad_norm": 0.6073912382125854,
      "learning_rate": 0.00010265341932464163,
      "loss": 1.1049,
      "step": 36420
    },
    {
      "epoch": 1.4760640991876177,
      "grad_norm": 0.6524277925491333,
      "learning_rate": 0.0001026109743439497,
      "loss": 1.1016,
      "step": 36430
    },
    {
      "epoch": 1.4764692773647212,
      "grad_norm": 0.6127753853797913,
      "learning_rate": 0.00010256852889254464,
      "loss": 1.0732,
      "step": 36440
    },
    {
      "epoch": 1.4768744555418245,
      "grad_norm": 0.6396984457969666,
      "learning_rate": 0.00010252608297807871,
      "loss": 1.0894,
      "step": 36450
    },
    {
      "epoch": 1.4772796337189278,
      "grad_norm": 0.6863594651222229,
      "learning_rate": 0.00010248363660820409,
      "loss": 1.1083,
      "step": 36460
    },
    {
      "epoch": 1.4776848118960313,
      "grad_norm": 0.6073252558708191,
      "learning_rate": 0.00010244118979057309,
      "loss": 1.0906,
      "step": 36470
    },
    {
      "epoch": 1.4780899900731348,
      "grad_norm": 0.533916711807251,
      "learning_rate": 0.00010239874253283821,
      "loss": 1.1052,
      "step": 36480
    },
    {
      "epoch": 1.478495168250238,
      "grad_norm": 0.6530753374099731,
      "learning_rate": 0.00010235629484265194,
      "loss": 1.0775,
      "step": 36490
    },
    {
      "epoch": 1.4789003464273414,
      "grad_norm": 0.5883758068084717,
      "learning_rate": 0.00010231384672766679,
      "loss": 1.0828,
      "step": 36500
    },
    {
      "epoch": 1.4793055246044449,
      "grad_norm": 0.6443793177604675,
      "learning_rate": 0.00010227139819553549,
      "loss": 1.0981,
      "step": 36510
    },
    {
      "epoch": 1.4797107027815481,
      "grad_norm": 0.6065267324447632,
      "learning_rate": 0.00010222894925391073,
      "loss": 1.1054,
      "step": 36520
    },
    {
      "epoch": 1.4801158809586514,
      "grad_norm": 0.584169864654541,
      "learning_rate": 0.00010218649991044532,
      "loss": 1.0991,
      "step": 36530
    },
    {
      "epoch": 1.480521059135755,
      "grad_norm": 0.5831345319747925,
      "learning_rate": 0.00010214405017279217,
      "loss": 1.0586,
      "step": 36540
    },
    {
      "epoch": 1.4809262373128584,
      "grad_norm": 0.5987511277198792,
      "learning_rate": 0.00010210160004860417,
      "loss": 1.0704,
      "step": 36550
    },
    {
      "epoch": 1.4813314154899617,
      "grad_norm": 0.6094339489936829,
      "learning_rate": 0.00010205914954553433,
      "loss": 1.1198,
      "step": 36560
    },
    {
      "epoch": 1.481736593667065,
      "grad_norm": 0.5896655917167664,
      "learning_rate": 0.0001020166986712358,
      "loss": 1.0606,
      "step": 36570
    },
    {
      "epoch": 1.4821417718441685,
      "grad_norm": 0.5944350957870483,
      "learning_rate": 0.0001019742474333617,
      "loss": 1.1183,
      "step": 36580
    },
    {
      "epoch": 1.4825469500212718,
      "grad_norm": 0.5621413588523865,
      "learning_rate": 0.00010193179583956523,
      "loss": 1.1424,
      "step": 36590
    },
    {
      "epoch": 1.4829521281983753,
      "grad_norm": 0.5802322626113892,
      "learning_rate": 0.00010188934389749968,
      "loss": 1.0884,
      "step": 36600
    },
    {
      "epoch": 1.4833573063754786,
      "grad_norm": 0.6104912161827087,
      "learning_rate": 0.0001018468916148184,
      "loss": 1.1007,
      "step": 36610
    },
    {
      "epoch": 1.483762484552582,
      "grad_norm": 0.6180727481842041,
      "learning_rate": 0.00010180443899917482,
      "loss": 1.0668,
      "step": 36620
    },
    {
      "epoch": 1.4841676627296854,
      "grad_norm": 0.6500835418701172,
      "learning_rate": 0.00010176198605822238,
      "loss": 1.0863,
      "step": 36630
    },
    {
      "epoch": 1.4845728409067886,
      "grad_norm": 0.6133726835250854,
      "learning_rate": 0.00010171953279961459,
      "loss": 1.0721,
      "step": 36640
    },
    {
      "epoch": 1.4849780190838922,
      "grad_norm": 0.6091869473457336,
      "learning_rate": 0.00010167707923100507,
      "loss": 1.0823,
      "step": 36650
    },
    {
      "epoch": 1.4853831972609957,
      "grad_norm": 0.5697134137153625,
      "learning_rate": 0.00010163462536004742,
      "loss": 1.1095,
      "step": 36660
    },
    {
      "epoch": 1.485788375438099,
      "grad_norm": 0.6629116535186768,
      "learning_rate": 0.00010159217119439533,
      "loss": 1.0785,
      "step": 36670
    },
    {
      "epoch": 1.4861935536152022,
      "grad_norm": 0.5888656973838806,
      "learning_rate": 0.0001015497167417026,
      "loss": 1.0574,
      "step": 36680
    },
    {
      "epoch": 1.4865987317923057,
      "grad_norm": 0.5856531858444214,
      "learning_rate": 0.00010150726200962294,
      "loss": 1.0435,
      "step": 36690
    },
    {
      "epoch": 1.487003909969409,
      "grad_norm": 0.6397324800491333,
      "learning_rate": 0.0001014648070058103,
      "loss": 1.0963,
      "step": 36700
    },
    {
      "epoch": 1.4874090881465123,
      "grad_norm": 0.6533519625663757,
      "learning_rate": 0.00010142235173791847,
      "loss": 1.0992,
      "step": 36710
    },
    {
      "epoch": 1.4878142663236158,
      "grad_norm": 0.6663082838058472,
      "learning_rate": 0.00010137989621360149,
      "loss": 1.071,
      "step": 36720
    },
    {
      "epoch": 1.4882194445007193,
      "grad_norm": 0.5572199821472168,
      "learning_rate": 0.00010133744044051328,
      "loss": 1.0862,
      "step": 36730
    },
    {
      "epoch": 1.4886246226778226,
      "grad_norm": 0.6073725819587708,
      "learning_rate": 0.00010129498442630791,
      "loss": 1.1091,
      "step": 36740
    },
    {
      "epoch": 1.4890298008549259,
      "grad_norm": 0.5905301570892334,
      "learning_rate": 0.00010125252817863944,
      "loss": 1.1032,
      "step": 36750
    },
    {
      "epoch": 1.4894349790320294,
      "grad_norm": 0.5685065984725952,
      "learning_rate": 0.00010121007170516202,
      "loss": 1.1444,
      "step": 36760
    },
    {
      "epoch": 1.4898401572091327,
      "grad_norm": 0.5926529169082642,
      "learning_rate": 0.0001011676150135298,
      "loss": 1.0966,
      "step": 36770
    },
    {
      "epoch": 1.4902453353862362,
      "grad_norm": 0.6164329648017883,
      "learning_rate": 0.00010112515811139693,
      "loss": 1.0869,
      "step": 36780
    },
    {
      "epoch": 1.4906505135633394,
      "grad_norm": 0.5323477983474731,
      "learning_rate": 0.00010108270100641773,
      "loss": 1.0938,
      "step": 36790
    },
    {
      "epoch": 1.491055691740443,
      "grad_norm": 0.5424414873123169,
      "learning_rate": 0.00010104024370624644,
      "loss": 1.0694,
      "step": 36800
    },
    {
      "epoch": 1.4914608699175462,
      "grad_norm": 0.612612247467041,
      "learning_rate": 0.00010099778621853733,
      "loss": 1.1177,
      "step": 36810
    },
    {
      "epoch": 1.4918660480946495,
      "grad_norm": 0.6570168733596802,
      "learning_rate": 0.0001009553285509448,
      "loss": 1.1033,
      "step": 36820
    },
    {
      "epoch": 1.492271226271753,
      "grad_norm": 0.6244915723800659,
      "learning_rate": 0.00010091287071112321,
      "loss": 1.0616,
      "step": 36830
    },
    {
      "epoch": 1.4926764044488565,
      "grad_norm": 0.608332633972168,
      "learning_rate": 0.00010087041270672695,
      "loss": 1.0745,
      "step": 36840
    },
    {
      "epoch": 1.4930815826259598,
      "grad_norm": 0.6677070260047913,
      "learning_rate": 0.00010082795454541048,
      "loss": 1.1086,
      "step": 36850
    },
    {
      "epoch": 1.493486760803063,
      "grad_norm": 0.5970058441162109,
      "learning_rate": 0.00010078549623482824,
      "loss": 1.0898,
      "step": 36860
    },
    {
      "epoch": 1.4938919389801666,
      "grad_norm": 0.606522262096405,
      "learning_rate": 0.00010074303778263474,
      "loss": 1.0794,
      "step": 36870
    },
    {
      "epoch": 1.4942971171572699,
      "grad_norm": 0.6687438488006592,
      "learning_rate": 0.00010070057919648452,
      "loss": 1.0835,
      "step": 36880
    },
    {
      "epoch": 1.4947022953343732,
      "grad_norm": 0.610047459602356,
      "learning_rate": 0.00010065812048403209,
      "loss": 1.1024,
      "step": 36890
    },
    {
      "epoch": 1.4951074735114767,
      "grad_norm": 0.5931307673454285,
      "learning_rate": 0.00010061566165293201,
      "loss": 1.0796,
      "step": 36900
    },
    {
      "epoch": 1.4955126516885802,
      "grad_norm": 0.6280748844146729,
      "learning_rate": 0.00010057320271083889,
      "loss": 1.0774,
      "step": 36910
    },
    {
      "epoch": 1.4959178298656834,
      "grad_norm": 0.6080884337425232,
      "learning_rate": 0.00010053074366540732,
      "loss": 1.1113,
      "step": 36920
    },
    {
      "epoch": 1.4963230080427867,
      "grad_norm": 0.5972120761871338,
      "learning_rate": 0.00010048828452429195,
      "loss": 1.0925,
      "step": 36930
    },
    {
      "epoch": 1.4967281862198902,
      "grad_norm": 0.6744397878646851,
      "learning_rate": 0.00010044582529514739,
      "loss": 1.0914,
      "step": 36940
    },
    {
      "epoch": 1.4971333643969935,
      "grad_norm": 0.601017951965332,
      "learning_rate": 0.0001004033659856283,
      "loss": 1.0689,
      "step": 36950
    },
    {
      "epoch": 1.497538542574097,
      "grad_norm": 0.5675349831581116,
      "learning_rate": 0.00010036090660338937,
      "loss": 1.0734,
      "step": 36960
    },
    {
      "epoch": 1.4979437207512003,
      "grad_norm": 0.5935280323028564,
      "learning_rate": 0.0001003184471560853,
      "loss": 1.092,
      "step": 36970
    },
    {
      "epoch": 1.4983488989283038,
      "grad_norm": 0.6260930299758911,
      "learning_rate": 0.00010027598765137072,
      "loss": 1.0977,
      "step": 36980
    },
    {
      "epoch": 1.498754077105407,
      "grad_norm": 0.5303331017494202,
      "learning_rate": 0.00010023352809690041,
      "loss": 1.1373,
      "step": 36990
    },
    {
      "epoch": 1.4991592552825104,
      "grad_norm": 0.5695198774337769,
      "learning_rate": 0.00010019106850032907,
      "loss": 1.0651,
      "step": 37000
    },
    {
      "epoch": 1.4995644334596139,
      "grad_norm": 0.668104350566864,
      "learning_rate": 0.00010014860886931139,
      "loss": 1.0856,
      "step": 37010
    },
    {
      "epoch": 1.4999696116367174,
      "grad_norm": 0.6332420110702515,
      "learning_rate": 0.00010010614921150214,
      "loss": 1.0759,
      "step": 37020
    },
    {
      "epoch": 1.5003747898138207,
      "grad_norm": 0.6350727081298828,
      "learning_rate": 0.00010006368953455602,
      "loss": 1.1008,
      "step": 37030
    },
    {
      "epoch": 1.500779967990924,
      "grad_norm": 0.522470235824585,
      "learning_rate": 0.0001000212298461278,
      "loss": 1.0446,
      "step": 37040
    },
    {
      "epoch": 1.5011851461680275,
      "grad_norm": 0.6215276122093201,
      "learning_rate": 9.997877015387222e-05,
      "loss": 1.1153,
      "step": 37050
    },
    {
      "epoch": 1.5015903243451307,
      "grad_norm": 0.7171666622161865,
      "learning_rate": 9.9936310465444e-05,
      "loss": 1.0998,
      "step": 37060
    },
    {
      "epoch": 1.501995502522234,
      "grad_norm": 0.5876935124397278,
      "learning_rate": 9.98938507884979e-05,
      "loss": 1.0861,
      "step": 37070
    },
    {
      "epoch": 1.5024006806993375,
      "grad_norm": 0.6662000417709351,
      "learning_rate": 9.985139113068865e-05,
      "loss": 1.0843,
      "step": 37080
    },
    {
      "epoch": 1.502805858876441,
      "grad_norm": 0.5971617698669434,
      "learning_rate": 9.980893149967098e-05,
      "loss": 1.0799,
      "step": 37090
    },
    {
      "epoch": 1.5032110370535443,
      "grad_norm": 0.6066619157791138,
      "learning_rate": 9.976647190309964e-05,
      "loss": 1.1227,
      "step": 37100
    },
    {
      "epoch": 1.5036162152306476,
      "grad_norm": 0.6394475698471069,
      "learning_rate": 9.972401234862928e-05,
      "loss": 1.0704,
      "step": 37110
    },
    {
      "epoch": 1.504021393407751,
      "grad_norm": 0.5686768889427185,
      "learning_rate": 9.968155284391474e-05,
      "loss": 1.0875,
      "step": 37120
    },
    {
      "epoch": 1.5044265715848546,
      "grad_norm": 0.5881919860839844,
      "learning_rate": 9.963909339661065e-05,
      "loss": 1.0655,
      "step": 37130
    },
    {
      "epoch": 1.5048317497619577,
      "grad_norm": 0.6650229692459106,
      "learning_rate": 9.959663401437174e-05,
      "loss": 1.0562,
      "step": 37140
    },
    {
      "epoch": 1.5052369279390612,
      "grad_norm": 0.576485276222229,
      "learning_rate": 9.955417470485265e-05,
      "loss": 1.1121,
      "step": 37150
    },
    {
      "epoch": 1.5056421061161647,
      "grad_norm": 0.5821657180786133,
      "learning_rate": 9.951171547570806e-05,
      "loss": 1.1054,
      "step": 37160
    },
    {
      "epoch": 1.506047284293268,
      "grad_norm": 0.6484360098838806,
      "learning_rate": 9.946925633459268e-05,
      "loss": 1.1103,
      "step": 37170
    },
    {
      "epoch": 1.5064524624703712,
      "grad_norm": 0.6982921957969666,
      "learning_rate": 9.942679728916112e-05,
      "loss": 1.0409,
      "step": 37180
    },
    {
      "epoch": 1.5068576406474747,
      "grad_norm": 0.6671663522720337,
      "learning_rate": 9.938433834706801e-05,
      "loss": 1.0947,
      "step": 37190
    },
    {
      "epoch": 1.5072628188245782,
      "grad_norm": 0.6114413142204285,
      "learning_rate": 9.934187951596795e-05,
      "loss": 1.1058,
      "step": 37200
    },
    {
      "epoch": 1.5076679970016815,
      "grad_norm": 0.6305966973304749,
      "learning_rate": 9.929942080351551e-05,
      "loss": 1.1062,
      "step": 37210
    },
    {
      "epoch": 1.5080731751787848,
      "grad_norm": 0.6196856498718262,
      "learning_rate": 9.925696221736525e-05,
      "loss": 1.128,
      "step": 37220
    },
    {
      "epoch": 1.5084783533558883,
      "grad_norm": 0.5612085461616516,
      "learning_rate": 9.921450376517176e-05,
      "loss": 1.0896,
      "step": 37230
    },
    {
      "epoch": 1.5088835315329916,
      "grad_norm": 0.6142746210098267,
      "learning_rate": 9.917204545458955e-05,
      "loss": 1.0815,
      "step": 37240
    },
    {
      "epoch": 1.5092887097100949,
      "grad_norm": 0.6135663390159607,
      "learning_rate": 9.912958729327309e-05,
      "loss": 1.0815,
      "step": 37250
    },
    {
      "epoch": 1.5096938878871984,
      "grad_norm": 0.5763309597969055,
      "learning_rate": 9.908712928887684e-05,
      "loss": 1.1156,
      "step": 37260
    },
    {
      "epoch": 1.5100990660643019,
      "grad_norm": 0.5741956233978271,
      "learning_rate": 9.904467144905525e-05,
      "loss": 1.0717,
      "step": 37270
    },
    {
      "epoch": 1.5105042442414052,
      "grad_norm": 0.6799144744873047,
      "learning_rate": 9.900221378146268e-05,
      "loss": 1.113,
      "step": 37280
    },
    {
      "epoch": 1.5109094224185085,
      "grad_norm": 0.5597561597824097,
      "learning_rate": 9.895975629375359e-05,
      "loss": 1.084,
      "step": 37290
    },
    {
      "epoch": 1.511314600595612,
      "grad_norm": 0.6127524375915527,
      "learning_rate": 9.89172989935823e-05,
      "loss": 1.0655,
      "step": 37300
    },
    {
      "epoch": 1.5117197787727155,
      "grad_norm": 0.5680673122406006,
      "learning_rate": 9.887484188860309e-05,
      "loss": 1.0893,
      "step": 37310
    },
    {
      "epoch": 1.5121249569498185,
      "grad_norm": 0.6223932504653931,
      "learning_rate": 9.883238498647025e-05,
      "loss": 1.0394,
      "step": 37320
    },
    {
      "epoch": 1.512530135126922,
      "grad_norm": 0.6610687971115112,
      "learning_rate": 9.878992829483798e-05,
      "loss": 1.0961,
      "step": 37330
    },
    {
      "epoch": 1.5129353133040255,
      "grad_norm": 0.6296534538269043,
      "learning_rate": 9.874747182136055e-05,
      "loss": 1.0642,
      "step": 37340
    },
    {
      "epoch": 1.5133404914811288,
      "grad_norm": 0.6366645097732544,
      "learning_rate": 9.87050155736921e-05,
      "loss": 1.0922,
      "step": 37350
    },
    {
      "epoch": 1.513745669658232,
      "grad_norm": 0.5641550421714783,
      "learning_rate": 9.866255955948676e-05,
      "loss": 1.1023,
      "step": 37360
    },
    {
      "epoch": 1.5141508478353356,
      "grad_norm": 0.5766530632972717,
      "learning_rate": 9.862010378639854e-05,
      "loss": 1.0796,
      "step": 37370
    },
    {
      "epoch": 1.514556026012439,
      "grad_norm": 0.629383385181427,
      "learning_rate": 9.857764826208155e-05,
      "loss": 1.1024,
      "step": 37380
    },
    {
      "epoch": 1.5149612041895424,
      "grad_norm": 0.6079122424125671,
      "learning_rate": 9.853519299418974e-05,
      "loss": 1.1129,
      "step": 37390
    },
    {
      "epoch": 1.5153663823666457,
      "grad_norm": 0.5600486397743225,
      "learning_rate": 9.849273799037708e-05,
      "loss": 1.107,
      "step": 37400
    },
    {
      "epoch": 1.5157715605437492,
      "grad_norm": 0.5765146017074585,
      "learning_rate": 9.845028325829744e-05,
      "loss": 1.088,
      "step": 37410
    },
    {
      "epoch": 1.5161767387208525,
      "grad_norm": 0.5608596801757812,
      "learning_rate": 9.84078288056047e-05,
      "loss": 1.0898,
      "step": 37420
    },
    {
      "epoch": 1.5165819168979557,
      "grad_norm": 0.6206350326538086,
      "learning_rate": 9.836537463995262e-05,
      "loss": 1.0665,
      "step": 37430
    },
    {
      "epoch": 1.5169870950750592,
      "grad_norm": 0.784157395362854,
      "learning_rate": 9.832292076899497e-05,
      "loss": 1.1241,
      "step": 37440
    },
    {
      "epoch": 1.5173922732521627,
      "grad_norm": 0.7105434536933899,
      "learning_rate": 9.828046720038542e-05,
      "loss": 1.135,
      "step": 37450
    },
    {
      "epoch": 1.517797451429266,
      "grad_norm": 0.5889471173286438,
      "learning_rate": 9.823801394177764e-05,
      "loss": 1.0351,
      "step": 37460
    },
    {
      "epoch": 1.5182026296063693,
      "grad_norm": 0.6353241205215454,
      "learning_rate": 9.81955610008252e-05,
      "loss": 1.1039,
      "step": 37470
    },
    {
      "epoch": 1.5186078077834728,
      "grad_norm": 0.6204172372817993,
      "learning_rate": 9.815310838518161e-05,
      "loss": 1.09,
      "step": 37480
    },
    {
      "epoch": 1.5190129859605763,
      "grad_norm": 0.5438039302825928,
      "learning_rate": 9.811065610250035e-05,
      "loss": 1.1065,
      "step": 37490
    },
    {
      "epoch": 1.5194181641376794,
      "grad_norm": 0.5423998236656189,
      "learning_rate": 9.806820416043478e-05,
      "loss": 1.0891,
      "step": 37500
    },
    {
      "epoch": 1.5198233423147829,
      "grad_norm": 0.5923877954483032,
      "learning_rate": 9.802575256663833e-05,
      "loss": 1.1184,
      "step": 37510
    },
    {
      "epoch": 1.5202285204918864,
      "grad_norm": 0.5731462240219116,
      "learning_rate": 9.798330132876423e-05,
      "loss": 1.0989,
      "step": 37520
    },
    {
      "epoch": 1.5206336986689897,
      "grad_norm": 0.5892829298973083,
      "learning_rate": 9.79408504544657e-05,
      "loss": 1.0818,
      "step": 37530
    },
    {
      "epoch": 1.521038876846093,
      "grad_norm": 0.6960386633872986,
      "learning_rate": 9.789839995139589e-05,
      "loss": 1.097,
      "step": 37540
    },
    {
      "epoch": 1.5214440550231965,
      "grad_norm": 0.5625761151313782,
      "learning_rate": 9.78559498272079e-05,
      "loss": 1.076,
      "step": 37550
    },
    {
      "epoch": 1.5218492332003,
      "grad_norm": 0.6747767925262451,
      "learning_rate": 9.781350008955468e-05,
      "loss": 1.1057,
      "step": 37560
    },
    {
      "epoch": 1.5222544113774032,
      "grad_norm": 0.6134365200996399,
      "learning_rate": 9.777105074608928e-05,
      "loss": 1.0838,
      "step": 37570
    },
    {
      "epoch": 1.5226595895545065,
      "grad_norm": 0.5866749882698059,
      "learning_rate": 9.772860180446452e-05,
      "loss": 1.0677,
      "step": 37580
    },
    {
      "epoch": 1.52306476773161,
      "grad_norm": 0.6485437154769897,
      "learning_rate": 9.768615327233322e-05,
      "loss": 1.0786,
      "step": 37590
    },
    {
      "epoch": 1.5234699459087133,
      "grad_norm": 0.5941844582557678,
      "learning_rate": 9.764370515734811e-05,
      "loss": 1.0973,
      "step": 37600
    },
    {
      "epoch": 1.5238751240858166,
      "grad_norm": 0.5568268895149231,
      "learning_rate": 9.760125746716178e-05,
      "loss": 1.0381,
      "step": 37610
    },
    {
      "epoch": 1.52428030226292,
      "grad_norm": 0.6648565530776978,
      "learning_rate": 9.755881020942691e-05,
      "loss": 1.0728,
      "step": 37620
    },
    {
      "epoch": 1.5246854804400236,
      "grad_norm": 0.59449702501297,
      "learning_rate": 9.751636339179595e-05,
      "loss": 1.0915,
      "step": 37630
    },
    {
      "epoch": 1.525090658617127,
      "grad_norm": 0.5744891166687012,
      "learning_rate": 9.747391702192132e-05,
      "loss": 1.0882,
      "step": 37640
    },
    {
      "epoch": 1.5254958367942302,
      "grad_norm": 0.6622604727745056,
      "learning_rate": 9.743147110745537e-05,
      "loss": 1.1268,
      "step": 37650
    },
    {
      "epoch": 1.5259010149713337,
      "grad_norm": 0.6324489116668701,
      "learning_rate": 9.738902565605032e-05,
      "loss": 1.1234,
      "step": 37660
    },
    {
      "epoch": 1.5263061931484372,
      "grad_norm": 0.6724990606307983,
      "learning_rate": 9.734658067535838e-05,
      "loss": 1.1196,
      "step": 37670
    },
    {
      "epoch": 1.5267113713255402,
      "grad_norm": 0.7306877970695496,
      "learning_rate": 9.730413617303165e-05,
      "loss": 1.0756,
      "step": 37680
    },
    {
      "epoch": 1.5271165495026437,
      "grad_norm": 0.6941239237785339,
      "learning_rate": 9.726169215672212e-05,
      "loss": 1.1347,
      "step": 37690
    },
    {
      "epoch": 1.5275217276797473,
      "grad_norm": 0.5901744961738586,
      "learning_rate": 9.721924863408167e-05,
      "loss": 1.0777,
      "step": 37700
    },
    {
      "epoch": 1.5279269058568505,
      "grad_norm": 0.7050279378890991,
      "learning_rate": 9.717680561276219e-05,
      "loss": 1.0749,
      "step": 37710
    },
    {
      "epoch": 1.5283320840339538,
      "grad_norm": 0.5323678851127625,
      "learning_rate": 9.713436310041538e-05,
      "loss": 1.0889,
      "step": 37720
    },
    {
      "epoch": 1.5287372622110573,
      "grad_norm": 0.5957754850387573,
      "learning_rate": 9.709192110469287e-05,
      "loss": 1.1414,
      "step": 37730
    },
    {
      "epoch": 1.5291424403881608,
      "grad_norm": 0.684363842010498,
      "learning_rate": 9.704947963324623e-05,
      "loss": 1.1008,
      "step": 37740
    },
    {
      "epoch": 1.529547618565264,
      "grad_norm": 0.5662672519683838,
      "learning_rate": 9.700703869372693e-05,
      "loss": 1.1273,
      "step": 37750
    },
    {
      "epoch": 1.5299527967423674,
      "grad_norm": 0.6172959208488464,
      "learning_rate": 9.696459829378633e-05,
      "loss": 1.1027,
      "step": 37760
    },
    {
      "epoch": 1.530357974919471,
      "grad_norm": 0.5194318294525146,
      "learning_rate": 9.692215844107567e-05,
      "loss": 1.0783,
      "step": 37770
    },
    {
      "epoch": 1.5307631530965742,
      "grad_norm": 0.7022966146469116,
      "learning_rate": 9.687971914324607e-05,
      "loss": 1.0684,
      "step": 37780
    },
    {
      "epoch": 1.5311683312736775,
      "grad_norm": 0.651706337928772,
      "learning_rate": 9.68372804079487e-05,
      "loss": 1.0922,
      "step": 37790
    },
    {
      "epoch": 1.531573509450781,
      "grad_norm": 0.6092678904533386,
      "learning_rate": 9.679484224283449e-05,
      "loss": 1.1002,
      "step": 37800
    },
    {
      "epoch": 1.5319786876278845,
      "grad_norm": 0.5481024384498596,
      "learning_rate": 9.675240465555426e-05,
      "loss": 1.0675,
      "step": 37810
    },
    {
      "epoch": 1.5323838658049878,
      "grad_norm": 0.5915434956550598,
      "learning_rate": 9.670996765375881e-05,
      "loss": 1.0696,
      "step": 37820
    },
    {
      "epoch": 1.532789043982091,
      "grad_norm": 0.6262772083282471,
      "learning_rate": 9.666753124509875e-05,
      "loss": 1.0932,
      "step": 37830
    },
    {
      "epoch": 1.5331942221591945,
      "grad_norm": 0.5769603252410889,
      "learning_rate": 9.662509543722461e-05,
      "loss": 1.0844,
      "step": 37840
    },
    {
      "epoch": 1.533599400336298,
      "grad_norm": 0.5739287734031677,
      "learning_rate": 9.658266023778689e-05,
      "loss": 1.0968,
      "step": 37850
    },
    {
      "epoch": 1.534004578513401,
      "grad_norm": 0.6166068911552429,
      "learning_rate": 9.65402256544359e-05,
      "loss": 1.0871,
      "step": 37860
    },
    {
      "epoch": 1.5344097566905046,
      "grad_norm": 0.6060196161270142,
      "learning_rate": 9.649779169482184e-05,
      "loss": 1.0876,
      "step": 37870
    },
    {
      "epoch": 1.5348149348676081,
      "grad_norm": 0.599209189414978,
      "learning_rate": 9.64553583665948e-05,
      "loss": 1.0915,
      "step": 37880
    },
    {
      "epoch": 1.5352201130447114,
      "grad_norm": 0.6452056169509888,
      "learning_rate": 9.64129256774048e-05,
      "loss": 1.0731,
      "step": 37890
    },
    {
      "epoch": 1.5356252912218147,
      "grad_norm": 0.6181835532188416,
      "learning_rate": 9.637049363490163e-05,
      "loss": 1.0685,
      "step": 37900
    },
    {
      "epoch": 1.5360304693989182,
      "grad_norm": 0.6041398644447327,
      "learning_rate": 9.632806224673517e-05,
      "loss": 1.1051,
      "step": 37910
    },
    {
      "epoch": 1.5364356475760217,
      "grad_norm": 0.5540357828140259,
      "learning_rate": 9.628563152055498e-05,
      "loss": 1.0765,
      "step": 37920
    },
    {
      "epoch": 1.536840825753125,
      "grad_norm": 0.6762358546257019,
      "learning_rate": 9.624320146401063e-05,
      "loss": 1.0967,
      "step": 37930
    },
    {
      "epoch": 1.5372460039302283,
      "grad_norm": 0.6156542301177979,
      "learning_rate": 9.620077208475146e-05,
      "loss": 1.0672,
      "step": 37940
    },
    {
      "epoch": 1.5376511821073318,
      "grad_norm": 0.6129539608955383,
      "learning_rate": 9.615834339042674e-05,
      "loss": 1.103,
      "step": 37950
    },
    {
      "epoch": 1.538056360284435,
      "grad_norm": 0.5505751371383667,
      "learning_rate": 9.611591538868569e-05,
      "loss": 1.0852,
      "step": 37960
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 0.6762579083442688,
      "learning_rate": 9.607348808717732e-05,
      "loss": 1.0823,
      "step": 37970
    },
    {
      "epoch": 1.5388667166386418,
      "grad_norm": 0.5867717266082764,
      "learning_rate": 9.603106149355052e-05,
      "loss": 1.0675,
      "step": 37980
    },
    {
      "epoch": 1.5392718948157453,
      "grad_norm": 0.6707775592803955,
      "learning_rate": 9.598863561545404e-05,
      "loss": 1.0667,
      "step": 37990
    },
    {
      "epoch": 1.5396770729928486,
      "grad_norm": 0.6111908555030823,
      "learning_rate": 9.594621046053655e-05,
      "loss": 1.0643,
      "step": 38000
    },
    {
      "epoch": 1.540082251169952,
      "grad_norm": 0.5623034834861755,
      "learning_rate": 9.590378603644654e-05,
      "loss": 1.0844,
      "step": 38010
    },
    {
      "epoch": 1.5404874293470554,
      "grad_norm": 0.6437550187110901,
      "learning_rate": 9.586136235083245e-05,
      "loss": 1.0446,
      "step": 38020
    },
    {
      "epoch": 1.540892607524159,
      "grad_norm": 0.6192774772644043,
      "learning_rate": 9.58189394113425e-05,
      "loss": 1.1085,
      "step": 38030
    },
    {
      "epoch": 1.541297785701262,
      "grad_norm": 0.5645702481269836,
      "learning_rate": 9.577651722562476e-05,
      "loss": 1.0727,
      "step": 38040
    },
    {
      "epoch": 1.5417029638783655,
      "grad_norm": 0.5665873885154724,
      "learning_rate": 9.573409580132728e-05,
      "loss": 1.0838,
      "step": 38050
    },
    {
      "epoch": 1.542108142055469,
      "grad_norm": 0.6304237842559814,
      "learning_rate": 9.569167514609786e-05,
      "loss": 1.1333,
      "step": 38060
    },
    {
      "epoch": 1.5425133202325723,
      "grad_norm": 0.6520474553108215,
      "learning_rate": 9.564925526758423e-05,
      "loss": 1.127,
      "step": 38070
    },
    {
      "epoch": 1.5429184984096755,
      "grad_norm": 0.6242173314094543,
      "learning_rate": 9.560683617343391e-05,
      "loss": 1.0865,
      "step": 38080
    },
    {
      "epoch": 1.543323676586779,
      "grad_norm": 0.5920267105102539,
      "learning_rate": 9.556441787129439e-05,
      "loss": 1.1186,
      "step": 38090
    },
    {
      "epoch": 1.5437288547638826,
      "grad_norm": 0.5540690422058105,
      "learning_rate": 9.55220003688129e-05,
      "loss": 1.1022,
      "step": 38100
    },
    {
      "epoch": 1.5441340329409858,
      "grad_norm": 0.6577634215354919,
      "learning_rate": 9.54795836736366e-05,
      "loss": 1.1066,
      "step": 38110
    },
    {
      "epoch": 1.5445392111180891,
      "grad_norm": 0.6142936944961548,
      "learning_rate": 9.543716779341241e-05,
      "loss": 1.0662,
      "step": 38120
    },
    {
      "epoch": 1.5449443892951926,
      "grad_norm": 0.628097414970398,
      "learning_rate": 9.539475273578729e-05,
      "loss": 1.0951,
      "step": 38130
    },
    {
      "epoch": 1.545349567472296,
      "grad_norm": 0.633327841758728,
      "learning_rate": 9.535233850840786e-05,
      "loss": 1.1315,
      "step": 38140
    },
    {
      "epoch": 1.5457547456493992,
      "grad_norm": 0.6506749987602234,
      "learning_rate": 9.530992511892068e-05,
      "loss": 1.0544,
      "step": 38150
    },
    {
      "epoch": 1.5461599238265027,
      "grad_norm": 0.6062018275260925,
      "learning_rate": 9.526751257497215e-05,
      "loss": 1.0972,
      "step": 38160
    },
    {
      "epoch": 1.5465651020036062,
      "grad_norm": 0.5924050211906433,
      "learning_rate": 9.522510088420851e-05,
      "loss": 1.0609,
      "step": 38170
    },
    {
      "epoch": 1.5469702801807095,
      "grad_norm": 0.7092388868331909,
      "learning_rate": 9.518269005427579e-05,
      "loss": 1.0822,
      "step": 38180
    },
    {
      "epoch": 1.5473754583578128,
      "grad_norm": 0.5658613443374634,
      "learning_rate": 9.514028009282002e-05,
      "loss": 1.0915,
      "step": 38190
    },
    {
      "epoch": 1.5477806365349163,
      "grad_norm": 0.5547269582748413,
      "learning_rate": 9.509787100748692e-05,
      "loss": 1.1108,
      "step": 38200
    },
    {
      "epoch": 1.5481858147120198,
      "grad_norm": 0.6566269993782043,
      "learning_rate": 9.505546280592214e-05,
      "loss": 1.094,
      "step": 38210
    },
    {
      "epoch": 1.5485909928891228,
      "grad_norm": 0.6539956331253052,
      "learning_rate": 9.501305549577111e-05,
      "loss": 1.0936,
      "step": 38220
    },
    {
      "epoch": 1.5489961710662263,
      "grad_norm": 0.574077844619751,
      "learning_rate": 9.497064908467909e-05,
      "loss": 1.0834,
      "step": 38230
    },
    {
      "epoch": 1.5494013492433298,
      "grad_norm": 0.5364807844161987,
      "learning_rate": 9.49282435802913e-05,
      "loss": 1.0639,
      "step": 38240
    },
    {
      "epoch": 1.5498065274204331,
      "grad_norm": 0.7028329372406006,
      "learning_rate": 9.488583899025268e-05,
      "loss": 1.1112,
      "step": 38250
    },
    {
      "epoch": 1.5502117055975364,
      "grad_norm": 0.7041266560554504,
      "learning_rate": 9.484343532220804e-05,
      "loss": 1.0918,
      "step": 38260
    },
    {
      "epoch": 1.55061688377464,
      "grad_norm": 0.6479509472846985,
      "learning_rate": 9.480103258380198e-05,
      "loss": 1.0966,
      "step": 38270
    },
    {
      "epoch": 1.5510220619517434,
      "grad_norm": 0.6425219178199768,
      "learning_rate": 9.475863078267904e-05,
      "loss": 1.0686,
      "step": 38280
    },
    {
      "epoch": 1.5514272401288467,
      "grad_norm": 0.6126973628997803,
      "learning_rate": 9.471622992648343e-05,
      "loss": 1.0866,
      "step": 38290
    },
    {
      "epoch": 1.55183241830595,
      "grad_norm": 0.6086312532424927,
      "learning_rate": 9.467383002285938e-05,
      "loss": 1.091,
      "step": 38300
    },
    {
      "epoch": 1.5522375964830535,
      "grad_norm": 0.6344898343086243,
      "learning_rate": 9.463143107945084e-05,
      "loss": 1.0964,
      "step": 38310
    },
    {
      "epoch": 1.5526427746601568,
      "grad_norm": 0.5694665312767029,
      "learning_rate": 9.458903310390156e-05,
      "loss": 1.0583,
      "step": 38320
    },
    {
      "epoch": 1.55304795283726,
      "grad_norm": 0.6014675498008728,
      "learning_rate": 9.454663610385516e-05,
      "loss": 1.0672,
      "step": 38330
    },
    {
      "epoch": 1.5534531310143636,
      "grad_norm": 0.5599460005760193,
      "learning_rate": 9.450424008695509e-05,
      "loss": 1.0552,
      "step": 38340
    },
    {
      "epoch": 1.553858309191467,
      "grad_norm": 0.6046348810195923,
      "learning_rate": 9.446184506084463e-05,
      "loss": 1.0777,
      "step": 38350
    },
    {
      "epoch": 1.5542634873685703,
      "grad_norm": 0.5129001140594482,
      "learning_rate": 9.441945103316683e-05,
      "loss": 1.1022,
      "step": 38360
    },
    {
      "epoch": 1.5546686655456736,
      "grad_norm": 0.640091598033905,
      "learning_rate": 9.437705801156464e-05,
      "loss": 1.0989,
      "step": 38370
    },
    {
      "epoch": 1.5550738437227771,
      "grad_norm": 0.7817403674125671,
      "learning_rate": 9.433466600368071e-05,
      "loss": 1.1541,
      "step": 38380
    },
    {
      "epoch": 1.5554790218998806,
      "grad_norm": 0.5818488597869873,
      "learning_rate": 9.429227501715765e-05,
      "loss": 1.121,
      "step": 38390
    },
    {
      "epoch": 1.5558842000769837,
      "grad_norm": 0.7617504596710205,
      "learning_rate": 9.424988505963776e-05,
      "loss": 1.081,
      "step": 38400
    },
    {
      "epoch": 1.5562893782540872,
      "grad_norm": 0.5552318692207336,
      "learning_rate": 9.420749613876325e-05,
      "loss": 1.0839,
      "step": 38410
    },
    {
      "epoch": 1.5566945564311907,
      "grad_norm": 0.6303594708442688,
      "learning_rate": 9.416510826217608e-05,
      "loss": 1.1144,
      "step": 38420
    },
    {
      "epoch": 1.557099734608294,
      "grad_norm": 0.6651632189750671,
      "learning_rate": 9.412272143751807e-05,
      "loss": 1.0936,
      "step": 38430
    },
    {
      "epoch": 1.5575049127853973,
      "grad_norm": 0.5950849652290344,
      "learning_rate": 9.40803356724308e-05,
      "loss": 1.0674,
      "step": 38440
    },
    {
      "epoch": 1.5579100909625008,
      "grad_norm": 0.559912383556366,
      "learning_rate": 9.40379509745557e-05,
      "loss": 1.1033,
      "step": 38450
    },
    {
      "epoch": 1.5583152691396043,
      "grad_norm": 0.7054072618484497,
      "learning_rate": 9.399556735153393e-05,
      "loss": 1.0789,
      "step": 38460
    },
    {
      "epoch": 1.5587204473167076,
      "grad_norm": 0.6914424300193787,
      "learning_rate": 9.395318481100661e-05,
      "loss": 1.059,
      "step": 38470
    },
    {
      "epoch": 1.5591256254938108,
      "grad_norm": 0.6692454814910889,
      "learning_rate": 9.391080336061454e-05,
      "loss": 1.1208,
      "step": 38480
    },
    {
      "epoch": 1.5595308036709143,
      "grad_norm": 0.6104021668434143,
      "learning_rate": 9.386842300799834e-05,
      "loss": 1.0917,
      "step": 38490
    },
    {
      "epoch": 1.5599359818480176,
      "grad_norm": 0.576413631439209,
      "learning_rate": 9.382604376079847e-05,
      "loss": 1.0939,
      "step": 38500
    },
    {
      "epoch": 1.560341160025121,
      "grad_norm": 0.5529304146766663,
      "learning_rate": 9.378366562665513e-05,
      "loss": 1.0348,
      "step": 38510
    },
    {
      "epoch": 1.5607463382022244,
      "grad_norm": 0.6399975419044495,
      "learning_rate": 9.374128861320836e-05,
      "loss": 1.1063,
      "step": 38520
    },
    {
      "epoch": 1.561151516379328,
      "grad_norm": 0.6199853420257568,
      "learning_rate": 9.369891272809805e-05,
      "loss": 1.1408,
      "step": 38530
    },
    {
      "epoch": 1.5615566945564312,
      "grad_norm": 0.5825318098068237,
      "learning_rate": 9.36565379789638e-05,
      "loss": 1.0669,
      "step": 38540
    },
    {
      "epoch": 1.5619618727335345,
      "grad_norm": 0.6446751952171326,
      "learning_rate": 9.361416437344503e-05,
      "loss": 1.1118,
      "step": 38550
    },
    {
      "epoch": 1.562367050910638,
      "grad_norm": 0.5801598429679871,
      "learning_rate": 9.357179191918099e-05,
      "loss": 1.1178,
      "step": 38560
    },
    {
      "epoch": 1.5627722290877415,
      "grad_norm": 0.8009796142578125,
      "learning_rate": 9.352942062381062e-05,
      "loss": 1.1216,
      "step": 38570
    },
    {
      "epoch": 1.5631774072648446,
      "grad_norm": 0.6592332720756531,
      "learning_rate": 9.348705049497282e-05,
      "loss": 1.0862,
      "step": 38580
    },
    {
      "epoch": 1.563582585441948,
      "grad_norm": 0.6327279806137085,
      "learning_rate": 9.344468154030615e-05,
      "loss": 1.132,
      "step": 38590
    },
    {
      "epoch": 1.5639877636190516,
      "grad_norm": 0.7034571766853333,
      "learning_rate": 9.340231376744896e-05,
      "loss": 1.0603,
      "step": 38600
    },
    {
      "epoch": 1.5643929417961548,
      "grad_norm": 0.5548397898674011,
      "learning_rate": 9.335994718403948e-05,
      "loss": 1.0836,
      "step": 38610
    },
    {
      "epoch": 1.5647981199732581,
      "grad_norm": 0.5942142605781555,
      "learning_rate": 9.331758179771561e-05,
      "loss": 1.1089,
      "step": 38620
    },
    {
      "epoch": 1.5652032981503616,
      "grad_norm": 0.6517189145088196,
      "learning_rate": 9.327521761611507e-05,
      "loss": 1.0949,
      "step": 38630
    },
    {
      "epoch": 1.5656084763274651,
      "grad_norm": 0.5941251516342163,
      "learning_rate": 9.323285464687548e-05,
      "loss": 1.0988,
      "step": 38640
    },
    {
      "epoch": 1.5660136545045684,
      "grad_norm": 0.6198177337646484,
      "learning_rate": 9.319049289763408e-05,
      "loss": 1.1075,
      "step": 38650
    },
    {
      "epoch": 1.5664188326816717,
      "grad_norm": 0.5610637664794922,
      "learning_rate": 9.314813237602796e-05,
      "loss": 1.0721,
      "step": 38660
    },
    {
      "epoch": 1.5668240108587752,
      "grad_norm": 0.607314944267273,
      "learning_rate": 9.310577308969396e-05,
      "loss": 1.0854,
      "step": 38670
    },
    {
      "epoch": 1.5672291890358785,
      "grad_norm": 0.5881009101867676,
      "learning_rate": 9.306341504626874e-05,
      "loss": 1.0685,
      "step": 38680
    },
    {
      "epoch": 1.5676343672129818,
      "grad_norm": 0.6872912645339966,
      "learning_rate": 9.302105825338876e-05,
      "loss": 1.0932,
      "step": 38690
    },
    {
      "epoch": 1.5680395453900853,
      "grad_norm": 0.6155185699462891,
      "learning_rate": 9.297870271869015e-05,
      "loss": 1.0966,
      "step": 38700
    },
    {
      "epoch": 1.5684447235671888,
      "grad_norm": 0.6043867468833923,
      "learning_rate": 9.29363484498089e-05,
      "loss": 1.0647,
      "step": 38710
    },
    {
      "epoch": 1.568849901744292,
      "grad_norm": 0.6379392743110657,
      "learning_rate": 9.289399545438073e-05,
      "loss": 1.0899,
      "step": 38720
    },
    {
      "epoch": 1.5692550799213953,
      "grad_norm": 0.6558389663696289,
      "learning_rate": 9.285164374004118e-05,
      "loss": 1.0834,
      "step": 38730
    },
    {
      "epoch": 1.5696602580984989,
      "grad_norm": 0.6114158630371094,
      "learning_rate": 9.280929331442547e-05,
      "loss": 1.1174,
      "step": 38740
    },
    {
      "epoch": 1.5700654362756024,
      "grad_norm": 0.5345343351364136,
      "learning_rate": 9.276694418516867e-05,
      "loss": 1.0363,
      "step": 38750
    },
    {
      "epoch": 1.5704706144527054,
      "grad_norm": 0.6120557188987732,
      "learning_rate": 9.272459635990562e-05,
      "loss": 1.0781,
      "step": 38760
    },
    {
      "epoch": 1.570875792629809,
      "grad_norm": 0.6571221351623535,
      "learning_rate": 9.268224984627086e-05,
      "loss": 1.0949,
      "step": 38770
    },
    {
      "epoch": 1.5712809708069124,
      "grad_norm": 0.7217812538146973,
      "learning_rate": 9.263990465189874e-05,
      "loss": 1.0738,
      "step": 38780
    },
    {
      "epoch": 1.5716861489840157,
      "grad_norm": 0.6045231223106384,
      "learning_rate": 9.259756078442335e-05,
      "loss": 1.0772,
      "step": 38790
    },
    {
      "epoch": 1.572091327161119,
      "grad_norm": 0.6246163845062256,
      "learning_rate": 9.255521825147852e-05,
      "loss": 1.1032,
      "step": 38800
    },
    {
      "epoch": 1.5724965053382225,
      "grad_norm": 0.5911117196083069,
      "learning_rate": 9.251287706069794e-05,
      "loss": 1.1264,
      "step": 38810
    },
    {
      "epoch": 1.572901683515326,
      "grad_norm": 0.5903120040893555,
      "learning_rate": 9.247053721971495e-05,
      "loss": 1.0965,
      "step": 38820
    },
    {
      "epoch": 1.5733068616924293,
      "grad_norm": 0.5445790886878967,
      "learning_rate": 9.242819873616268e-05,
      "loss": 1.0747,
      "step": 38830
    },
    {
      "epoch": 1.5737120398695326,
      "grad_norm": 0.6243996024131775,
      "learning_rate": 9.238586161767402e-05,
      "loss": 1.0856,
      "step": 38840
    },
    {
      "epoch": 1.574117218046636,
      "grad_norm": 0.6954445242881775,
      "learning_rate": 9.234352587188158e-05,
      "loss": 1.0894,
      "step": 38850
    },
    {
      "epoch": 1.5745223962237394,
      "grad_norm": 0.5146361589431763,
      "learning_rate": 9.230119150641784e-05,
      "loss": 1.0894,
      "step": 38860
    },
    {
      "epoch": 1.5749275744008426,
      "grad_norm": 0.607353925704956,
      "learning_rate": 9.225885852891491e-05,
      "loss": 1.0327,
      "step": 38870
    },
    {
      "epoch": 1.5753327525779461,
      "grad_norm": 0.6767659187316895,
      "learning_rate": 9.221652694700467e-05,
      "loss": 1.0867,
      "step": 38880
    },
    {
      "epoch": 1.5757379307550496,
      "grad_norm": 0.6017717123031616,
      "learning_rate": 9.217419676831876e-05,
      "loss": 1.1092,
      "step": 38890
    },
    {
      "epoch": 1.576143108932153,
      "grad_norm": 0.6909428238868713,
      "learning_rate": 9.213186800048861e-05,
      "loss": 1.1219,
      "step": 38900
    },
    {
      "epoch": 1.5765482871092562,
      "grad_norm": 0.5006800889968872,
      "learning_rate": 9.208954065114526e-05,
      "loss": 1.1009,
      "step": 38910
    },
    {
      "epoch": 1.5769534652863597,
      "grad_norm": 0.6178410649299622,
      "learning_rate": 9.204721472791974e-05,
      "loss": 1.0822,
      "step": 38920
    },
    {
      "epoch": 1.5773586434634632,
      "grad_norm": 0.5487187504768372,
      "learning_rate": 9.200489023844258e-05,
      "loss": 1.1225,
      "step": 38930
    },
    {
      "epoch": 1.5777638216405663,
      "grad_norm": 0.56340491771698,
      "learning_rate": 9.19625671903442e-05,
      "loss": 1.0734,
      "step": 38940
    },
    {
      "epoch": 1.5781689998176698,
      "grad_norm": 0.6653925776481628,
      "learning_rate": 9.192024559125465e-05,
      "loss": 1.1009,
      "step": 38950
    },
    {
      "epoch": 1.5785741779947733,
      "grad_norm": 0.6719014048576355,
      "learning_rate": 9.18779254488038e-05,
      "loss": 1.1026,
      "step": 38960
    },
    {
      "epoch": 1.5789793561718766,
      "grad_norm": 0.6308223009109497,
      "learning_rate": 9.183560677062119e-05,
      "loss": 1.1044,
      "step": 38970
    },
    {
      "epoch": 1.5793845343489799,
      "grad_norm": 0.6283626556396484,
      "learning_rate": 9.179328956433623e-05,
      "loss": 1.0906,
      "step": 38980
    },
    {
      "epoch": 1.5797897125260834,
      "grad_norm": 0.6139721274375916,
      "learning_rate": 9.175097383757794e-05,
      "loss": 1.1084,
      "step": 38990
    },
    {
      "epoch": 1.5801948907031869,
      "grad_norm": 0.6013203263282776,
      "learning_rate": 9.170865959797509e-05,
      "loss": 1.1144,
      "step": 39000
    },
    {
      "epoch": 1.5806000688802901,
      "grad_norm": 0.6077088117599487,
      "learning_rate": 9.166634685315617e-05,
      "loss": 1.1118,
      "step": 39010
    },
    {
      "epoch": 1.5810052470573934,
      "grad_norm": 0.6360870599746704,
      "learning_rate": 9.162403561074947e-05,
      "loss": 1.1135,
      "step": 39020
    },
    {
      "epoch": 1.581410425234497,
      "grad_norm": 0.5303450226783752,
      "learning_rate": 9.158172587838298e-05,
      "loss": 1.0911,
      "step": 39030
    },
    {
      "epoch": 1.5818156034116002,
      "grad_norm": 0.5624047517776489,
      "learning_rate": 9.153941766368439e-05,
      "loss": 1.1294,
      "step": 39040
    },
    {
      "epoch": 1.5822207815887035,
      "grad_norm": 0.6138178706169128,
      "learning_rate": 9.149711097428111e-05,
      "loss": 1.0732,
      "step": 39050
    },
    {
      "epoch": 1.582625959765807,
      "grad_norm": 0.5909848809242249,
      "learning_rate": 9.145480581780033e-05,
      "loss": 1.0592,
      "step": 39060
    },
    {
      "epoch": 1.5830311379429105,
      "grad_norm": 0.7190127372741699,
      "learning_rate": 9.141250220186892e-05,
      "loss": 1.0465,
      "step": 39070
    },
    {
      "epoch": 1.5834363161200138,
      "grad_norm": 0.6280826330184937,
      "learning_rate": 9.137020013411347e-05,
      "loss": 1.0798,
      "step": 39080
    },
    {
      "epoch": 1.583841494297117,
      "grad_norm": 0.5749551653862,
      "learning_rate": 9.132789962216031e-05,
      "loss": 1.0953,
      "step": 39090
    },
    {
      "epoch": 1.5842466724742206,
      "grad_norm": 0.6618676781654358,
      "learning_rate": 9.128560067363551e-05,
      "loss": 1.0693,
      "step": 39100
    },
    {
      "epoch": 1.584651850651324,
      "grad_norm": 0.5931825637817383,
      "learning_rate": 9.124330329616482e-05,
      "loss": 1.0823,
      "step": 39110
    },
    {
      "epoch": 1.5850570288284271,
      "grad_norm": 0.6341961026191711,
      "learning_rate": 9.12010074973737e-05,
      "loss": 1.0466,
      "step": 39120
    },
    {
      "epoch": 1.5854622070055306,
      "grad_norm": 0.6672579646110535,
      "learning_rate": 9.115871328488737e-05,
      "loss": 1.1192,
      "step": 39130
    },
    {
      "epoch": 1.5858673851826341,
      "grad_norm": 0.5738404393196106,
      "learning_rate": 9.111642066633067e-05,
      "loss": 1.0663,
      "step": 39140
    },
    {
      "epoch": 1.5862725633597374,
      "grad_norm": 0.5870886445045471,
      "learning_rate": 9.107412964932832e-05,
      "loss": 1.0829,
      "step": 39150
    },
    {
      "epoch": 1.5866777415368407,
      "grad_norm": 0.6038975119590759,
      "learning_rate": 9.103184024150462e-05,
      "loss": 1.0888,
      "step": 39160
    },
    {
      "epoch": 1.5870829197139442,
      "grad_norm": 0.639144778251648,
      "learning_rate": 9.098955245048359e-05,
      "loss": 1.1131,
      "step": 39170
    },
    {
      "epoch": 1.5874880978910477,
      "grad_norm": 0.6082796454429626,
      "learning_rate": 9.094726628388899e-05,
      "loss": 1.0886,
      "step": 39180
    },
    {
      "epoch": 1.587893276068151,
      "grad_norm": 0.6729987859725952,
      "learning_rate": 9.090498174934424e-05,
      "loss": 1.061,
      "step": 39190
    },
    {
      "epoch": 1.5882984542452543,
      "grad_norm": 0.6638504266738892,
      "learning_rate": 9.086269885447258e-05,
      "loss": 1.0737,
      "step": 39200
    },
    {
      "epoch": 1.5887036324223578,
      "grad_norm": 0.652356743812561,
      "learning_rate": 9.082041760689685e-05,
      "loss": 1.1125,
      "step": 39210
    },
    {
      "epoch": 1.589108810599461,
      "grad_norm": 0.594770610332489,
      "learning_rate": 9.07781380142396e-05,
      "loss": 1.0807,
      "step": 39220
    },
    {
      "epoch": 1.5895139887765644,
      "grad_norm": 0.5903393626213074,
      "learning_rate": 9.073586008412311e-05,
      "loss": 1.0845,
      "step": 39230
    },
    {
      "epoch": 1.5899191669536679,
      "grad_norm": 0.6025105118751526,
      "learning_rate": 9.069358382416937e-05,
      "loss": 1.0789,
      "step": 39240
    },
    {
      "epoch": 1.5903243451307714,
      "grad_norm": 0.7120552062988281,
      "learning_rate": 9.065130924199998e-05,
      "loss": 1.1148,
      "step": 39250
    },
    {
      "epoch": 1.5907295233078746,
      "grad_norm": 0.6624370217323303,
      "learning_rate": 9.060903634523643e-05,
      "loss": 1.0957,
      "step": 39260
    },
    {
      "epoch": 1.591134701484978,
      "grad_norm": 0.6254958510398865,
      "learning_rate": 9.05667651414997e-05,
      "loss": 1.0854,
      "step": 39270
    },
    {
      "epoch": 1.5915398796620814,
      "grad_norm": 0.622671902179718,
      "learning_rate": 9.05244956384106e-05,
      "loss": 1.0899,
      "step": 39280
    },
    {
      "epoch": 1.591945057839185,
      "grad_norm": 0.5928722620010376,
      "learning_rate": 9.048222784358954e-05,
      "loss": 1.0584,
      "step": 39290
    },
    {
      "epoch": 1.592350236016288,
      "grad_norm": 0.634277880191803,
      "learning_rate": 9.043996176465665e-05,
      "loss": 1.0883,
      "step": 39300
    },
    {
      "epoch": 1.5927554141933915,
      "grad_norm": 0.6084487438201904,
      "learning_rate": 9.039769740923183e-05,
      "loss": 1.0887,
      "step": 39310
    },
    {
      "epoch": 1.593160592370495,
      "grad_norm": 0.6041915416717529,
      "learning_rate": 9.035543478493458e-05,
      "loss": 1.0638,
      "step": 39320
    },
    {
      "epoch": 1.5935657705475983,
      "grad_norm": 0.5692616105079651,
      "learning_rate": 9.031317389938411e-05,
      "loss": 1.0733,
      "step": 39330
    },
    {
      "epoch": 1.5939709487247016,
      "grad_norm": 0.6208325028419495,
      "learning_rate": 9.027091476019928e-05,
      "loss": 1.0929,
      "step": 39340
    },
    {
      "epoch": 1.594376126901805,
      "grad_norm": 0.5855532884597778,
      "learning_rate": 9.022865737499876e-05,
      "loss": 1.0867,
      "step": 39350
    },
    {
      "epoch": 1.5947813050789086,
      "grad_norm": 0.6911576390266418,
      "learning_rate": 9.018640175140074e-05,
      "loss": 1.1172,
      "step": 39360
    },
    {
      "epoch": 1.5951864832560119,
      "grad_norm": 0.5716893672943115,
      "learning_rate": 9.014414789702324e-05,
      "loss": 1.1018,
      "step": 39370
    },
    {
      "epoch": 1.5955916614331151,
      "grad_norm": 0.6040748357772827,
      "learning_rate": 9.01018958194838e-05,
      "loss": 1.0374,
      "step": 39380
    },
    {
      "epoch": 1.5959968396102187,
      "grad_norm": 0.6257567405700684,
      "learning_rate": 9.005964552639984e-05,
      "loss": 1.082,
      "step": 39390
    },
    {
      "epoch": 1.596402017787322,
      "grad_norm": 0.6275560259819031,
      "learning_rate": 9.001739702538829e-05,
      "loss": 1.0978,
      "step": 39400
    },
    {
      "epoch": 1.5968071959644252,
      "grad_norm": 0.6203012466430664,
      "learning_rate": 8.997515032406582e-05,
      "loss": 1.0653,
      "step": 39410
    },
    {
      "epoch": 1.5972123741415287,
      "grad_norm": 0.6383571028709412,
      "learning_rate": 8.993290543004876e-05,
      "loss": 1.0878,
      "step": 39420
    },
    {
      "epoch": 1.5976175523186322,
      "grad_norm": 0.568600594997406,
      "learning_rate": 8.989066235095317e-05,
      "loss": 1.0439,
      "step": 39430
    },
    {
      "epoch": 1.5980227304957355,
      "grad_norm": 0.5528169274330139,
      "learning_rate": 8.984842109439472e-05,
      "loss": 1.1286,
      "step": 39440
    },
    {
      "epoch": 1.5984279086728388,
      "grad_norm": 0.6062717437744141,
      "learning_rate": 8.980618166798876e-05,
      "loss": 1.1052,
      "step": 39450
    },
    {
      "epoch": 1.5988330868499423,
      "grad_norm": 0.6953077912330627,
      "learning_rate": 8.976394407935034e-05,
      "loss": 1.0897,
      "step": 39460
    },
    {
      "epoch": 1.5992382650270458,
      "grad_norm": 0.601433277130127,
      "learning_rate": 8.97217083360941e-05,
      "loss": 1.0855,
      "step": 39470
    },
    {
      "epoch": 1.5996434432041489,
      "grad_norm": 0.6150017380714417,
      "learning_rate": 8.96794744458345e-05,
      "loss": 1.0862,
      "step": 39480
    },
    {
      "epoch": 1.6000486213812524,
      "grad_norm": 0.638403594493866,
      "learning_rate": 8.963724241618552e-05,
      "loss": 1.1081,
      "step": 39490
    },
    {
      "epoch": 1.6004537995583559,
      "grad_norm": 0.600587785243988,
      "learning_rate": 8.959501225476088e-05,
      "loss": 1.103,
      "step": 39500
    },
    {
      "epoch": 1.6008589777354592,
      "grad_norm": 0.5986045002937317,
      "learning_rate": 8.955278396917393e-05,
      "loss": 1.0321,
      "step": 39510
    },
    {
      "epoch": 1.6012641559125624,
      "grad_norm": 0.6670452356338501,
      "learning_rate": 8.95105575670377e-05,
      "loss": 1.0699,
      "step": 39520
    },
    {
      "epoch": 1.601669334089666,
      "grad_norm": 0.6330640316009521,
      "learning_rate": 8.94683330559648e-05,
      "loss": 1.0747,
      "step": 39530
    },
    {
      "epoch": 1.6020745122667694,
      "grad_norm": 0.5920957922935486,
      "learning_rate": 8.94261104435677e-05,
      "loss": 1.0955,
      "step": 39540
    },
    {
      "epoch": 1.6024796904438727,
      "grad_norm": 0.6082499623298645,
      "learning_rate": 8.938388973745832e-05,
      "loss": 1.0765,
      "step": 39550
    },
    {
      "epoch": 1.602884868620976,
      "grad_norm": 0.5481220483779907,
      "learning_rate": 8.934167094524835e-05,
      "loss": 1.0667,
      "step": 39560
    },
    {
      "epoch": 1.6032900467980795,
      "grad_norm": 0.6848596334457397,
      "learning_rate": 8.929945407454907e-05,
      "loss": 1.0934,
      "step": 39570
    },
    {
      "epoch": 1.6036952249751828,
      "grad_norm": 0.5654980540275574,
      "learning_rate": 8.925723913297146e-05,
      "loss": 1.0651,
      "step": 39580
    },
    {
      "epoch": 1.604100403152286,
      "grad_norm": 0.6066490411758423,
      "learning_rate": 8.92150261281261e-05,
      "loss": 1.1098,
      "step": 39590
    },
    {
      "epoch": 1.6045055813293896,
      "grad_norm": 0.6112165451049805,
      "learning_rate": 8.917281506762335e-05,
      "loss": 1.1016,
      "step": 39600
    },
    {
      "epoch": 1.604910759506493,
      "grad_norm": 0.576171338558197,
      "learning_rate": 8.913060595907305e-05,
      "loss": 1.063,
      "step": 39610
    },
    {
      "epoch": 1.6053159376835964,
      "grad_norm": 0.6219948530197144,
      "learning_rate": 8.90883988100848e-05,
      "loss": 1.0837,
      "step": 39620
    },
    {
      "epoch": 1.6057211158606997,
      "grad_norm": 0.6182501316070557,
      "learning_rate": 8.90461936282678e-05,
      "loss": 1.0624,
      "step": 39630
    },
    {
      "epoch": 1.6061262940378032,
      "grad_norm": 0.5886044502258301,
      "learning_rate": 8.900399042123086e-05,
      "loss": 1.1246,
      "step": 39640
    },
    {
      "epoch": 1.6065314722149067,
      "grad_norm": 0.6841354370117188,
      "learning_rate": 8.896178919658257e-05,
      "loss": 1.0821,
      "step": 39650
    },
    {
      "epoch": 1.6069366503920097,
      "grad_norm": 0.6246191263198853,
      "learning_rate": 8.891958996193105e-05,
      "loss": 1.1082,
      "step": 39660
    },
    {
      "epoch": 1.6073418285691132,
      "grad_norm": 0.6086409091949463,
      "learning_rate": 8.887739272488406e-05,
      "loss": 1.0933,
      "step": 39670
    },
    {
      "epoch": 1.6077470067462167,
      "grad_norm": 0.6315511465072632,
      "learning_rate": 8.8835197493049e-05,
      "loss": 1.1012,
      "step": 39680
    },
    {
      "epoch": 1.60815218492332,
      "grad_norm": 0.5941274762153625,
      "learning_rate": 8.8793004274033e-05,
      "loss": 1.0745,
      "step": 39690
    },
    {
      "epoch": 1.6085573631004233,
      "grad_norm": 0.699793815612793,
      "learning_rate": 8.875081307544272e-05,
      "loss": 1.1045,
      "step": 39700
    },
    {
      "epoch": 1.6089625412775268,
      "grad_norm": 0.6342272162437439,
      "learning_rate": 8.870862390488452e-05,
      "loss": 1.0683,
      "step": 39710
    },
    {
      "epoch": 1.6093677194546303,
      "grad_norm": 0.6315333247184753,
      "learning_rate": 8.866643676996435e-05,
      "loss": 1.0802,
      "step": 39720
    },
    {
      "epoch": 1.6097728976317336,
      "grad_norm": 0.6373916864395142,
      "learning_rate": 8.862425167828781e-05,
      "loss": 1.0809,
      "step": 39730
    },
    {
      "epoch": 1.6101780758088369,
      "grad_norm": 0.7128381133079529,
      "learning_rate": 8.858206863746018e-05,
      "loss": 1.0713,
      "step": 39740
    },
    {
      "epoch": 1.6105832539859404,
      "grad_norm": 0.5565304160118103,
      "learning_rate": 8.853988765508625e-05,
      "loss": 1.1112,
      "step": 39750
    },
    {
      "epoch": 1.6109884321630437,
      "grad_norm": 0.625450074672699,
      "learning_rate": 8.849770873877057e-05,
      "loss": 1.1171,
      "step": 39760
    },
    {
      "epoch": 1.611393610340147,
      "grad_norm": 0.6005278825759888,
      "learning_rate": 8.845553189611727e-05,
      "loss": 1.1131,
      "step": 39770
    },
    {
      "epoch": 1.6117987885172504,
      "grad_norm": 0.6605473160743713,
      "learning_rate": 8.841335713473008e-05,
      "loss": 1.066,
      "step": 39780
    },
    {
      "epoch": 1.612203966694354,
      "grad_norm": 0.6042067408561707,
      "learning_rate": 8.837118446221238e-05,
      "loss": 1.063,
      "step": 39790
    },
    {
      "epoch": 1.6126091448714572,
      "grad_norm": 0.5892220139503479,
      "learning_rate": 8.832901388616715e-05,
      "loss": 1.0631,
      "step": 39800
    },
    {
      "epoch": 1.6130143230485605,
      "grad_norm": 0.612393319606781,
      "learning_rate": 8.828684541419696e-05,
      "loss": 1.0692,
      "step": 39810
    },
    {
      "epoch": 1.613419501225664,
      "grad_norm": 0.7139731049537659,
      "learning_rate": 8.824467905390418e-05,
      "loss": 1.1044,
      "step": 39820
    },
    {
      "epoch": 1.6138246794027675,
      "grad_norm": 0.6660493612289429,
      "learning_rate": 8.820251481289058e-05,
      "loss": 1.0633,
      "step": 39830
    },
    {
      "epoch": 1.6142298575798706,
      "grad_norm": 0.6602833867073059,
      "learning_rate": 8.816035269875765e-05,
      "loss": 1.0844,
      "step": 39840
    },
    {
      "epoch": 1.614635035756974,
      "grad_norm": 0.6375397443771362,
      "learning_rate": 8.81181927191065e-05,
      "loss": 1.0889,
      "step": 39850
    },
    {
      "epoch": 1.6150402139340776,
      "grad_norm": 0.6376544237136841,
      "learning_rate": 8.807603488153782e-05,
      "loss": 1.1296,
      "step": 39860
    },
    {
      "epoch": 1.6154453921111809,
      "grad_norm": 0.5930672287940979,
      "learning_rate": 8.80338791936519e-05,
      "loss": 1.0774,
      "step": 39870
    },
    {
      "epoch": 1.6158505702882842,
      "grad_norm": 0.6160508394241333,
      "learning_rate": 8.799172566304874e-05,
      "loss": 1.11,
      "step": 39880
    },
    {
      "epoch": 1.6162557484653877,
      "grad_norm": 0.5299475193023682,
      "learning_rate": 8.794957429732787e-05,
      "loss": 1.0685,
      "step": 39890
    },
    {
      "epoch": 1.6166609266424912,
      "grad_norm": 0.5838728547096252,
      "learning_rate": 8.790742510408844e-05,
      "loss": 1.1137,
      "step": 39900
    },
    {
      "epoch": 1.6170661048195945,
      "grad_norm": 0.595874547958374,
      "learning_rate": 8.786527809092919e-05,
      "loss": 1.0752,
      "step": 39910
    },
    {
      "epoch": 1.6174712829966977,
      "grad_norm": 0.6063977479934692,
      "learning_rate": 8.782313326544847e-05,
      "loss": 1.0404,
      "step": 39920
    },
    {
      "epoch": 1.6178764611738012,
      "grad_norm": 0.5461052656173706,
      "learning_rate": 8.778099063524436e-05,
      "loss": 1.0973,
      "step": 39930
    },
    {
      "epoch": 1.6182816393509045,
      "grad_norm": 0.5443772077560425,
      "learning_rate": 8.773885020791436e-05,
      "loss": 1.0737,
      "step": 39940
    },
    {
      "epoch": 1.6186868175280078,
      "grad_norm": 0.6033807396888733,
      "learning_rate": 8.769671199105565e-05,
      "loss": 1.0399,
      "step": 39950
    },
    {
      "epoch": 1.6190919957051113,
      "grad_norm": 0.5636721253395081,
      "learning_rate": 8.765457599226506e-05,
      "loss": 1.0939,
      "step": 39960
    },
    {
      "epoch": 1.6194971738822148,
      "grad_norm": 0.6716282963752747,
      "learning_rate": 8.761244221913894e-05,
      "loss": 1.0858,
      "step": 39970
    },
    {
      "epoch": 1.619902352059318,
      "grad_norm": 0.6178957223892212,
      "learning_rate": 8.757031067927325e-05,
      "loss": 1.053,
      "step": 39980
    },
    {
      "epoch": 1.6203075302364214,
      "grad_norm": 0.6940179467201233,
      "learning_rate": 8.752818138026364e-05,
      "loss": 1.1088,
      "step": 39990
    },
    {
      "epoch": 1.6207127084135249,
      "grad_norm": 0.6220199465751648,
      "learning_rate": 8.748605432970525e-05,
      "loss": 1.0796,
      "step": 40000
    },
    {
      "epoch": 1.6211178865906284,
      "grad_norm": 0.5971586108207703,
      "learning_rate": 8.744392953519286e-05,
      "loss": 1.088,
      "step": 40010
    },
    {
      "epoch": 1.6215230647677314,
      "grad_norm": 0.5990147590637207,
      "learning_rate": 8.74018070043208e-05,
      "loss": 1.0969,
      "step": 40020
    },
    {
      "epoch": 1.621928242944835,
      "grad_norm": 0.641177237033844,
      "learning_rate": 8.73596867446831e-05,
      "loss": 1.1163,
      "step": 40030
    },
    {
      "epoch": 1.6223334211219385,
      "grad_norm": 0.5541775226593018,
      "learning_rate": 8.731756876387322e-05,
      "loss": 1.1409,
      "step": 40040
    },
    {
      "epoch": 1.6227385992990417,
      "grad_norm": 0.5820834040641785,
      "learning_rate": 8.727545306948436e-05,
      "loss": 1.1161,
      "step": 40050
    },
    {
      "epoch": 1.623143777476145,
      "grad_norm": 0.6210418939590454,
      "learning_rate": 8.72333396691092e-05,
      "loss": 1.0826,
      "step": 40060
    },
    {
      "epoch": 1.6235489556532485,
      "grad_norm": 0.6438102126121521,
      "learning_rate": 8.719122857034012e-05,
      "loss": 1.0932,
      "step": 40070
    },
    {
      "epoch": 1.623954133830352,
      "grad_norm": 0.6240463852882385,
      "learning_rate": 8.714911978076896e-05,
      "loss": 1.1364,
      "step": 40080
    },
    {
      "epoch": 1.6243593120074553,
      "grad_norm": 0.6653884053230286,
      "learning_rate": 8.710701330798719e-05,
      "loss": 1.0815,
      "step": 40090
    },
    {
      "epoch": 1.6247644901845586,
      "grad_norm": 0.5975614786148071,
      "learning_rate": 8.706490915958589e-05,
      "loss": 1.107,
      "step": 40100
    },
    {
      "epoch": 1.625169668361662,
      "grad_norm": 0.6499710083007812,
      "learning_rate": 8.702280734315574e-05,
      "loss": 1.0986,
      "step": 40110
    },
    {
      "epoch": 1.6255748465387654,
      "grad_norm": 0.6058189272880554,
      "learning_rate": 8.698070786628693e-05,
      "loss": 1.093,
      "step": 40120
    },
    {
      "epoch": 1.6259800247158687,
      "grad_norm": 0.608938455581665,
      "learning_rate": 8.693861073656925e-05,
      "loss": 1.0989,
      "step": 40130
    },
    {
      "epoch": 1.6263852028929722,
      "grad_norm": 0.5834254026412964,
      "learning_rate": 8.689651596159211e-05,
      "loss": 1.0662,
      "step": 40140
    },
    {
      "epoch": 1.6267903810700757,
      "grad_norm": 0.6277144551277161,
      "learning_rate": 8.685442354894439e-05,
      "loss": 1.0827,
      "step": 40150
    },
    {
      "epoch": 1.627195559247179,
      "grad_norm": 0.607833743095398,
      "learning_rate": 8.68123335062147e-05,
      "loss": 1.0721,
      "step": 40160
    },
    {
      "epoch": 1.6276007374242822,
      "grad_norm": 0.6204822063446045,
      "learning_rate": 8.677024584099113e-05,
      "loss": 1.0868,
      "step": 40170
    },
    {
      "epoch": 1.6280059156013857,
      "grad_norm": 0.6538118124008179,
      "learning_rate": 8.672816056086132e-05,
      "loss": 1.0776,
      "step": 40180
    },
    {
      "epoch": 1.6284110937784892,
      "grad_norm": 0.6493998765945435,
      "learning_rate": 8.668607767341252e-05,
      "loss": 1.0947,
      "step": 40190
    },
    {
      "epoch": 1.6288162719555923,
      "grad_norm": 0.6866775751113892,
      "learning_rate": 8.664399718623155e-05,
      "loss": 1.0974,
      "step": 40200
    },
    {
      "epoch": 1.6292214501326958,
      "grad_norm": 0.5597275495529175,
      "learning_rate": 8.660191910690473e-05,
      "loss": 1.0312,
      "step": 40210
    },
    {
      "epoch": 1.6296266283097993,
      "grad_norm": 0.6652094125747681,
      "learning_rate": 8.65598434430181e-05,
      "loss": 1.041,
      "step": 40220
    },
    {
      "epoch": 1.6300318064869026,
      "grad_norm": 0.5362501740455627,
      "learning_rate": 8.651777020215712e-05,
      "loss": 1.0589,
      "step": 40230
    },
    {
      "epoch": 1.6304369846640059,
      "grad_norm": 0.5634346604347229,
      "learning_rate": 8.647569939190687e-05,
      "loss": 1.0603,
      "step": 40240
    },
    {
      "epoch": 1.6308421628411094,
      "grad_norm": 0.6009073257446289,
      "learning_rate": 8.643363101985196e-05,
      "loss": 1.1112,
      "step": 40250
    },
    {
      "epoch": 1.631247341018213,
      "grad_norm": 0.613074541091919,
      "learning_rate": 8.639156509357658e-05,
      "loss": 1.0642,
      "step": 40260
    },
    {
      "epoch": 1.6316525191953162,
      "grad_norm": 0.539750337600708,
      "learning_rate": 8.634950162066452e-05,
      "loss": 1.0736,
      "step": 40270
    },
    {
      "epoch": 1.6320576973724195,
      "grad_norm": 0.6343671679496765,
      "learning_rate": 8.630744060869909e-05,
      "loss": 1.0525,
      "step": 40280
    },
    {
      "epoch": 1.632462875549523,
      "grad_norm": 0.5611103773117065,
      "learning_rate": 8.626538206526311e-05,
      "loss": 1.087,
      "step": 40290
    },
    {
      "epoch": 1.6328680537266262,
      "grad_norm": 0.5328253507614136,
      "learning_rate": 8.622332599793906e-05,
      "loss": 1.0904,
      "step": 40300
    },
    {
      "epoch": 1.6332732319037295,
      "grad_norm": 0.5955448150634766,
      "learning_rate": 8.618127241430885e-05,
      "loss": 1.0465,
      "step": 40310
    },
    {
      "epoch": 1.633678410080833,
      "grad_norm": 0.5576790571212769,
      "learning_rate": 8.613922132195402e-05,
      "loss": 1.0619,
      "step": 40320
    },
    {
      "epoch": 1.6340835882579365,
      "grad_norm": 0.6001816987991333,
      "learning_rate": 8.609717272845571e-05,
      "loss": 1.0977,
      "step": 40330
    },
    {
      "epoch": 1.6344887664350398,
      "grad_norm": 0.5778788924217224,
      "learning_rate": 8.60551266413945e-05,
      "loss": 1.0618,
      "step": 40340
    },
    {
      "epoch": 1.634893944612143,
      "grad_norm": 0.5315482020378113,
      "learning_rate": 8.601308306835057e-05,
      "loss": 1.0657,
      "step": 40350
    },
    {
      "epoch": 1.6352991227892466,
      "grad_norm": 0.5891866087913513,
      "learning_rate": 8.597104201690364e-05,
      "loss": 1.0637,
      "step": 40360
    },
    {
      "epoch": 1.63570430096635,
      "grad_norm": 0.6044672727584839,
      "learning_rate": 8.592900349463297e-05,
      "loss": 1.0655,
      "step": 40370
    },
    {
      "epoch": 1.6361094791434532,
      "grad_norm": 0.6423035264015198,
      "learning_rate": 8.58869675091174e-05,
      "loss": 1.0542,
      "step": 40380
    },
    {
      "epoch": 1.6365146573205567,
      "grad_norm": 0.6340887546539307,
      "learning_rate": 8.584493406793526e-05,
      "loss": 1.0744,
      "step": 40390
    },
    {
      "epoch": 1.6369198354976602,
      "grad_norm": 0.5769607424736023,
      "learning_rate": 8.580290317866446e-05,
      "loss": 1.1086,
      "step": 40400
    },
    {
      "epoch": 1.6373250136747635,
      "grad_norm": 0.6048576831817627,
      "learning_rate": 8.576087484888244e-05,
      "loss": 1.0681,
      "step": 40410
    },
    {
      "epoch": 1.6377301918518667,
      "grad_norm": 0.6626694798469543,
      "learning_rate": 8.571884908616616e-05,
      "loss": 1.0701,
      "step": 40420
    },
    {
      "epoch": 1.6381353700289703,
      "grad_norm": 0.6239895224571228,
      "learning_rate": 8.56768258980921e-05,
      "loss": 1.0793,
      "step": 40430
    },
    {
      "epoch": 1.6385405482060738,
      "grad_norm": 0.6267605423927307,
      "learning_rate": 8.563480529223638e-05,
      "loss": 1.0931,
      "step": 40440
    },
    {
      "epoch": 1.638945726383177,
      "grad_norm": 0.5992596745491028,
      "learning_rate": 8.559278727617455e-05,
      "loss": 1.11,
      "step": 40450
    },
    {
      "epoch": 1.6393509045602803,
      "grad_norm": 0.6366693377494812,
      "learning_rate": 8.55507718574817e-05,
      "loss": 1.0451,
      "step": 40460
    },
    {
      "epoch": 1.6397560827373838,
      "grad_norm": 0.601140558719635,
      "learning_rate": 8.550875904373253e-05,
      "loss": 1.1154,
      "step": 40470
    },
    {
      "epoch": 1.640161260914487,
      "grad_norm": 0.6160435080528259,
      "learning_rate": 8.546674884250115e-05,
      "loss": 1.0891,
      "step": 40480
    },
    {
      "epoch": 1.6405664390915904,
      "grad_norm": 0.5704943537712097,
      "learning_rate": 8.542474126136128e-05,
      "loss": 1.105,
      "step": 40490
    },
    {
      "epoch": 1.640971617268694,
      "grad_norm": 0.5755941271781921,
      "learning_rate": 8.538273630788621e-05,
      "loss": 1.0944,
      "step": 40500
    },
    {
      "epoch": 1.6413767954457974,
      "grad_norm": 0.5969797968864441,
      "learning_rate": 8.534073398964866e-05,
      "loss": 1.0637,
      "step": 40510
    },
    {
      "epoch": 1.6417819736229007,
      "grad_norm": 0.5483152866363525,
      "learning_rate": 8.529873431422092e-05,
      "loss": 1.0982,
      "step": 40520
    },
    {
      "epoch": 1.642187151800004,
      "grad_norm": 0.6910666823387146,
      "learning_rate": 8.525673728917479e-05,
      "loss": 1.0891,
      "step": 40530
    },
    {
      "epoch": 1.6425923299771075,
      "grad_norm": 0.6160481572151184,
      "learning_rate": 8.521474292208157e-05,
      "loss": 1.1246,
      "step": 40540
    },
    {
      "epoch": 1.642997508154211,
      "grad_norm": 0.6011239290237427,
      "learning_rate": 8.51727512205122e-05,
      "loss": 1.0945,
      "step": 40550
    },
    {
      "epoch": 1.643402686331314,
      "grad_norm": 0.6205071210861206,
      "learning_rate": 8.513076219203699e-05,
      "loss": 1.1133,
      "step": 40560
    },
    {
      "epoch": 1.6438078645084175,
      "grad_norm": 0.5777871608734131,
      "learning_rate": 8.508877584422584e-05,
      "loss": 1.1281,
      "step": 40570
    },
    {
      "epoch": 1.644213042685521,
      "grad_norm": 0.5639024376869202,
      "learning_rate": 8.504679218464816e-05,
      "loss": 1.0627,
      "step": 40580
    },
    {
      "epoch": 1.6446182208626243,
      "grad_norm": 0.615260124206543,
      "learning_rate": 8.500481122087285e-05,
      "loss": 1.1111,
      "step": 40590
    },
    {
      "epoch": 1.6450233990397276,
      "grad_norm": 0.5976471304893494,
      "learning_rate": 8.496283296046834e-05,
      "loss": 1.0611,
      "step": 40600
    },
    {
      "epoch": 1.6454285772168311,
      "grad_norm": 0.6440523266792297,
      "learning_rate": 8.492085741100262e-05,
      "loss": 1.0641,
      "step": 40610
    },
    {
      "epoch": 1.6458337553939346,
      "grad_norm": 0.5765237808227539,
      "learning_rate": 8.487888458004315e-05,
      "loss": 1.1165,
      "step": 40620
    },
    {
      "epoch": 1.646238933571038,
      "grad_norm": 0.6476341485977173,
      "learning_rate": 8.483691447515685e-05,
      "loss": 1.1093,
      "step": 40630
    },
    {
      "epoch": 1.6466441117481412,
      "grad_norm": 0.6461557149887085,
      "learning_rate": 8.479494710391025e-05,
      "loss": 1.0403,
      "step": 40640
    },
    {
      "epoch": 1.6470492899252447,
      "grad_norm": 0.629202663898468,
      "learning_rate": 8.475298247386927e-05,
      "loss": 1.0787,
      "step": 40650
    },
    {
      "epoch": 1.647454468102348,
      "grad_norm": 0.639281690120697,
      "learning_rate": 8.471102059259944e-05,
      "loss": 1.084,
      "step": 40660
    },
    {
      "epoch": 1.6478596462794513,
      "grad_norm": 0.5566146969795227,
      "learning_rate": 8.466906146766578e-05,
      "loss": 1.0563,
      "step": 40670
    },
    {
      "epoch": 1.6482648244565548,
      "grad_norm": 0.6295660138130188,
      "learning_rate": 8.462710510663275e-05,
      "loss": 1.0882,
      "step": 40680
    },
    {
      "epoch": 1.6486700026336583,
      "grad_norm": 0.5749549865722656,
      "learning_rate": 8.458515151706435e-05,
      "loss": 1.0744,
      "step": 40690
    },
    {
      "epoch": 1.6490751808107615,
      "grad_norm": 0.6582788825035095,
      "learning_rate": 8.454320070652411e-05,
      "loss": 1.0664,
      "step": 40700
    },
    {
      "epoch": 1.6494803589878648,
      "grad_norm": 0.6075201034545898,
      "learning_rate": 8.450125268257499e-05,
      "loss": 1.0789,
      "step": 40710
    },
    {
      "epoch": 1.6498855371649683,
      "grad_norm": 0.627381443977356,
      "learning_rate": 8.445930745277953e-05,
      "loss": 1.1022,
      "step": 40720
    },
    {
      "epoch": 1.6502907153420718,
      "grad_norm": 0.5821949243545532,
      "learning_rate": 8.441736502469966e-05,
      "loss": 1.1184,
      "step": 40730
    },
    {
      "epoch": 1.650695893519175,
      "grad_norm": 0.6098766922950745,
      "learning_rate": 8.437542540589694e-05,
      "loss": 1.1152,
      "step": 40740
    },
    {
      "epoch": 1.6511010716962784,
      "grad_norm": 0.6266768574714661,
      "learning_rate": 8.433348860393232e-05,
      "loss": 1.1099,
      "step": 40750
    },
    {
      "epoch": 1.651506249873382,
      "grad_norm": 0.5564292669296265,
      "learning_rate": 8.429155462636626e-05,
      "loss": 1.0749,
      "step": 40760
    },
    {
      "epoch": 1.6519114280504852,
      "grad_norm": 0.5888624787330627,
      "learning_rate": 8.42496234807587e-05,
      "loss": 1.0794,
      "step": 40770
    },
    {
      "epoch": 1.6523166062275885,
      "grad_norm": 0.5652872323989868,
      "learning_rate": 8.420769517466919e-05,
      "loss": 1.0849,
      "step": 40780
    },
    {
      "epoch": 1.652721784404692,
      "grad_norm": 0.599294900894165,
      "learning_rate": 8.41657697156566e-05,
      "loss": 1.1186,
      "step": 40790
    },
    {
      "epoch": 1.6531269625817955,
      "grad_norm": 0.6153682470321655,
      "learning_rate": 8.412384711127935e-05,
      "loss": 1.0675,
      "step": 40800
    },
    {
      "epoch": 1.6535321407588988,
      "grad_norm": 0.5572031140327454,
      "learning_rate": 8.40819273690954e-05,
      "loss": 1.0841,
      "step": 40810
    },
    {
      "epoch": 1.653937318936002,
      "grad_norm": 0.6122613549232483,
      "learning_rate": 8.404001049666211e-05,
      "loss": 1.0693,
      "step": 40820
    },
    {
      "epoch": 1.6543424971131055,
      "grad_norm": 0.6667152643203735,
      "learning_rate": 8.399809650153633e-05,
      "loss": 1.0945,
      "step": 40830
    },
    {
      "epoch": 1.6547476752902088,
      "grad_norm": 0.5740948915481567,
      "learning_rate": 8.39561853912745e-05,
      "loss": 1.1109,
      "step": 40840
    },
    {
      "epoch": 1.6551528534673121,
      "grad_norm": 0.5862547159194946,
      "learning_rate": 8.391427717343244e-05,
      "loss": 1.0747,
      "step": 40850
    },
    {
      "epoch": 1.6555580316444156,
      "grad_norm": 0.5960913896560669,
      "learning_rate": 8.387237185556545e-05,
      "loss": 1.0529,
      "step": 40860
    },
    {
      "epoch": 1.6559632098215191,
      "grad_norm": 0.5823866724967957,
      "learning_rate": 8.383046944522833e-05,
      "loss": 1.099,
      "step": 40870
    },
    {
      "epoch": 1.6563683879986224,
      "grad_norm": 0.5843397378921509,
      "learning_rate": 8.378856994997531e-05,
      "loss": 1.0329,
      "step": 40880
    },
    {
      "epoch": 1.6567735661757257,
      "grad_norm": 0.58575838804245,
      "learning_rate": 8.374667337736023e-05,
      "loss": 1.0952,
      "step": 40890
    },
    {
      "epoch": 1.6571787443528292,
      "grad_norm": 0.6204386949539185,
      "learning_rate": 8.370477973493627e-05,
      "loss": 1.0551,
      "step": 40900
    },
    {
      "epoch": 1.6575839225299327,
      "grad_norm": 0.6167457699775696,
      "learning_rate": 8.366288903025612e-05,
      "loss": 1.0602,
      "step": 40910
    },
    {
      "epoch": 1.6579891007070358,
      "grad_norm": 0.5577325224876404,
      "learning_rate": 8.362100127087193e-05,
      "loss": 1.0894,
      "step": 40920
    },
    {
      "epoch": 1.6583942788841393,
      "grad_norm": 0.5810073614120483,
      "learning_rate": 8.357911646433535e-05,
      "loss": 1.0887,
      "step": 40930
    },
    {
      "epoch": 1.6587994570612428,
      "grad_norm": 0.6831967830657959,
      "learning_rate": 8.353723461819743e-05,
      "loss": 1.0752,
      "step": 40940
    },
    {
      "epoch": 1.659204635238346,
      "grad_norm": 0.716757595539093,
      "learning_rate": 8.349535574000881e-05,
      "loss": 1.0902,
      "step": 40950
    },
    {
      "epoch": 1.6596098134154493,
      "grad_norm": 0.6159638166427612,
      "learning_rate": 8.34534798373195e-05,
      "loss": 1.0913,
      "step": 40960
    },
    {
      "epoch": 1.6600149915925528,
      "grad_norm": 0.5965061783790588,
      "learning_rate": 8.341160691767897e-05,
      "loss": 1.1025,
      "step": 40970
    },
    {
      "epoch": 1.6604201697696563,
      "grad_norm": 0.6997594237327576,
      "learning_rate": 8.336973698863618e-05,
      "loss": 1.1032,
      "step": 40980
    },
    {
      "epoch": 1.6608253479467596,
      "grad_norm": 0.5979862809181213,
      "learning_rate": 8.332787005773956e-05,
      "loss": 1.0722,
      "step": 40990
    },
    {
      "epoch": 1.661230526123863,
      "grad_norm": 0.5905948281288147,
      "learning_rate": 8.3286006132537e-05,
      "loss": 1.0762,
      "step": 41000
    },
    {
      "epoch": 1.6616357043009664,
      "grad_norm": 0.708157479763031,
      "learning_rate": 8.324414522057582e-05,
      "loss": 1.081,
      "step": 41010
    },
    {
      "epoch": 1.6620408824780697,
      "grad_norm": 0.5689798593521118,
      "learning_rate": 8.32022873294028e-05,
      "loss": 1.0807,
      "step": 41020
    },
    {
      "epoch": 1.662446060655173,
      "grad_norm": 0.5819019675254822,
      "learning_rate": 8.316043246656418e-05,
      "loss": 1.133,
      "step": 41030
    },
    {
      "epoch": 1.6628512388322765,
      "grad_norm": 0.6250677108764648,
      "learning_rate": 8.311858063960572e-05,
      "loss": 1.0725,
      "step": 41040
    },
    {
      "epoch": 1.66325641700938,
      "grad_norm": 0.6704633831977844,
      "learning_rate": 8.30767318560725e-05,
      "loss": 1.0842,
      "step": 41050
    },
    {
      "epoch": 1.6636615951864833,
      "grad_norm": 0.7360610961914062,
      "learning_rate": 8.303488612350917e-05,
      "loss": 1.0858,
      "step": 41060
    },
    {
      "epoch": 1.6640667733635865,
      "grad_norm": 0.6989070177078247,
      "learning_rate": 8.299304344945977e-05,
      "loss": 1.0622,
      "step": 41070
    },
    {
      "epoch": 1.66447195154069,
      "grad_norm": 0.5714001059532166,
      "learning_rate": 8.295120384146782e-05,
      "loss": 1.0715,
      "step": 41080
    },
    {
      "epoch": 1.6648771297177936,
      "grad_norm": 0.6496096849441528,
      "learning_rate": 8.290936730707626e-05,
      "loss": 1.0636,
      "step": 41090
    },
    {
      "epoch": 1.6652823078948966,
      "grad_norm": 0.5581532716751099,
      "learning_rate": 8.286753385382748e-05,
      "loss": 1.0882,
      "step": 41100
    },
    {
      "epoch": 1.6656874860720001,
      "grad_norm": 0.5957270264625549,
      "learning_rate": 8.282570348926328e-05,
      "loss": 1.0551,
      "step": 41110
    },
    {
      "epoch": 1.6660926642491036,
      "grad_norm": 0.6788472533226013,
      "learning_rate": 8.278387622092505e-05,
      "loss": 1.0614,
      "step": 41120
    },
    {
      "epoch": 1.666497842426207,
      "grad_norm": 0.5656567811965942,
      "learning_rate": 8.274205205635345e-05,
      "loss": 1.067,
      "step": 41130
    },
    {
      "epoch": 1.6669030206033102,
      "grad_norm": 0.6129136681556702,
      "learning_rate": 8.270023100308865e-05,
      "loss": 1.1244,
      "step": 41140
    },
    {
      "epoch": 1.6673081987804137,
      "grad_norm": 0.6843184232711792,
      "learning_rate": 8.265841306867027e-05,
      "loss": 1.1035,
      "step": 41150
    },
    {
      "epoch": 1.6677133769575172,
      "grad_norm": 0.6451422572135925,
      "learning_rate": 8.26165982606373e-05,
      "loss": 1.1382,
      "step": 41160
    },
    {
      "epoch": 1.6681185551346205,
      "grad_norm": 0.5695357918739319,
      "learning_rate": 8.25747865865283e-05,
      "loss": 1.0622,
      "step": 41170
    },
    {
      "epoch": 1.6685237333117238,
      "grad_norm": 0.6317389011383057,
      "learning_rate": 8.253297805388116e-05,
      "loss": 1.0986,
      "step": 41180
    },
    {
      "epoch": 1.6689289114888273,
      "grad_norm": 0.5949646830558777,
      "learning_rate": 8.249117267023323e-05,
      "loss": 1.0798,
      "step": 41190
    },
    {
      "epoch": 1.6693340896659306,
      "grad_norm": 0.5747407078742981,
      "learning_rate": 8.244937044312127e-05,
      "loss": 1.1061,
      "step": 41200
    },
    {
      "epoch": 1.6697392678430338,
      "grad_norm": 0.567299485206604,
      "learning_rate": 8.240757138008149e-05,
      "loss": 1.0648,
      "step": 41210
    },
    {
      "epoch": 1.6701444460201373,
      "grad_norm": 0.6568788290023804,
      "learning_rate": 8.23657754886495e-05,
      "loss": 1.0683,
      "step": 41220
    },
    {
      "epoch": 1.6705496241972408,
      "grad_norm": 0.5857599377632141,
      "learning_rate": 8.232398277636046e-05,
      "loss": 1.0663,
      "step": 41230
    },
    {
      "epoch": 1.6709548023743441,
      "grad_norm": 0.5675964951515198,
      "learning_rate": 8.228219325074883e-05,
      "loss": 1.1187,
      "step": 41240
    },
    {
      "epoch": 1.6713599805514474,
      "grad_norm": 0.6106472611427307,
      "learning_rate": 8.224040691934852e-05,
      "loss": 1.1318,
      "step": 41250
    },
    {
      "epoch": 1.671765158728551,
      "grad_norm": 0.6154898405075073,
      "learning_rate": 8.219862378969287e-05,
      "loss": 1.0875,
      "step": 41260
    },
    {
      "epoch": 1.6721703369056544,
      "grad_norm": 0.5511752367019653,
      "learning_rate": 8.215684386931465e-05,
      "loss": 1.0582,
      "step": 41270
    },
    {
      "epoch": 1.6725755150827575,
      "grad_norm": 0.5830888152122498,
      "learning_rate": 8.211506716574602e-05,
      "loss": 1.0932,
      "step": 41280
    },
    {
      "epoch": 1.672980693259861,
      "grad_norm": 0.5677667260169983,
      "learning_rate": 8.207329368651868e-05,
      "loss": 1.0996,
      "step": 41290
    },
    {
      "epoch": 1.6733858714369645,
      "grad_norm": 0.6216022372245789,
      "learning_rate": 8.203152343916361e-05,
      "loss": 1.0835,
      "step": 41300
    },
    {
      "epoch": 1.6737910496140678,
      "grad_norm": 0.5340301990509033,
      "learning_rate": 8.198975643121125e-05,
      "loss": 1.0967,
      "step": 41310
    },
    {
      "epoch": 1.674196227791171,
      "grad_norm": 0.6516919136047363,
      "learning_rate": 8.194799267019144e-05,
      "loss": 1.0933,
      "step": 41320
    },
    {
      "epoch": 1.6746014059682746,
      "grad_norm": 0.6240261197090149,
      "learning_rate": 8.190623216363348e-05,
      "loss": 1.0954,
      "step": 41330
    },
    {
      "epoch": 1.675006584145378,
      "grad_norm": 0.6708238124847412,
      "learning_rate": 8.18644749190661e-05,
      "loss": 1.0579,
      "step": 41340
    },
    {
      "epoch": 1.6754117623224813,
      "grad_norm": 0.5357533097267151,
      "learning_rate": 8.182272094401735e-05,
      "loss": 1.0894,
      "step": 41350
    },
    {
      "epoch": 1.6758169404995846,
      "grad_norm": 0.6665723919868469,
      "learning_rate": 8.178097024601475e-05,
      "loss": 1.0791,
      "step": 41360
    },
    {
      "epoch": 1.6762221186766881,
      "grad_norm": 0.6504266858100891,
      "learning_rate": 8.173922283258524e-05,
      "loss": 1.0905,
      "step": 41370
    },
    {
      "epoch": 1.6766272968537914,
      "grad_norm": 0.6282165050506592,
      "learning_rate": 8.169747871125515e-05,
      "loss": 1.0975,
      "step": 41380
    },
    {
      "epoch": 1.6770324750308947,
      "grad_norm": 0.5665832757949829,
      "learning_rate": 8.165573788955018e-05,
      "loss": 1.0478,
      "step": 41390
    },
    {
      "epoch": 1.6774376532079982,
      "grad_norm": 0.6236062049865723,
      "learning_rate": 8.161400037499551e-05,
      "loss": 1.0898,
      "step": 41400
    },
    {
      "epoch": 1.6778428313851017,
      "grad_norm": 0.569297730922699,
      "learning_rate": 8.157226617511569e-05,
      "loss": 1.0853,
      "step": 41410
    },
    {
      "epoch": 1.678248009562205,
      "grad_norm": 1.106619954109192,
      "learning_rate": 8.153053529743465e-05,
      "loss": 1.1154,
      "step": 41420
    },
    {
      "epoch": 1.6786531877393083,
      "grad_norm": 0.6145444512367249,
      "learning_rate": 8.148880774947574e-05,
      "loss": 1.1005,
      "step": 41430
    },
    {
      "epoch": 1.6790583659164118,
      "grad_norm": 0.5769622325897217,
      "learning_rate": 8.144708353876172e-05,
      "loss": 1.1573,
      "step": 41440
    },
    {
      "epoch": 1.6794635440935153,
      "grad_norm": 0.6475668549537659,
      "learning_rate": 8.140536267281468e-05,
      "loss": 1.0904,
      "step": 41450
    },
    {
      "epoch": 1.6798687222706183,
      "grad_norm": 0.6542167067527771,
      "learning_rate": 8.136364515915626e-05,
      "loss": 1.0511,
      "step": 41460
    },
    {
      "epoch": 1.6802739004477218,
      "grad_norm": 0.6032489538192749,
      "learning_rate": 8.132193100530734e-05,
      "loss": 1.1073,
      "step": 41470
    },
    {
      "epoch": 1.6806790786248254,
      "grad_norm": 0.5692014694213867,
      "learning_rate": 8.128022021878827e-05,
      "loss": 1.0779,
      "step": 41480
    },
    {
      "epoch": 1.6810842568019286,
      "grad_norm": 0.6242772340774536,
      "learning_rate": 8.123851280711877e-05,
      "loss": 1.0485,
      "step": 41490
    },
    {
      "epoch": 1.681489434979032,
      "grad_norm": 0.6253157258033752,
      "learning_rate": 8.119680877781792e-05,
      "loss": 1.0923,
      "step": 41500
    },
    {
      "epoch": 1.6818946131561354,
      "grad_norm": 0.5916028618812561,
      "learning_rate": 8.11551081384043e-05,
      "loss": 1.1263,
      "step": 41510
    },
    {
      "epoch": 1.682299791333239,
      "grad_norm": 0.6092944741249084,
      "learning_rate": 8.111341089639579e-05,
      "loss": 1.0862,
      "step": 41520
    },
    {
      "epoch": 1.6827049695103422,
      "grad_norm": 0.5257608294487,
      "learning_rate": 8.107171705930966e-05,
      "loss": 1.0911,
      "step": 41530
    },
    {
      "epoch": 1.6831101476874455,
      "grad_norm": 0.608534574508667,
      "learning_rate": 8.103002663466258e-05,
      "loss": 1.0559,
      "step": 41540
    },
    {
      "epoch": 1.683515325864549,
      "grad_norm": 0.6477605700492859,
      "learning_rate": 8.098833962997061e-05,
      "loss": 1.07,
      "step": 41550
    },
    {
      "epoch": 1.6839205040416523,
      "grad_norm": 0.5437607169151306,
      "learning_rate": 8.094665605274913e-05,
      "loss": 1.0452,
      "step": 41560
    },
    {
      "epoch": 1.6843256822187556,
      "grad_norm": 0.6297755837440491,
      "learning_rate": 8.090497591051308e-05,
      "loss": 1.1025,
      "step": 41570
    },
    {
      "epoch": 1.684730860395859,
      "grad_norm": 0.5964361429214478,
      "learning_rate": 8.08632992107766e-05,
      "loss": 1.0948,
      "step": 41580
    },
    {
      "epoch": 1.6851360385729626,
      "grad_norm": 0.6963759660720825,
      "learning_rate": 8.082162596105326e-05,
      "loss": 1.0425,
      "step": 41590
    },
    {
      "epoch": 1.6855412167500659,
      "grad_norm": 0.5937255024909973,
      "learning_rate": 8.077995616885604e-05,
      "loss": 1.1125,
      "step": 41600
    },
    {
      "epoch": 1.6859463949271691,
      "grad_norm": 0.6546024084091187,
      "learning_rate": 8.073828984169722e-05,
      "loss": 1.0927,
      "step": 41610
    },
    {
      "epoch": 1.6863515731042726,
      "grad_norm": 0.57801753282547,
      "learning_rate": 8.069662698708862e-05,
      "loss": 1.0432,
      "step": 41620
    },
    {
      "epoch": 1.6867567512813761,
      "grad_norm": 0.6321067810058594,
      "learning_rate": 8.065496761254126e-05,
      "loss": 1.0564,
      "step": 41630
    },
    {
      "epoch": 1.6871619294584792,
      "grad_norm": 0.5949912667274475,
      "learning_rate": 8.061331172556559e-05,
      "loss": 1.1078,
      "step": 41640
    },
    {
      "epoch": 1.6875671076355827,
      "grad_norm": 0.659115195274353,
      "learning_rate": 8.057165933367145e-05,
      "loss": 1.0926,
      "step": 41650
    },
    {
      "epoch": 1.6879722858126862,
      "grad_norm": 0.6424001455307007,
      "learning_rate": 8.053001044436803e-05,
      "loss": 1.1129,
      "step": 41660
    },
    {
      "epoch": 1.6883774639897895,
      "grad_norm": 0.5578034520149231,
      "learning_rate": 8.048836506516392e-05,
      "loss": 1.109,
      "step": 41670
    },
    {
      "epoch": 1.6887826421668928,
      "grad_norm": 0.6404385566711426,
      "learning_rate": 8.044672320356706e-05,
      "loss": 1.0823,
      "step": 41680
    },
    {
      "epoch": 1.6891878203439963,
      "grad_norm": 0.50143963098526,
      "learning_rate": 8.040508486708474e-05,
      "loss": 1.0844,
      "step": 41690
    },
    {
      "epoch": 1.6895929985210998,
      "grad_norm": 0.6037933826446533,
      "learning_rate": 8.036345006322359e-05,
      "loss": 1.0838,
      "step": 41700
    },
    {
      "epoch": 1.689998176698203,
      "grad_norm": 0.6772459745407104,
      "learning_rate": 8.03218187994897e-05,
      "loss": 1.104,
      "step": 41710
    },
    {
      "epoch": 1.6904033548753064,
      "grad_norm": 0.6678712368011475,
      "learning_rate": 8.028019108338842e-05,
      "loss": 1.0969,
      "step": 41720
    },
    {
      "epoch": 1.6908085330524099,
      "grad_norm": 0.6103267669677734,
      "learning_rate": 8.023856692242449e-05,
      "loss": 1.1237,
      "step": 41730
    },
    {
      "epoch": 1.6912137112295131,
      "grad_norm": 0.5825545191764832,
      "learning_rate": 8.019694632410205e-05,
      "loss": 1.0843,
      "step": 41740
    },
    {
      "epoch": 1.6916188894066164,
      "grad_norm": 0.6241534948348999,
      "learning_rate": 8.015532929592457e-05,
      "loss": 1.0888,
      "step": 41750
    },
    {
      "epoch": 1.69202406758372,
      "grad_norm": 0.594817578792572,
      "learning_rate": 8.011371584539487e-05,
      "loss": 1.0532,
      "step": 41760
    },
    {
      "epoch": 1.6924292457608234,
      "grad_norm": 0.6272482872009277,
      "learning_rate": 8.007210598001512e-05,
      "loss": 1.0811,
      "step": 41770
    },
    {
      "epoch": 1.6928344239379267,
      "grad_norm": 0.6640413999557495,
      "learning_rate": 8.00304997072868e-05,
      "loss": 1.08,
      "step": 41780
    },
    {
      "epoch": 1.69323960211503,
      "grad_norm": 0.5955937504768372,
      "learning_rate": 7.998889703471091e-05,
      "loss": 1.0748,
      "step": 41790
    },
    {
      "epoch": 1.6936447802921335,
      "grad_norm": 0.6034173369407654,
      "learning_rate": 7.994729796978761e-05,
      "loss": 1.0776,
      "step": 41800
    },
    {
      "epoch": 1.694049958469237,
      "grad_norm": 0.5841090679168701,
      "learning_rate": 7.99057025200165e-05,
      "loss": 1.1454,
      "step": 41810
    },
    {
      "epoch": 1.69445513664634,
      "grad_norm": 0.6329482793807983,
      "learning_rate": 7.98641106928965e-05,
      "loss": 1.0602,
      "step": 41820
    },
    {
      "epoch": 1.6948603148234436,
      "grad_norm": 0.5933811068534851,
      "learning_rate": 7.982252249592593e-05,
      "loss": 1.0766,
      "step": 41830
    },
    {
      "epoch": 1.695265493000547,
      "grad_norm": 0.6235707402229309,
      "learning_rate": 7.978093793660233e-05,
      "loss": 1.0955,
      "step": 41840
    },
    {
      "epoch": 1.6956706711776504,
      "grad_norm": 0.6618982553482056,
      "learning_rate": 7.973935702242278e-05,
      "loss": 1.1113,
      "step": 41850
    },
    {
      "epoch": 1.6960758493547536,
      "grad_norm": 0.6178349256515503,
      "learning_rate": 7.969777976088354e-05,
      "loss": 1.0734,
      "step": 41860
    },
    {
      "epoch": 1.6964810275318571,
      "grad_norm": 0.5246672630310059,
      "learning_rate": 7.965620615948026e-05,
      "loss": 1.0493,
      "step": 41870
    },
    {
      "epoch": 1.6968862057089606,
      "grad_norm": 0.532846987247467,
      "learning_rate": 7.961463622570795e-05,
      "loss": 1.0994,
      "step": 41880
    },
    {
      "epoch": 1.697291383886064,
      "grad_norm": 0.5877938866615295,
      "learning_rate": 7.957306996706093e-05,
      "loss": 1.0791,
      "step": 41890
    },
    {
      "epoch": 1.6976965620631672,
      "grad_norm": 0.5789476037025452,
      "learning_rate": 7.953150739103284e-05,
      "loss": 1.0948,
      "step": 41900
    },
    {
      "epoch": 1.6981017402402707,
      "grad_norm": 0.6186649203300476,
      "learning_rate": 7.948994850511677e-05,
      "loss": 1.1056,
      "step": 41910
    },
    {
      "epoch": 1.698506918417374,
      "grad_norm": 0.6360160112380981,
      "learning_rate": 7.9448393316805e-05,
      "loss": 1.0756,
      "step": 41920
    },
    {
      "epoch": 1.6989120965944773,
      "grad_norm": 0.591882050037384,
      "learning_rate": 7.940684183358924e-05,
      "loss": 1.0883,
      "step": 41930
    },
    {
      "epoch": 1.6993172747715808,
      "grad_norm": 0.5510368943214417,
      "learning_rate": 7.936529406296047e-05,
      "loss": 1.0379,
      "step": 41940
    },
    {
      "epoch": 1.6997224529486843,
      "grad_norm": 0.7873096466064453,
      "learning_rate": 7.932375001240898e-05,
      "loss": 1.1099,
      "step": 41950
    },
    {
      "epoch": 1.7001276311257876,
      "grad_norm": 0.6100546717643738,
      "learning_rate": 7.928220968942455e-05,
      "loss": 1.0803,
      "step": 41960
    },
    {
      "epoch": 1.7005328093028909,
      "grad_norm": 0.6516473293304443,
      "learning_rate": 7.924067310149611e-05,
      "loss": 1.1285,
      "step": 41970
    },
    {
      "epoch": 1.7009379874799944,
      "grad_norm": 0.6563409566879272,
      "learning_rate": 7.9199140256112e-05,
      "loss": 1.1305,
      "step": 41980
    },
    {
      "epoch": 1.7013431656570979,
      "grad_norm": 0.6524419188499451,
      "learning_rate": 7.915761116075983e-05,
      "loss": 1.091,
      "step": 41990
    },
    {
      "epoch": 1.701748343834201,
      "grad_norm": 0.5856846570968628,
      "learning_rate": 7.911608582292661e-05,
      "loss": 1.0641,
      "step": 42000
    },
    {
      "epoch": 1.7021535220113044,
      "grad_norm": 0.5868181586265564,
      "learning_rate": 7.907456425009861e-05,
      "loss": 1.0602,
      "step": 42010
    },
    {
      "epoch": 1.702558700188408,
      "grad_norm": 0.582825243473053,
      "learning_rate": 7.903304644976148e-05,
      "loss": 1.0609,
      "step": 42020
    },
    {
      "epoch": 1.7029638783655112,
      "grad_norm": 0.7013672590255737,
      "learning_rate": 7.89915324294001e-05,
      "loss": 1.0683,
      "step": 42030
    },
    {
      "epoch": 1.7033690565426145,
      "grad_norm": 0.6720321774482727,
      "learning_rate": 7.895002219649878e-05,
      "loss": 1.082,
      "step": 42040
    },
    {
      "epoch": 1.703774234719718,
      "grad_norm": 0.5609248876571655,
      "learning_rate": 7.890851575854108e-05,
      "loss": 1.0628,
      "step": 42050
    },
    {
      "epoch": 1.7041794128968215,
      "grad_norm": 0.6618645787239075,
      "learning_rate": 7.886701312300986e-05,
      "loss": 1.1067,
      "step": 42060
    },
    {
      "epoch": 1.7045845910739248,
      "grad_norm": 0.6162015199661255,
      "learning_rate": 7.882551429738729e-05,
      "loss": 1.0732,
      "step": 42070
    },
    {
      "epoch": 1.704989769251028,
      "grad_norm": 0.5910059213638306,
      "learning_rate": 7.878401928915496e-05,
      "loss": 1.0716,
      "step": 42080
    },
    {
      "epoch": 1.7053949474281316,
      "grad_norm": 0.6165366172790527,
      "learning_rate": 7.874252810579367e-05,
      "loss": 1.0961,
      "step": 42090
    },
    {
      "epoch": 1.7058001256052349,
      "grad_norm": 0.632723867893219,
      "learning_rate": 7.870104075478354e-05,
      "loss": 1.0914,
      "step": 42100
    },
    {
      "epoch": 1.7062053037823381,
      "grad_norm": 0.5970762968063354,
      "learning_rate": 7.865955724360404e-05,
      "loss": 1.0926,
      "step": 42110
    },
    {
      "epoch": 1.7066104819594416,
      "grad_norm": 0.662216305732727,
      "learning_rate": 7.861807757973387e-05,
      "loss": 1.0904,
      "step": 42120
    },
    {
      "epoch": 1.7070156601365452,
      "grad_norm": 0.5830686092376709,
      "learning_rate": 7.857660177065115e-05,
      "loss": 1.072,
      "step": 42130
    },
    {
      "epoch": 1.7074208383136484,
      "grad_norm": 0.6249591112136841,
      "learning_rate": 7.853512982383322e-05,
      "loss": 1.0803,
      "step": 42140
    },
    {
      "epoch": 1.7078260164907517,
      "grad_norm": 0.606816828250885,
      "learning_rate": 7.849366174675677e-05,
      "loss": 1.0916,
      "step": 42150
    },
    {
      "epoch": 1.7082311946678552,
      "grad_norm": 0.6057868003845215,
      "learning_rate": 7.845219754689772e-05,
      "loss": 1.0946,
      "step": 42160
    },
    {
      "epoch": 1.7086363728449587,
      "grad_norm": 0.6349979639053345,
      "learning_rate": 7.84107372317314e-05,
      "loss": 1.0916,
      "step": 42170
    },
    {
      "epoch": 1.7090415510220618,
      "grad_norm": 0.6006426811218262,
      "learning_rate": 7.836928080873229e-05,
      "loss": 1.0844,
      "step": 42180
    },
    {
      "epoch": 1.7094467291991653,
      "grad_norm": 0.6190152168273926,
      "learning_rate": 7.832782828537437e-05,
      "loss": 1.1108,
      "step": 42190
    },
    {
      "epoch": 1.7098519073762688,
      "grad_norm": 0.6387402415275574,
      "learning_rate": 7.828637966913075e-05,
      "loss": 1.094,
      "step": 42200
    },
    {
      "epoch": 1.710257085553372,
      "grad_norm": 0.6757359504699707,
      "learning_rate": 7.82449349674739e-05,
      "loss": 1.0681,
      "step": 42210
    },
    {
      "epoch": 1.7106622637304754,
      "grad_norm": 0.6308721303939819,
      "learning_rate": 7.820349418787558e-05,
      "loss": 1.0904,
      "step": 42220
    },
    {
      "epoch": 1.7110674419075789,
      "grad_norm": 0.7610599994659424,
      "learning_rate": 7.81620573378068e-05,
      "loss": 1.0575,
      "step": 42230
    },
    {
      "epoch": 1.7114726200846824,
      "grad_norm": 0.6256186366081238,
      "learning_rate": 7.812062442473798e-05,
      "loss": 1.0722,
      "step": 42240
    },
    {
      "epoch": 1.7118777982617857,
      "grad_norm": 0.7230322360992432,
      "learning_rate": 7.80791954561387e-05,
      "loss": 1.1034,
      "step": 42250
    },
    {
      "epoch": 1.712282976438889,
      "grad_norm": 0.6067711114883423,
      "learning_rate": 7.803777043947789e-05,
      "loss": 1.0868,
      "step": 42260
    },
    {
      "epoch": 1.7126881546159924,
      "grad_norm": 0.6005358695983887,
      "learning_rate": 7.799634938222376e-05,
      "loss": 1.0666,
      "step": 42270
    },
    {
      "epoch": 1.7130933327930957,
      "grad_norm": 0.6411232948303223,
      "learning_rate": 7.795493229184379e-05,
      "loss": 1.0875,
      "step": 42280
    },
    {
      "epoch": 1.713498510970199,
      "grad_norm": 0.6443528532981873,
      "learning_rate": 7.791351917580472e-05,
      "loss": 1.1082,
      "step": 42290
    },
    {
      "epoch": 1.7139036891473025,
      "grad_norm": 0.6795934438705444,
      "learning_rate": 7.787211004157273e-05,
      "loss": 1.0567,
      "step": 42300
    },
    {
      "epoch": 1.714308867324406,
      "grad_norm": 0.688274621963501,
      "learning_rate": 7.783070489661306e-05,
      "loss": 1.0897,
      "step": 42310
    },
    {
      "epoch": 1.7147140455015093,
      "grad_norm": 0.6286495327949524,
      "learning_rate": 7.778930374839038e-05,
      "loss": 1.097,
      "step": 42320
    },
    {
      "epoch": 1.7151192236786126,
      "grad_norm": 0.6837913990020752,
      "learning_rate": 7.774790660436858e-05,
      "loss": 1.0865,
      "step": 42330
    },
    {
      "epoch": 1.715524401855716,
      "grad_norm": 0.5706535577774048,
      "learning_rate": 7.770651347201085e-05,
      "loss": 1.09,
      "step": 42340
    },
    {
      "epoch": 1.7159295800328196,
      "grad_norm": 0.5848308205604553,
      "learning_rate": 7.766512435877962e-05,
      "loss": 1.1311,
      "step": 42350
    },
    {
      "epoch": 1.7163347582099227,
      "grad_norm": 0.6235148906707764,
      "learning_rate": 7.762373927213667e-05,
      "loss": 1.0744,
      "step": 42360
    },
    {
      "epoch": 1.7167399363870262,
      "grad_norm": 0.583672046661377,
      "learning_rate": 7.758235821954297e-05,
      "loss": 1.077,
      "step": 42370
    },
    {
      "epoch": 1.7171451145641297,
      "grad_norm": 0.5354995727539062,
      "learning_rate": 7.754098120845885e-05,
      "loss": 1.0866,
      "step": 42380
    },
    {
      "epoch": 1.717550292741233,
      "grad_norm": 0.5851695537567139,
      "learning_rate": 7.749960824634382e-05,
      "loss": 1.1037,
      "step": 42390
    },
    {
      "epoch": 1.7179554709183362,
      "grad_norm": 0.6210392117500305,
      "learning_rate": 7.745823934065671e-05,
      "loss": 1.0556,
      "step": 42400
    },
    {
      "epoch": 1.7183606490954397,
      "grad_norm": 0.603327214717865,
      "learning_rate": 7.74168744988556e-05,
      "loss": 1.0918,
      "step": 42410
    },
    {
      "epoch": 1.7187658272725432,
      "grad_norm": 0.567908525466919,
      "learning_rate": 7.737551372839791e-05,
      "loss": 1.123,
      "step": 42420
    },
    {
      "epoch": 1.7191710054496465,
      "grad_norm": 0.6880656480789185,
      "learning_rate": 7.733415703674021e-05,
      "loss": 1.0901,
      "step": 42430
    },
    {
      "epoch": 1.7195761836267498,
      "grad_norm": 0.6616040468215942,
      "learning_rate": 7.72928044313384e-05,
      "loss": 1.1099,
      "step": 42440
    },
    {
      "epoch": 1.7199813618038533,
      "grad_norm": 0.5772875547409058,
      "learning_rate": 7.725145591964763e-05,
      "loss": 1.0713,
      "step": 42450
    },
    {
      "epoch": 1.7203865399809566,
      "grad_norm": 0.7275482416152954,
      "learning_rate": 7.721011150912228e-05,
      "loss": 1.0749,
      "step": 42460
    },
    {
      "epoch": 1.7207917181580599,
      "grad_norm": 0.6244465112686157,
      "learning_rate": 7.716877120721611e-05,
      "loss": 1.0858,
      "step": 42470
    },
    {
      "epoch": 1.7211968963351634,
      "grad_norm": 0.6336467862129211,
      "learning_rate": 7.7127435021382e-05,
      "loss": 1.0829,
      "step": 42480
    },
    {
      "epoch": 1.7216020745122669,
      "grad_norm": 0.5654087662696838,
      "learning_rate": 7.708610295907218e-05,
      "loss": 1.105,
      "step": 42490
    },
    {
      "epoch": 1.7220072526893702,
      "grad_norm": 0.6063326597213745,
      "learning_rate": 7.704477502773803e-05,
      "loss": 1.0797,
      "step": 42500
    },
    {
      "epoch": 1.7224124308664734,
      "grad_norm": 0.5860738754272461,
      "learning_rate": 7.700345123483032e-05,
      "loss": 1.0298,
      "step": 42510
    },
    {
      "epoch": 1.722817609043577,
      "grad_norm": 0.6219716668128967,
      "learning_rate": 7.696213158779895e-05,
      "loss": 1.1224,
      "step": 42520
    },
    {
      "epoch": 1.7232227872206805,
      "grad_norm": 0.5905763506889343,
      "learning_rate": 7.692081609409321e-05,
      "loss": 1.0837,
      "step": 42530
    },
    {
      "epoch": 1.7236279653977835,
      "grad_norm": 0.6365659832954407,
      "learning_rate": 7.68795047611615e-05,
      "loss": 1.0886,
      "step": 42540
    },
    {
      "epoch": 1.724033143574887,
      "grad_norm": 0.601661741733551,
      "learning_rate": 7.683819759645157e-05,
      "loss": 1.0929,
      "step": 42550
    },
    {
      "epoch": 1.7244383217519905,
      "grad_norm": 0.5781034231185913,
      "learning_rate": 7.679689460741036e-05,
      "loss": 1.0877,
      "step": 42560
    },
    {
      "epoch": 1.7248434999290938,
      "grad_norm": 0.6078129410743713,
      "learning_rate": 7.675559580148402e-05,
      "loss": 1.1049,
      "step": 42570
    },
    {
      "epoch": 1.725248678106197,
      "grad_norm": 0.6642013192176819,
      "learning_rate": 7.671430118611812e-05,
      "loss": 1.0479,
      "step": 42580
    },
    {
      "epoch": 1.7256538562833006,
      "grad_norm": 0.562610387802124,
      "learning_rate": 7.667301076875731e-05,
      "loss": 1.1123,
      "step": 42590
    },
    {
      "epoch": 1.726059034460404,
      "grad_norm": 0.5873477458953857,
      "learning_rate": 7.66317245568455e-05,
      "loss": 1.0985,
      "step": 42600
    },
    {
      "epoch": 1.7264642126375074,
      "grad_norm": 0.6712811589241028,
      "learning_rate": 7.659044255782593e-05,
      "loss": 1.1289,
      "step": 42610
    },
    {
      "epoch": 1.7268693908146107,
      "grad_norm": 0.5659089088439941,
      "learning_rate": 7.654916477914096e-05,
      "loss": 1.0909,
      "step": 42620
    },
    {
      "epoch": 1.7272745689917142,
      "grad_norm": 0.6328191161155701,
      "learning_rate": 7.650789122823225e-05,
      "loss": 1.0921,
      "step": 42630
    },
    {
      "epoch": 1.7276797471688174,
      "grad_norm": 0.6005386710166931,
      "learning_rate": 7.646662191254079e-05,
      "loss": 1.0508,
      "step": 42640
    },
    {
      "epoch": 1.7280849253459207,
      "grad_norm": 0.615166187286377,
      "learning_rate": 7.642535683950666e-05,
      "loss": 1.0885,
      "step": 42650
    },
    {
      "epoch": 1.7284901035230242,
      "grad_norm": 0.6085776090621948,
      "learning_rate": 7.638409601656923e-05,
      "loss": 1.0709,
      "step": 42660
    },
    {
      "epoch": 1.7288952817001277,
      "grad_norm": 0.7155930995941162,
      "learning_rate": 7.634283945116709e-05,
      "loss": 1.04,
      "step": 42670
    },
    {
      "epoch": 1.729300459877231,
      "grad_norm": 0.5784938335418701,
      "learning_rate": 7.630158715073813e-05,
      "loss": 1.0574,
      "step": 42680
    },
    {
      "epoch": 1.7297056380543343,
      "grad_norm": 0.6009300947189331,
      "learning_rate": 7.626033912271936e-05,
      "loss": 1.0801,
      "step": 42690
    },
    {
      "epoch": 1.7301108162314378,
      "grad_norm": 0.580877423286438,
      "learning_rate": 7.621909537454714e-05,
      "loss": 1.0772,
      "step": 42700
    },
    {
      "epoch": 1.7305159944085413,
      "grad_norm": 0.5727139711380005,
      "learning_rate": 7.617785591365694e-05,
      "loss": 1.0942,
      "step": 42710
    },
    {
      "epoch": 1.7309211725856444,
      "grad_norm": 0.5385617017745972,
      "learning_rate": 7.613662074748356e-05,
      "loss": 1.0862,
      "step": 42720
    },
    {
      "epoch": 1.7313263507627479,
      "grad_norm": 0.6322827935218811,
      "learning_rate": 7.609538988346096e-05,
      "loss": 1.0572,
      "step": 42730
    },
    {
      "epoch": 1.7317315289398514,
      "grad_norm": 0.5766319632530212,
      "learning_rate": 7.605416332902233e-05,
      "loss": 1.0917,
      "step": 42740
    },
    {
      "epoch": 1.7321367071169547,
      "grad_norm": 0.6075160503387451,
      "learning_rate": 7.601294109160012e-05,
      "loss": 1.0707,
      "step": 42750
    },
    {
      "epoch": 1.732541885294058,
      "grad_norm": 0.6175504922866821,
      "learning_rate": 7.5971723178626e-05,
      "loss": 1.104,
      "step": 42760
    },
    {
      "epoch": 1.7329470634711615,
      "grad_norm": 0.6501725316047668,
      "learning_rate": 7.593050959753081e-05,
      "loss": 1.07,
      "step": 42770
    },
    {
      "epoch": 1.733352241648265,
      "grad_norm": 0.6015642285346985,
      "learning_rate": 7.588930035574465e-05,
      "loss": 1.0789,
      "step": 42780
    },
    {
      "epoch": 1.7337574198253682,
      "grad_norm": 0.6304374933242798,
      "learning_rate": 7.584809546069684e-05,
      "loss": 1.0837,
      "step": 42790
    },
    {
      "epoch": 1.7341625980024715,
      "grad_norm": 0.6161757707595825,
      "learning_rate": 7.580689491981582e-05,
      "loss": 1.0644,
      "step": 42800
    },
    {
      "epoch": 1.734567776179575,
      "grad_norm": 0.6000008583068848,
      "learning_rate": 7.576569874052944e-05,
      "loss": 1.0765,
      "step": 42810
    },
    {
      "epoch": 1.7349729543566783,
      "grad_norm": 0.6347299814224243,
      "learning_rate": 7.572450693026462e-05,
      "loss": 1.102,
      "step": 42820
    },
    {
      "epoch": 1.7353781325337816,
      "grad_norm": 0.7259131073951721,
      "learning_rate": 7.568331949644749e-05,
      "loss": 1.0787,
      "step": 42830
    },
    {
      "epoch": 1.735783310710885,
      "grad_norm": 0.5990109443664551,
      "learning_rate": 7.564213644650346e-05,
      "loss": 1.0497,
      "step": 42840
    },
    {
      "epoch": 1.7361884888879886,
      "grad_norm": 0.5936081409454346,
      "learning_rate": 7.560095778785708e-05,
      "loss": 1.1024,
      "step": 42850
    },
    {
      "epoch": 1.7365936670650919,
      "grad_norm": 0.6973927021026611,
      "learning_rate": 7.555978352793218e-05,
      "loss": 1.0729,
      "step": 42860
    },
    {
      "epoch": 1.7369988452421952,
      "grad_norm": 0.6047586798667908,
      "learning_rate": 7.551861367415176e-05,
      "loss": 1.0857,
      "step": 42870
    },
    {
      "epoch": 1.7374040234192987,
      "grad_norm": 0.6165871620178223,
      "learning_rate": 7.547744823393801e-05,
      "loss": 1.0771,
      "step": 42880
    },
    {
      "epoch": 1.7378092015964022,
      "grad_norm": 0.6361507177352905,
      "learning_rate": 7.543628721471233e-05,
      "loss": 1.0793,
      "step": 42890
    },
    {
      "epoch": 1.7382143797735052,
      "grad_norm": 0.6371106505393982,
      "learning_rate": 7.539513062389536e-05,
      "loss": 1.0903,
      "step": 42900
    },
    {
      "epoch": 1.7386195579506087,
      "grad_norm": 0.6197506785392761,
      "learning_rate": 7.535397846890685e-05,
      "loss": 1.0899,
      "step": 42910
    },
    {
      "epoch": 1.7390247361277122,
      "grad_norm": 0.623746931552887,
      "learning_rate": 7.531283075716591e-05,
      "loss": 1.0933,
      "step": 42920
    },
    {
      "epoch": 1.7394299143048155,
      "grad_norm": 0.5851884484291077,
      "learning_rate": 7.527168749609073e-05,
      "loss": 1.1018,
      "step": 42930
    },
    {
      "epoch": 1.7398350924819188,
      "grad_norm": 0.5747804641723633,
      "learning_rate": 7.523054869309868e-05,
      "loss": 1.0856,
      "step": 42940
    },
    {
      "epoch": 1.7402402706590223,
      "grad_norm": 0.6425610780715942,
      "learning_rate": 7.518941435560639e-05,
      "loss": 1.0782,
      "step": 42950
    },
    {
      "epoch": 1.7406454488361258,
      "grad_norm": 0.6578708291053772,
      "learning_rate": 7.514828449102966e-05,
      "loss": 1.0935,
      "step": 42960
    },
    {
      "epoch": 1.741050627013229,
      "grad_norm": 0.6534647941589355,
      "learning_rate": 7.510715910678348e-05,
      "loss": 1.1209,
      "step": 42970
    },
    {
      "epoch": 1.7414558051903324,
      "grad_norm": 0.5272698402404785,
      "learning_rate": 7.506603821028207e-05,
      "loss": 1.0334,
      "step": 42980
    },
    {
      "epoch": 1.7418609833674359,
      "grad_norm": 0.6369834542274475,
      "learning_rate": 7.502492180893878e-05,
      "loss": 1.1074,
      "step": 42990
    },
    {
      "epoch": 1.7422661615445392,
      "grad_norm": 0.7020087838172913,
      "learning_rate": 7.498380991016618e-05,
      "loss": 1.0971,
      "step": 43000
    },
    {
      "epoch": 1.7426713397216425,
      "grad_norm": 0.5271212458610535,
      "learning_rate": 7.494270252137605e-05,
      "loss": 1.0305,
      "step": 43010
    },
    {
      "epoch": 1.743076517898746,
      "grad_norm": 0.596669614315033,
      "learning_rate": 7.490159964997931e-05,
      "loss": 1.1364,
      "step": 43020
    },
    {
      "epoch": 1.7434816960758495,
      "grad_norm": 0.5959604382514954,
      "learning_rate": 7.486050130338612e-05,
      "loss": 1.1219,
      "step": 43030
    },
    {
      "epoch": 1.7438868742529527,
      "grad_norm": 0.582251787185669,
      "learning_rate": 7.481940748900576e-05,
      "loss": 1.0785,
      "step": 43040
    },
    {
      "epoch": 1.744292052430056,
      "grad_norm": 0.6245517730712891,
      "learning_rate": 7.477831821424675e-05,
      "loss": 1.1082,
      "step": 43050
    },
    {
      "epoch": 1.7446972306071595,
      "grad_norm": 0.6348722577095032,
      "learning_rate": 7.473723348651678e-05,
      "loss": 1.081,
      "step": 43060
    },
    {
      "epoch": 1.745102408784263,
      "grad_norm": 0.6295092701911926,
      "learning_rate": 7.469615331322268e-05,
      "loss": 1.0844,
      "step": 43070
    },
    {
      "epoch": 1.745507586961366,
      "grad_norm": 0.6042221784591675,
      "learning_rate": 7.465507770177046e-05,
      "loss": 1.1052,
      "step": 43080
    },
    {
      "epoch": 1.7459127651384696,
      "grad_norm": 0.5502166152000427,
      "learning_rate": 7.461400665956541e-05,
      "loss": 1.0826,
      "step": 43090
    },
    {
      "epoch": 1.746317943315573,
      "grad_norm": 0.606246292591095,
      "learning_rate": 7.457294019401191e-05,
      "loss": 1.0884,
      "step": 43100
    },
    {
      "epoch": 1.7467231214926764,
      "grad_norm": 0.6135533452033997,
      "learning_rate": 7.45318783125135e-05,
      "loss": 1.0556,
      "step": 43110
    },
    {
      "epoch": 1.7471282996697797,
      "grad_norm": 0.5769482851028442,
      "learning_rate": 7.449082102247291e-05,
      "loss": 1.0673,
      "step": 43120
    },
    {
      "epoch": 1.7475334778468832,
      "grad_norm": 0.6512131690979004,
      "learning_rate": 7.44497683312921e-05,
      "loss": 1.0826,
      "step": 43130
    },
    {
      "epoch": 1.7479386560239867,
      "grad_norm": 0.585924506187439,
      "learning_rate": 7.440872024637205e-05,
      "loss": 1.0577,
      "step": 43140
    },
    {
      "epoch": 1.74834383420109,
      "grad_norm": 0.6175044178962708,
      "learning_rate": 7.436767677511314e-05,
      "loss": 1.0998,
      "step": 43150
    },
    {
      "epoch": 1.7487490123781932,
      "grad_norm": 0.6337620615959167,
      "learning_rate": 7.432663792491472e-05,
      "loss": 1.0771,
      "step": 43160
    },
    {
      "epoch": 1.7491541905552968,
      "grad_norm": 0.5720326900482178,
      "learning_rate": 7.428560370317542e-05,
      "loss": 1.0247,
      "step": 43170
    },
    {
      "epoch": 1.7495593687324,
      "grad_norm": 0.6628276109695435,
      "learning_rate": 7.424457411729296e-05,
      "loss": 1.0673,
      "step": 43180
    },
    {
      "epoch": 1.7499645469095033,
      "grad_norm": 0.5770721435546875,
      "learning_rate": 7.420354917466423e-05,
      "loss": 1.0718,
      "step": 43190
    },
    {
      "epoch": 1.7503697250866068,
      "grad_norm": 0.594289243221283,
      "learning_rate": 7.41625288826854e-05,
      "loss": 1.0918,
      "step": 43200
    },
    {
      "epoch": 1.7507749032637103,
      "grad_norm": 0.6583713293075562,
      "learning_rate": 7.412151324875164e-05,
      "loss": 1.1158,
      "step": 43210
    },
    {
      "epoch": 1.7511800814408136,
      "grad_norm": 0.5816066265106201,
      "learning_rate": 7.408050228025738e-05,
      "loss": 1.0867,
      "step": 43220
    },
    {
      "epoch": 1.751585259617917,
      "grad_norm": 0.6325855851173401,
      "learning_rate": 7.403949598459618e-05,
      "loss": 1.0705,
      "step": 43230
    },
    {
      "epoch": 1.7519904377950204,
      "grad_norm": 0.6092843413352966,
      "learning_rate": 7.399849436916077e-05,
      "loss": 1.1107,
      "step": 43240
    },
    {
      "epoch": 1.752395615972124,
      "grad_norm": 0.6766306161880493,
      "learning_rate": 7.395749744134294e-05,
      "loss": 1.1007,
      "step": 43250
    },
    {
      "epoch": 1.752800794149227,
      "grad_norm": 0.5424476265907288,
      "learning_rate": 7.391650520853386e-05,
      "loss": 1.092,
      "step": 43260
    },
    {
      "epoch": 1.7532059723263305,
      "grad_norm": 0.6181219220161438,
      "learning_rate": 7.387551767812364e-05,
      "loss": 1.0867,
      "step": 43270
    },
    {
      "epoch": 1.753611150503434,
      "grad_norm": 0.5975931882858276,
      "learning_rate": 7.383453485750163e-05,
      "loss": 1.0877,
      "step": 43280
    },
    {
      "epoch": 1.7540163286805373,
      "grad_norm": 0.6279584765434265,
      "learning_rate": 7.379355675405629e-05,
      "loss": 1.0879,
      "step": 43290
    },
    {
      "epoch": 1.7544215068576405,
      "grad_norm": 0.55378657579422,
      "learning_rate": 7.375258337517528e-05,
      "loss": 1.0533,
      "step": 43300
    },
    {
      "epoch": 1.754826685034744,
      "grad_norm": 0.6549389958381653,
      "learning_rate": 7.371161472824536e-05,
      "loss": 1.0947,
      "step": 43310
    },
    {
      "epoch": 1.7552318632118475,
      "grad_norm": 0.62941575050354,
      "learning_rate": 7.367065082065252e-05,
      "loss": 1.0537,
      "step": 43320
    },
    {
      "epoch": 1.7556370413889508,
      "grad_norm": 0.6556832194328308,
      "learning_rate": 7.362969165978179e-05,
      "loss": 1.0904,
      "step": 43330
    },
    {
      "epoch": 1.756042219566054,
      "grad_norm": 0.6234332919120789,
      "learning_rate": 7.358873725301738e-05,
      "loss": 1.0844,
      "step": 43340
    },
    {
      "epoch": 1.7564473977431576,
      "grad_norm": 0.7334731817245483,
      "learning_rate": 7.354778760774272e-05,
      "loss": 1.1097,
      "step": 43350
    },
    {
      "epoch": 1.756852575920261,
      "grad_norm": 0.6254906058311462,
      "learning_rate": 7.350684273134023e-05,
      "loss": 1.075,
      "step": 43360
    },
    {
      "epoch": 1.7572577540973642,
      "grad_norm": 0.6174623966217041,
      "learning_rate": 7.346590263119164e-05,
      "loss": 1.085,
      "step": 43370
    },
    {
      "epoch": 1.7576629322744677,
      "grad_norm": 0.6149426698684692,
      "learning_rate": 7.342496731467767e-05,
      "loss": 1.0985,
      "step": 43380
    },
    {
      "epoch": 1.7580681104515712,
      "grad_norm": 0.5668327808380127,
      "learning_rate": 7.338403678917829e-05,
      "loss": 1.094,
      "step": 43390
    },
    {
      "epoch": 1.7584732886286745,
      "grad_norm": 0.5944305062294006,
      "learning_rate": 7.334311106207252e-05,
      "loss": 1.1011,
      "step": 43400
    },
    {
      "epoch": 1.7588784668057778,
      "grad_norm": 0.6785864233970642,
      "learning_rate": 7.33021901407386e-05,
      "loss": 1.1272,
      "step": 43410
    },
    {
      "epoch": 1.7592836449828813,
      "grad_norm": 0.6031008958816528,
      "learning_rate": 7.326127403255379e-05,
      "loss": 1.1047,
      "step": 43420
    },
    {
      "epoch": 1.7596888231599848,
      "grad_norm": 0.593094527721405,
      "learning_rate": 7.322036274489461e-05,
      "loss": 1.0857,
      "step": 43430
    },
    {
      "epoch": 1.7600940013370878,
      "grad_norm": 0.5727397799491882,
      "learning_rate": 7.317945628513665e-05,
      "loss": 1.0712,
      "step": 43440
    },
    {
      "epoch": 1.7604991795141913,
      "grad_norm": 0.6378394961357117,
      "learning_rate": 7.31385546606546e-05,
      "loss": 1.0934,
      "step": 43450
    },
    {
      "epoch": 1.7609043576912948,
      "grad_norm": 0.735921323299408,
      "learning_rate": 7.309765787882235e-05,
      "loss": 1.0872,
      "step": 43460
    },
    {
      "epoch": 1.7613095358683981,
      "grad_norm": 0.6385384798049927,
      "learning_rate": 7.305676594701278e-05,
      "loss": 1.09,
      "step": 43470
    },
    {
      "epoch": 1.7617147140455014,
      "grad_norm": 0.6381013989448547,
      "learning_rate": 7.301587887259812e-05,
      "loss": 1.0618,
      "step": 43480
    },
    {
      "epoch": 1.762119892222605,
      "grad_norm": 0.6355891823768616,
      "learning_rate": 7.297499666294951e-05,
      "loss": 1.0619,
      "step": 43490
    },
    {
      "epoch": 1.7625250703997084,
      "grad_norm": 0.6312190294265747,
      "learning_rate": 7.293411932543734e-05,
      "loss": 1.1268,
      "step": 43500
    },
    {
      "epoch": 1.7629302485768117,
      "grad_norm": 0.726867139339447,
      "learning_rate": 7.289324686743106e-05,
      "loss": 1.1009,
      "step": 43510
    },
    {
      "epoch": 1.763335426753915,
      "grad_norm": 0.7145953178405762,
      "learning_rate": 7.285237929629928e-05,
      "loss": 1.0855,
      "step": 43520
    },
    {
      "epoch": 1.7637406049310185,
      "grad_norm": 0.6660085320472717,
      "learning_rate": 7.281151661940964e-05,
      "loss": 1.0838,
      "step": 43530
    },
    {
      "epoch": 1.7641457831081218,
      "grad_norm": 0.6392118334770203,
      "learning_rate": 7.277065884412908e-05,
      "loss": 1.0521,
      "step": 43540
    },
    {
      "epoch": 1.764550961285225,
      "grad_norm": 0.5790143013000488,
      "learning_rate": 7.272980597782348e-05,
      "loss": 1.0929,
      "step": 43550
    },
    {
      "epoch": 1.7649561394623285,
      "grad_norm": 0.595791757106781,
      "learning_rate": 7.268895802785788e-05,
      "loss": 1.0911,
      "step": 43560
    },
    {
      "epoch": 1.765361317639432,
      "grad_norm": 0.6016699075698853,
      "learning_rate": 7.264811500159652e-05,
      "loss": 1.0719,
      "step": 43570
    },
    {
      "epoch": 1.7657664958165353,
      "grad_norm": 0.5523437261581421,
      "learning_rate": 7.260727690640262e-05,
      "loss": 1.1182,
      "step": 43580
    },
    {
      "epoch": 1.7661716739936386,
      "grad_norm": 0.6251005530357361,
      "learning_rate": 7.256644374963857e-05,
      "loss": 1.051,
      "step": 43590
    },
    {
      "epoch": 1.7665768521707421,
      "grad_norm": 0.5913191437721252,
      "learning_rate": 7.252561553866592e-05,
      "loss": 1.0793,
      "step": 43600
    },
    {
      "epoch": 1.7669820303478456,
      "grad_norm": 0.6476854085922241,
      "learning_rate": 7.248479228084529e-05,
      "loss": 1.0869,
      "step": 43610
    },
    {
      "epoch": 1.7673872085249487,
      "grad_norm": 0.56357342004776,
      "learning_rate": 7.244397398353637e-05,
      "loss": 1.097,
      "step": 43620
    },
    {
      "epoch": 1.7677923867020522,
      "grad_norm": 0.6322107315063477,
      "learning_rate": 7.240316065409801e-05,
      "loss": 1.0926,
      "step": 43630
    },
    {
      "epoch": 1.7681975648791557,
      "grad_norm": 0.5937241315841675,
      "learning_rate": 7.236235229988808e-05,
      "loss": 1.0867,
      "step": 43640
    },
    {
      "epoch": 1.768602743056259,
      "grad_norm": 0.7642531991004944,
      "learning_rate": 7.23215489282637e-05,
      "loss": 1.0494,
      "step": 43650
    },
    {
      "epoch": 1.7690079212333623,
      "grad_norm": 0.6182461977005005,
      "learning_rate": 7.228075054658096e-05,
      "loss": 1.0454,
      "step": 43660
    },
    {
      "epoch": 1.7694130994104658,
      "grad_norm": 0.5716409087181091,
      "learning_rate": 7.223995716219511e-05,
      "loss": 1.0608,
      "step": 43670
    },
    {
      "epoch": 1.7698182775875693,
      "grad_norm": 0.6383996605873108,
      "learning_rate": 7.219916878246045e-05,
      "loss": 1.1032,
      "step": 43680
    },
    {
      "epoch": 1.7702234557646725,
      "grad_norm": 0.6008943319320679,
      "learning_rate": 7.215838541473045e-05,
      "loss": 1.0661,
      "step": 43690
    },
    {
      "epoch": 1.7706286339417758,
      "grad_norm": 0.5702589154243469,
      "learning_rate": 7.211760706635763e-05,
      "loss": 1.1033,
      "step": 43700
    },
    {
      "epoch": 1.7710338121188793,
      "grad_norm": 0.5940125584602356,
      "learning_rate": 7.207683374469362e-05,
      "loss": 1.0876,
      "step": 43710
    },
    {
      "epoch": 1.7714389902959826,
      "grad_norm": 0.6159881949424744,
      "learning_rate": 7.203606545708914e-05,
      "loss": 1.0676,
      "step": 43720
    },
    {
      "epoch": 1.771844168473086,
      "grad_norm": 0.574105441570282,
      "learning_rate": 7.199530221089398e-05,
      "loss": 1.1003,
      "step": 43730
    },
    {
      "epoch": 1.7722493466501894,
      "grad_norm": 0.6247740983963013,
      "learning_rate": 7.19545440134571e-05,
      "loss": 1.0748,
      "step": 43740
    },
    {
      "epoch": 1.772654524827293,
      "grad_norm": 0.5844593644142151,
      "learning_rate": 7.19137908721264e-05,
      "loss": 1.0741,
      "step": 43750
    },
    {
      "epoch": 1.7730597030043962,
      "grad_norm": 0.6064063310623169,
      "learning_rate": 7.187304279424899e-05,
      "loss": 1.1135,
      "step": 43760
    },
    {
      "epoch": 1.7734648811814995,
      "grad_norm": 0.6328258514404297,
      "learning_rate": 7.18322997871711e-05,
      "loss": 1.0945,
      "step": 43770
    },
    {
      "epoch": 1.773870059358603,
      "grad_norm": 0.7519960999488831,
      "learning_rate": 7.179156185823792e-05,
      "loss": 1.0884,
      "step": 43780
    },
    {
      "epoch": 1.7742752375357065,
      "grad_norm": 0.7395449280738831,
      "learning_rate": 7.17508290147938e-05,
      "loss": 1.0825,
      "step": 43790
    },
    {
      "epoch": 1.7746804157128095,
      "grad_norm": 0.5618184804916382,
      "learning_rate": 7.171010126418218e-05,
      "loss": 1.0768,
      "step": 43800
    },
    {
      "epoch": 1.775085593889913,
      "grad_norm": 0.6558398604393005,
      "learning_rate": 7.166937861374552e-05,
      "loss": 1.0603,
      "step": 43810
    },
    {
      "epoch": 1.7754907720670166,
      "grad_norm": 0.6446412801742554,
      "learning_rate": 7.162866107082544e-05,
      "loss": 1.0601,
      "step": 43820
    },
    {
      "epoch": 1.7758959502441198,
      "grad_norm": 0.5918707251548767,
      "learning_rate": 7.158794864276261e-05,
      "loss": 1.0979,
      "step": 43830
    },
    {
      "epoch": 1.7763011284212231,
      "grad_norm": 0.5496525764465332,
      "learning_rate": 7.154724133689677e-05,
      "loss": 1.1124,
      "step": 43840
    },
    {
      "epoch": 1.7767063065983266,
      "grad_norm": 0.6576629281044006,
      "learning_rate": 7.150653916056671e-05,
      "loss": 1.0664,
      "step": 43850
    },
    {
      "epoch": 1.7771114847754301,
      "grad_norm": 0.6951850652694702,
      "learning_rate": 7.146584212111032e-05,
      "loss": 1.0574,
      "step": 43860
    },
    {
      "epoch": 1.7775166629525334,
      "grad_norm": 0.5932276844978333,
      "learning_rate": 7.142515022586456e-05,
      "loss": 1.1072,
      "step": 43870
    },
    {
      "epoch": 1.7779218411296367,
      "grad_norm": 0.6034003496170044,
      "learning_rate": 7.138446348216551e-05,
      "loss": 1.0792,
      "step": 43880
    },
    {
      "epoch": 1.7783270193067402,
      "grad_norm": 0.6389949321746826,
      "learning_rate": 7.134378189734826e-05,
      "loss": 1.0808,
      "step": 43890
    },
    {
      "epoch": 1.7787321974838435,
      "grad_norm": 0.5962080359458923,
      "learning_rate": 7.130310547874698e-05,
      "loss": 1.1082,
      "step": 43900
    },
    {
      "epoch": 1.7791373756609468,
      "grad_norm": 0.60044264793396,
      "learning_rate": 7.126243423369494e-05,
      "loss": 1.0672,
      "step": 43910
    },
    {
      "epoch": 1.7795425538380503,
      "grad_norm": 0.646701991558075,
      "learning_rate": 7.12217681695244e-05,
      "loss": 1.0654,
      "step": 43920
    },
    {
      "epoch": 1.7799477320151538,
      "grad_norm": 0.6609534025192261,
      "learning_rate": 7.118110729356682e-05,
      "loss": 1.0678,
      "step": 43930
    },
    {
      "epoch": 1.780352910192257,
      "grad_norm": 0.6478208899497986,
      "learning_rate": 7.114045161315261e-05,
      "loss": 1.0659,
      "step": 43940
    },
    {
      "epoch": 1.7807580883693603,
      "grad_norm": 0.5828163623809814,
      "learning_rate": 7.109980113561128e-05,
      "loss": 1.1093,
      "step": 43950
    },
    {
      "epoch": 1.7811632665464638,
      "grad_norm": 0.6822598576545715,
      "learning_rate": 7.105915586827139e-05,
      "loss": 1.0803,
      "step": 43960
    },
    {
      "epoch": 1.7815684447235673,
      "grad_norm": 0.5759316086769104,
      "learning_rate": 7.101851581846055e-05,
      "loss": 1.0627,
      "step": 43970
    },
    {
      "epoch": 1.7819736229006704,
      "grad_norm": 0.5914221405982971,
      "learning_rate": 7.09778809935055e-05,
      "loss": 1.0837,
      "step": 43980
    },
    {
      "epoch": 1.782378801077774,
      "grad_norm": 0.5556907653808594,
      "learning_rate": 7.0937251400732e-05,
      "loss": 1.0992,
      "step": 43990
    },
    {
      "epoch": 1.7827839792548774,
      "grad_norm": 0.5216357707977295,
      "learning_rate": 7.089662704746482e-05,
      "loss": 1.0869,
      "step": 44000
    },
    {
      "epoch": 1.7831891574319807,
      "grad_norm": 0.6363734602928162,
      "learning_rate": 7.085600794102783e-05,
      "loss": 1.1095,
      "step": 44010
    },
    {
      "epoch": 1.783594335609084,
      "grad_norm": 0.6781261563301086,
      "learning_rate": 7.081539408874393e-05,
      "loss": 1.0715,
      "step": 44020
    },
    {
      "epoch": 1.7839995137861875,
      "grad_norm": 0.6798776984214783,
      "learning_rate": 7.077478549793515e-05,
      "loss": 1.082,
      "step": 44030
    },
    {
      "epoch": 1.784404691963291,
      "grad_norm": 0.6036969423294067,
      "learning_rate": 7.073418217592242e-05,
      "loss": 1.0934,
      "step": 44040
    },
    {
      "epoch": 1.7848098701403943,
      "grad_norm": 0.5762577056884766,
      "learning_rate": 7.069358413002585e-05,
      "loss": 1.0577,
      "step": 44050
    },
    {
      "epoch": 1.7852150483174976,
      "grad_norm": 0.5598288774490356,
      "learning_rate": 7.06529913675646e-05,
      "loss": 1.0994,
      "step": 44060
    },
    {
      "epoch": 1.785620226494601,
      "grad_norm": 0.639845609664917,
      "learning_rate": 7.061240389585682e-05,
      "loss": 1.1114,
      "step": 44070
    },
    {
      "epoch": 1.7860254046717043,
      "grad_norm": 0.7116144299507141,
      "learning_rate": 7.057182172221967e-05,
      "loss": 1.0829,
      "step": 44080
    },
    {
      "epoch": 1.7864305828488076,
      "grad_norm": 0.6173812747001648,
      "learning_rate": 7.053124485396942e-05,
      "loss": 1.0817,
      "step": 44090
    },
    {
      "epoch": 1.7868357610259111,
      "grad_norm": 0.6736747622489929,
      "learning_rate": 7.049067329842144e-05,
      "loss": 1.0624,
      "step": 44100
    },
    {
      "epoch": 1.7872409392030146,
      "grad_norm": 0.5541752576828003,
      "learning_rate": 7.045010706289003e-05,
      "loss": 1.0815,
      "step": 44110
    },
    {
      "epoch": 1.787646117380118,
      "grad_norm": 0.5496410727500916,
      "learning_rate": 7.040954615468855e-05,
      "loss": 1.057,
      "step": 44120
    },
    {
      "epoch": 1.7880512955572212,
      "grad_norm": 0.6310923099517822,
      "learning_rate": 7.036899058112947e-05,
      "loss": 1.0828,
      "step": 44130
    },
    {
      "epoch": 1.7884564737343247,
      "grad_norm": 0.6157212853431702,
      "learning_rate": 7.032844034952421e-05,
      "loss": 1.1068,
      "step": 44140
    },
    {
      "epoch": 1.7888616519114282,
      "grad_norm": 0.6041539907455444,
      "learning_rate": 7.028789546718326e-05,
      "loss": 1.0975,
      "step": 44150
    },
    {
      "epoch": 1.7892668300885313,
      "grad_norm": 0.6385085582733154,
      "learning_rate": 7.02473559414162e-05,
      "loss": 1.0861,
      "step": 44160
    },
    {
      "epoch": 1.7896720082656348,
      "grad_norm": 0.6585001945495605,
      "learning_rate": 7.02068217795316e-05,
      "loss": 1.0766,
      "step": 44170
    },
    {
      "epoch": 1.7900771864427383,
      "grad_norm": 0.5684599876403809,
      "learning_rate": 7.016629298883704e-05,
      "loss": 1.1036,
      "step": 44180
    },
    {
      "epoch": 1.7904823646198416,
      "grad_norm": 0.6298700571060181,
      "learning_rate": 7.012576957663916e-05,
      "loss": 1.0744,
      "step": 44190
    },
    {
      "epoch": 1.7908875427969448,
      "grad_norm": 0.5893180966377258,
      "learning_rate": 7.008525155024361e-05,
      "loss": 1.0676,
      "step": 44200
    },
    {
      "epoch": 1.7912927209740483,
      "grad_norm": 0.6228063702583313,
      "learning_rate": 7.004473891695506e-05,
      "loss": 1.0909,
      "step": 44210
    },
    {
      "epoch": 1.7916978991511519,
      "grad_norm": 0.646539568901062,
      "learning_rate": 7.00042316840773e-05,
      "loss": 1.0645,
      "step": 44220
    },
    {
      "epoch": 1.7921030773282551,
      "grad_norm": 0.6576654314994812,
      "learning_rate": 6.996372985891305e-05,
      "loss": 1.0746,
      "step": 44230
    },
    {
      "epoch": 1.7925082555053584,
      "grad_norm": 0.6219380497932434,
      "learning_rate": 6.992323344876405e-05,
      "loss": 1.0797,
      "step": 44240
    },
    {
      "epoch": 1.792913433682462,
      "grad_norm": 0.7096465229988098,
      "learning_rate": 6.988274246093113e-05,
      "loss": 1.044,
      "step": 44250
    },
    {
      "epoch": 1.7933186118595652,
      "grad_norm": 0.6157895922660828,
      "learning_rate": 6.984225690271405e-05,
      "loss": 1.0764,
      "step": 44260
    },
    {
      "epoch": 1.7937237900366685,
      "grad_norm": 0.5979365110397339,
      "learning_rate": 6.980177678141175e-05,
      "loss": 1.0976,
      "step": 44270
    },
    {
      "epoch": 1.794128968213772,
      "grad_norm": 0.5950358510017395,
      "learning_rate": 6.976130210432204e-05,
      "loss": 1.1041,
      "step": 44280
    },
    {
      "epoch": 1.7945341463908755,
      "grad_norm": 0.5818246603012085,
      "learning_rate": 6.972083287874177e-05,
      "loss": 1.0444,
      "step": 44290
    },
    {
      "epoch": 1.7949393245679788,
      "grad_norm": 0.6101977229118347,
      "learning_rate": 6.968036911196687e-05,
      "loss": 1.0514,
      "step": 44300
    },
    {
      "epoch": 1.795344502745082,
      "grad_norm": 0.5694460868835449,
      "learning_rate": 6.963991081129223e-05,
      "loss": 1.1204,
      "step": 44310
    },
    {
      "epoch": 1.7957496809221856,
      "grad_norm": 0.6903136968612671,
      "learning_rate": 6.959945798401177e-05,
      "loss": 1.0696,
      "step": 44320
    },
    {
      "epoch": 1.796154859099289,
      "grad_norm": 0.6616891026496887,
      "learning_rate": 6.955901063741848e-05,
      "loss": 1.0429,
      "step": 44330
    },
    {
      "epoch": 1.7965600372763921,
      "grad_norm": 0.7091642618179321,
      "learning_rate": 6.951856877880427e-05,
      "loss": 1.1274,
      "step": 44340
    },
    {
      "epoch": 1.7969652154534956,
      "grad_norm": 0.6450273394584656,
      "learning_rate": 6.947813241546008e-05,
      "loss": 1.1283,
      "step": 44350
    },
    {
      "epoch": 1.7973703936305991,
      "grad_norm": 0.6492212414741516,
      "learning_rate": 6.943770155467593e-05,
      "loss": 1.0709,
      "step": 44360
    },
    {
      "epoch": 1.7977755718077024,
      "grad_norm": 0.7310017347335815,
      "learning_rate": 6.939727620374077e-05,
      "loss": 1.0815,
      "step": 44370
    },
    {
      "epoch": 1.7981807499848057,
      "grad_norm": 0.623670756816864,
      "learning_rate": 6.935685636994257e-05,
      "loss": 1.1051,
      "step": 44380
    },
    {
      "epoch": 1.7985859281619092,
      "grad_norm": 0.6074561476707458,
      "learning_rate": 6.931644206056835e-05,
      "loss": 1.0775,
      "step": 44390
    },
    {
      "epoch": 1.7989911063390127,
      "grad_norm": 0.6005268096923828,
      "learning_rate": 6.927603328290411e-05,
      "loss": 1.0669,
      "step": 44400
    },
    {
      "epoch": 1.799396284516116,
      "grad_norm": 0.8310745358467102,
      "learning_rate": 6.923563004423484e-05,
      "loss": 1.0652,
      "step": 44410
    },
    {
      "epoch": 1.7998014626932193,
      "grad_norm": 0.5809043645858765,
      "learning_rate": 6.919523235184452e-05,
      "loss": 1.0465,
      "step": 44420
    },
    {
      "epoch": 1.8002066408703228,
      "grad_norm": 0.5754411220550537,
      "learning_rate": 6.915484021301613e-05,
      "loss": 1.0653,
      "step": 44430
    },
    {
      "epoch": 1.800611819047426,
      "grad_norm": 1.0274511575698853,
      "learning_rate": 6.911445363503172e-05,
      "loss": 1.0779,
      "step": 44440
    },
    {
      "epoch": 1.8010169972245293,
      "grad_norm": 0.5630729794502258,
      "learning_rate": 6.90740726251723e-05,
      "loss": 1.0656,
      "step": 44450
    },
    {
      "epoch": 1.8014221754016329,
      "grad_norm": 0.5680816769599915,
      "learning_rate": 6.90336971907178e-05,
      "loss": 1.1276,
      "step": 44460
    },
    {
      "epoch": 1.8018273535787364,
      "grad_norm": 0.5740541815757751,
      "learning_rate": 6.899332733894722e-05,
      "loss": 1.0574,
      "step": 44470
    },
    {
      "epoch": 1.8022325317558396,
      "grad_norm": 0.5940175652503967,
      "learning_rate": 6.895296307713856e-05,
      "loss": 1.0441,
      "step": 44480
    },
    {
      "epoch": 1.802637709932943,
      "grad_norm": 0.66058748960495,
      "learning_rate": 6.891260441256875e-05,
      "loss": 1.0605,
      "step": 44490
    },
    {
      "epoch": 1.8030428881100464,
      "grad_norm": 0.6275665760040283,
      "learning_rate": 6.887225135251381e-05,
      "loss": 1.0898,
      "step": 44500
    },
    {
      "epoch": 1.80344806628715,
      "grad_norm": 0.620156466960907,
      "learning_rate": 6.883190390424867e-05,
      "loss": 1.0645,
      "step": 44510
    },
    {
      "epoch": 1.803853244464253,
      "grad_norm": 0.6047785878181458,
      "learning_rate": 6.879156207504729e-05,
      "loss": 1.0628,
      "step": 44520
    },
    {
      "epoch": 1.8042584226413565,
      "grad_norm": 0.6813251376152039,
      "learning_rate": 6.875122587218255e-05,
      "loss": 1.0974,
      "step": 44530
    },
    {
      "epoch": 1.80466360081846,
      "grad_norm": 0.6084280014038086,
      "learning_rate": 6.871089530292636e-05,
      "loss": 1.0688,
      "step": 44540
    },
    {
      "epoch": 1.8050687789955633,
      "grad_norm": 0.5863734483718872,
      "learning_rate": 6.867057037454969e-05,
      "loss": 1.0554,
      "step": 44550
    },
    {
      "epoch": 1.8054739571726666,
      "grad_norm": 0.5725157260894775,
      "learning_rate": 6.863025109432238e-05,
      "loss": 1.0743,
      "step": 44560
    },
    {
      "epoch": 1.80587913534977,
      "grad_norm": 0.6385300755500793,
      "learning_rate": 6.858993746951328e-05,
      "loss": 1.0726,
      "step": 44570
    },
    {
      "epoch": 1.8062843135268736,
      "grad_norm": 0.6154051423072815,
      "learning_rate": 6.854962950739025e-05,
      "loss": 1.1265,
      "step": 44580
    },
    {
      "epoch": 1.8066894917039769,
      "grad_norm": 0.6082088351249695,
      "learning_rate": 6.850932721522009e-05,
      "loss": 1.0805,
      "step": 44590
    },
    {
      "epoch": 1.8070946698810801,
      "grad_norm": 0.6077187061309814,
      "learning_rate": 6.84690306002686e-05,
      "loss": 1.0545,
      "step": 44600
    },
    {
      "epoch": 1.8074998480581836,
      "grad_norm": 0.62650465965271,
      "learning_rate": 6.84287396698006e-05,
      "loss": 1.0734,
      "step": 44610
    },
    {
      "epoch": 1.807905026235287,
      "grad_norm": 0.6332207322120667,
      "learning_rate": 6.838845443107981e-05,
      "loss": 1.086,
      "step": 44620
    },
    {
      "epoch": 1.8083102044123902,
      "grad_norm": 0.571375846862793,
      "learning_rate": 6.834817489136898e-05,
      "loss": 1.0687,
      "step": 44630
    },
    {
      "epoch": 1.8087153825894937,
      "grad_norm": 0.5459672212600708,
      "learning_rate": 6.830790105792973e-05,
      "loss": 1.0732,
      "step": 44640
    },
    {
      "epoch": 1.8091205607665972,
      "grad_norm": 0.6983458995819092,
      "learning_rate": 6.826763293802282e-05,
      "loss": 1.0901,
      "step": 44650
    },
    {
      "epoch": 1.8095257389437005,
      "grad_norm": 0.6485315561294556,
      "learning_rate": 6.822737053890785e-05,
      "loss": 1.0725,
      "step": 44660
    },
    {
      "epoch": 1.8099309171208038,
      "grad_norm": 0.6755208969116211,
      "learning_rate": 6.818711386784345e-05,
      "loss": 1.0905,
      "step": 44670
    },
    {
      "epoch": 1.8103360952979073,
      "grad_norm": 0.6264453530311584,
      "learning_rate": 6.814686293208714e-05,
      "loss": 1.0884,
      "step": 44680
    },
    {
      "epoch": 1.8107412734750108,
      "grad_norm": 0.6265476942062378,
      "learning_rate": 6.810661773889553e-05,
      "loss": 1.1184,
      "step": 44690
    },
    {
      "epoch": 1.8111464516521139,
      "grad_norm": 0.6697673797607422,
      "learning_rate": 6.806637829552409e-05,
      "loss": 1.0694,
      "step": 44700
    },
    {
      "epoch": 1.8115516298292174,
      "grad_norm": 0.5995756983757019,
      "learning_rate": 6.802614460922728e-05,
      "loss": 1.1149,
      "step": 44710
    },
    {
      "epoch": 1.8119568080063209,
      "grad_norm": 0.6760508418083191,
      "learning_rate": 6.798591668725855e-05,
      "loss": 1.0876,
      "step": 44720
    },
    {
      "epoch": 1.8123619861834241,
      "grad_norm": 0.5963722467422485,
      "learning_rate": 6.794569453687029e-05,
      "loss": 1.1051,
      "step": 44730
    },
    {
      "epoch": 1.8127671643605274,
      "grad_norm": 0.6349040269851685,
      "learning_rate": 6.790547816531387e-05,
      "loss": 1.1156,
      "step": 44740
    },
    {
      "epoch": 1.813172342537631,
      "grad_norm": 0.5527036190032959,
      "learning_rate": 6.786526757983958e-05,
      "loss": 1.0578,
      "step": 44750
    },
    {
      "epoch": 1.8135775207147344,
      "grad_norm": 0.5545629262924194,
      "learning_rate": 6.782506278769667e-05,
      "loss": 1.0537,
      "step": 44760
    },
    {
      "epoch": 1.8139826988918377,
      "grad_norm": 0.5638221502304077,
      "learning_rate": 6.778486379613336e-05,
      "loss": 1.0692,
      "step": 44770
    },
    {
      "epoch": 1.814387877068941,
      "grad_norm": 0.6569054126739502,
      "learning_rate": 6.774467061239687e-05,
      "loss": 1.075,
      "step": 44780
    },
    {
      "epoch": 1.8147930552460445,
      "grad_norm": 0.6804246306419373,
      "learning_rate": 6.77044832437333e-05,
      "loss": 1.0815,
      "step": 44790
    },
    {
      "epoch": 1.8151982334231478,
      "grad_norm": 0.6071582436561584,
      "learning_rate": 6.766430169738774e-05,
      "loss": 1.0839,
      "step": 44800
    },
    {
      "epoch": 1.815603411600251,
      "grad_norm": 0.5721884965896606,
      "learning_rate": 6.762412598060422e-05,
      "loss": 1.0878,
      "step": 44810
    },
    {
      "epoch": 1.8160085897773546,
      "grad_norm": 0.6740723848342896,
      "learning_rate": 6.758395610062573e-05,
      "loss": 1.0843,
      "step": 44820
    },
    {
      "epoch": 1.816413767954458,
      "grad_norm": 0.6246472001075745,
      "learning_rate": 6.754379206469412e-05,
      "loss": 1.0788,
      "step": 44830
    },
    {
      "epoch": 1.8168189461315614,
      "grad_norm": 0.5982715487480164,
      "learning_rate": 6.75036338800504e-05,
      "loss": 1.0622,
      "step": 44840
    },
    {
      "epoch": 1.8172241243086646,
      "grad_norm": 0.7128490805625916,
      "learning_rate": 6.74634815539343e-05,
      "loss": 1.0991,
      "step": 44850
    },
    {
      "epoch": 1.8176293024857682,
      "grad_norm": 0.6357847452163696,
      "learning_rate": 6.742333509358462e-05,
      "loss": 1.0815,
      "step": 44860
    },
    {
      "epoch": 1.8180344806628717,
      "grad_norm": 0.5676137804985046,
      "learning_rate": 6.738319450623904e-05,
      "loss": 1.0446,
      "step": 44870
    },
    {
      "epoch": 1.8184396588399747,
      "grad_norm": 0.5770993828773499,
      "learning_rate": 6.734305979913417e-05,
      "loss": 1.0935,
      "step": 44880
    },
    {
      "epoch": 1.8188448370170782,
      "grad_norm": 0.654691219329834,
      "learning_rate": 6.730293097950571e-05,
      "loss": 1.0595,
      "step": 44890
    },
    {
      "epoch": 1.8192500151941817,
      "grad_norm": 0.600538969039917,
      "learning_rate": 6.72628080545881e-05,
      "loss": 1.1102,
      "step": 44900
    },
    {
      "epoch": 1.819655193371285,
      "grad_norm": 0.6289283037185669,
      "learning_rate": 6.722269103161483e-05,
      "loss": 1.0866,
      "step": 44910
    },
    {
      "epoch": 1.8200603715483883,
      "grad_norm": 0.6126197576522827,
      "learning_rate": 6.718257991781828e-05,
      "loss": 1.0993,
      "step": 44920
    },
    {
      "epoch": 1.8204655497254918,
      "grad_norm": 0.5652665495872498,
      "learning_rate": 6.714247472042982e-05,
      "loss": 1.0825,
      "step": 44930
    },
    {
      "epoch": 1.8208707279025953,
      "grad_norm": 0.6207419037818909,
      "learning_rate": 6.710237544667963e-05,
      "loss": 1.0568,
      "step": 44940
    },
    {
      "epoch": 1.8212759060796986,
      "grad_norm": 0.5753518342971802,
      "learning_rate": 6.706228210379703e-05,
      "loss": 1.0711,
      "step": 44950
    },
    {
      "epoch": 1.8216810842568019,
      "grad_norm": 0.635013997554779,
      "learning_rate": 6.70221946990101e-05,
      "loss": 1.0844,
      "step": 44960
    },
    {
      "epoch": 1.8220862624339054,
      "grad_norm": 0.6305503845214844,
      "learning_rate": 6.698211323954589e-05,
      "loss": 1.0739,
      "step": 44970
    },
    {
      "epoch": 1.8224914406110087,
      "grad_norm": 0.6183366775512695,
      "learning_rate": 6.694203773263034e-05,
      "loss": 1.1032,
      "step": 44980
    },
    {
      "epoch": 1.822896618788112,
      "grad_norm": 0.63092440366745,
      "learning_rate": 6.690196818548846e-05,
      "loss": 1.0723,
      "step": 44990
    },
    {
      "epoch": 1.8233017969652154,
      "grad_norm": 0.6244257688522339,
      "learning_rate": 6.686190460534402e-05,
      "loss": 1.0528,
      "step": 45000
    },
    {
      "epoch": 1.823706975142319,
      "grad_norm": 0.6372984051704407,
      "learning_rate": 6.682184699941984e-05,
      "loss": 1.0537,
      "step": 45010
    },
    {
      "epoch": 1.8241121533194222,
      "grad_norm": 0.6497765183448792,
      "learning_rate": 6.678179537493754e-05,
      "loss": 1.0956,
      "step": 45020
    },
    {
      "epoch": 1.8245173314965255,
      "grad_norm": 0.6502107381820679,
      "learning_rate": 6.674174973911778e-05,
      "loss": 1.0403,
      "step": 45030
    },
    {
      "epoch": 1.824922509673629,
      "grad_norm": 0.6743943691253662,
      "learning_rate": 6.67017100991801e-05,
      "loss": 1.0794,
      "step": 45040
    },
    {
      "epoch": 1.8253276878507325,
      "grad_norm": 0.5960996150970459,
      "learning_rate": 6.666167646234286e-05,
      "loss": 1.0699,
      "step": 45050
    },
    {
      "epoch": 1.8257328660278356,
      "grad_norm": 0.5895029306411743,
      "learning_rate": 6.662164883582354e-05,
      "loss": 1.0862,
      "step": 45060
    },
    {
      "epoch": 1.826138044204939,
      "grad_norm": 0.5468940734863281,
      "learning_rate": 6.658162722683836e-05,
      "loss": 1.108,
      "step": 45070
    },
    {
      "epoch": 1.8265432223820426,
      "grad_norm": 0.6110679507255554,
      "learning_rate": 6.654161164260253e-05,
      "loss": 1.0903,
      "step": 45080
    },
    {
      "epoch": 1.8269484005591459,
      "grad_norm": 0.5472179055213928,
      "learning_rate": 6.650160209033017e-05,
      "loss": 1.0709,
      "step": 45090
    },
    {
      "epoch": 1.8273535787362492,
      "grad_norm": 0.6353256106376648,
      "learning_rate": 6.646159857723427e-05,
      "loss": 1.0875,
      "step": 45100
    },
    {
      "epoch": 1.8277587569133527,
      "grad_norm": 0.6040241122245789,
      "learning_rate": 6.642160111052677e-05,
      "loss": 1.0563,
      "step": 45110
    },
    {
      "epoch": 1.8281639350904562,
      "grad_norm": 0.5715081691741943,
      "learning_rate": 6.638160969741857e-05,
      "loss": 1.0634,
      "step": 45120
    },
    {
      "epoch": 1.8285691132675594,
      "grad_norm": 0.6174970865249634,
      "learning_rate": 6.63416243451194e-05,
      "loss": 1.0873,
      "step": 45130
    },
    {
      "epoch": 1.8289742914446627,
      "grad_norm": 0.6156062483787537,
      "learning_rate": 6.630164506083788e-05,
      "loss": 1.0573,
      "step": 45140
    },
    {
      "epoch": 1.8293794696217662,
      "grad_norm": 0.5615379214286804,
      "learning_rate": 6.626167185178163e-05,
      "loss": 1.0482,
      "step": 45150
    },
    {
      "epoch": 1.8297846477988695,
      "grad_norm": 0.6470584869384766,
      "learning_rate": 6.622170472515704e-05,
      "loss": 1.0999,
      "step": 45160
    },
    {
      "epoch": 1.8301898259759728,
      "grad_norm": 0.5713909268379211,
      "learning_rate": 6.61817436881696e-05,
      "loss": 1.0408,
      "step": 45170
    },
    {
      "epoch": 1.8305950041530763,
      "grad_norm": 0.6763114929199219,
      "learning_rate": 6.614178874802352e-05,
      "loss": 1.0578,
      "step": 45180
    },
    {
      "epoch": 1.8310001823301798,
      "grad_norm": 0.5583401322364807,
      "learning_rate": 6.610183991192199e-05,
      "loss": 1.0616,
      "step": 45190
    },
    {
      "epoch": 1.831405360507283,
      "grad_norm": 0.6045337319374084,
      "learning_rate": 6.60618971870671e-05,
      "loss": 1.1167,
      "step": 45200
    },
    {
      "epoch": 1.8318105386843864,
      "grad_norm": 0.6106401085853577,
      "learning_rate": 6.60219605806598e-05,
      "loss": 1.0651,
      "step": 45210
    },
    {
      "epoch": 1.8322157168614899,
      "grad_norm": 0.7472437620162964,
      "learning_rate": 6.598203009989994e-05,
      "loss": 1.0908,
      "step": 45220
    },
    {
      "epoch": 1.8326208950385934,
      "grad_norm": 0.7096888422966003,
      "learning_rate": 6.594210575198636e-05,
      "loss": 1.1138,
      "step": 45230
    },
    {
      "epoch": 1.8330260732156964,
      "grad_norm": 0.6444381475448608,
      "learning_rate": 6.59021875441167e-05,
      "loss": 1.0815,
      "step": 45240
    },
    {
      "epoch": 1.8334312513928,
      "grad_norm": 0.6761157512664795,
      "learning_rate": 6.58622754834875e-05,
      "loss": 1.0621,
      "step": 45250
    },
    {
      "epoch": 1.8338364295699034,
      "grad_norm": 0.6878507137298584,
      "learning_rate": 6.582236957729422e-05,
      "loss": 1.1094,
      "step": 45260
    },
    {
      "epoch": 1.8342416077470067,
      "grad_norm": 0.62828129529953,
      "learning_rate": 6.578246983273118e-05,
      "loss": 1.0887,
      "step": 45270
    },
    {
      "epoch": 1.83464678592411,
      "grad_norm": 0.6259804964065552,
      "learning_rate": 6.574257625699157e-05,
      "loss": 1.0564,
      "step": 45280
    },
    {
      "epoch": 1.8350519641012135,
      "grad_norm": 0.6140628457069397,
      "learning_rate": 6.57026888572676e-05,
      "loss": 1.1113,
      "step": 45290
    },
    {
      "epoch": 1.835457142278317,
      "grad_norm": 0.5958295464515686,
      "learning_rate": 6.566280764075024e-05,
      "loss": 1.0767,
      "step": 45300
    },
    {
      "epoch": 1.8358623204554203,
      "grad_norm": 0.5905841588973999,
      "learning_rate": 6.562293261462935e-05,
      "loss": 1.0768,
      "step": 45310
    },
    {
      "epoch": 1.8362674986325236,
      "grad_norm": 0.5721420645713806,
      "learning_rate": 6.558306378609369e-05,
      "loss": 1.0706,
      "step": 45320
    },
    {
      "epoch": 1.836672676809627,
      "grad_norm": 0.6186130046844482,
      "learning_rate": 6.554320116233095e-05,
      "loss": 1.0875,
      "step": 45330
    },
    {
      "epoch": 1.8370778549867304,
      "grad_norm": 0.5949694514274597,
      "learning_rate": 6.550334475052767e-05,
      "loss": 1.042,
      "step": 45340
    },
    {
      "epoch": 1.8374830331638337,
      "grad_norm": 0.5867165923118591,
      "learning_rate": 6.546349455786926e-05,
      "loss": 1.0668,
      "step": 45350
    },
    {
      "epoch": 1.8378882113409372,
      "grad_norm": 0.6452912092208862,
      "learning_rate": 6.542365059153996e-05,
      "loss": 1.0764,
      "step": 45360
    },
    {
      "epoch": 1.8382933895180407,
      "grad_norm": 0.6266553401947021,
      "learning_rate": 6.538381285872302e-05,
      "loss": 1.0823,
      "step": 45370
    },
    {
      "epoch": 1.838698567695144,
      "grad_norm": 0.6432468891143799,
      "learning_rate": 6.534398136660045e-05,
      "loss": 1.053,
      "step": 45380
    },
    {
      "epoch": 1.8391037458722472,
      "grad_norm": 0.6723993420600891,
      "learning_rate": 6.530415612235316e-05,
      "loss": 1.0583,
      "step": 45390
    },
    {
      "epoch": 1.8395089240493507,
      "grad_norm": 0.6069017052650452,
      "learning_rate": 6.526433713316095e-05,
      "loss": 1.0905,
      "step": 45400
    },
    {
      "epoch": 1.8399141022264542,
      "grad_norm": 0.6103132963180542,
      "learning_rate": 6.522452440620254e-05,
      "loss": 1.0579,
      "step": 45410
    },
    {
      "epoch": 1.8403192804035573,
      "grad_norm": 0.5918545126914978,
      "learning_rate": 6.518471794865542e-05,
      "loss": 1.0908,
      "step": 45420
    },
    {
      "epoch": 1.8407244585806608,
      "grad_norm": 0.6256740689277649,
      "learning_rate": 6.5144917767696e-05,
      "loss": 1.0757,
      "step": 45430
    },
    {
      "epoch": 1.8411296367577643,
      "grad_norm": 0.6480696201324463,
      "learning_rate": 6.510512387049959e-05,
      "loss": 1.0893,
      "step": 45440
    },
    {
      "epoch": 1.8415348149348676,
      "grad_norm": 0.5501043796539307,
      "learning_rate": 6.506533626424025e-05,
      "loss": 1.0502,
      "step": 45450
    },
    {
      "epoch": 1.8419399931119709,
      "grad_norm": 0.6220190525054932,
      "learning_rate": 6.50255549560911e-05,
      "loss": 1.0876,
      "step": 45460
    },
    {
      "epoch": 1.8423451712890744,
      "grad_norm": 0.6318048238754272,
      "learning_rate": 6.4985779953224e-05,
      "loss": 1.0504,
      "step": 45470
    },
    {
      "epoch": 1.8427503494661779,
      "grad_norm": 0.5425092577934265,
      "learning_rate": 6.494601126280963e-05,
      "loss": 1.0529,
      "step": 45480
    },
    {
      "epoch": 1.8431555276432812,
      "grad_norm": 0.5990374684333801,
      "learning_rate": 6.490624889201762e-05,
      "loss": 1.1127,
      "step": 45490
    },
    {
      "epoch": 1.8435607058203844,
      "grad_norm": 0.5437769293785095,
      "learning_rate": 6.48664928480164e-05,
      "loss": 1.06,
      "step": 45500
    },
    {
      "epoch": 1.843965883997488,
      "grad_norm": 0.6215543150901794,
      "learning_rate": 6.482674313797335e-05,
      "loss": 1.078,
      "step": 45510
    },
    {
      "epoch": 1.8443710621745912,
      "grad_norm": 0.6219103932380676,
      "learning_rate": 6.478699976905463e-05,
      "loss": 1.0578,
      "step": 45520
    },
    {
      "epoch": 1.8447762403516945,
      "grad_norm": 0.6462129354476929,
      "learning_rate": 6.474726274842527e-05,
      "loss": 1.0744,
      "step": 45530
    },
    {
      "epoch": 1.845181418528798,
      "grad_norm": 0.667094886302948,
      "learning_rate": 6.470753208324914e-05,
      "loss": 1.0704,
      "step": 45540
    },
    {
      "epoch": 1.8455865967059015,
      "grad_norm": 0.586952805519104,
      "learning_rate": 6.466780778068903e-05,
      "loss": 1.0671,
      "step": 45550
    },
    {
      "epoch": 1.8459917748830048,
      "grad_norm": 0.6948442459106445,
      "learning_rate": 6.462808984790645e-05,
      "loss": 1.049,
      "step": 45560
    },
    {
      "epoch": 1.846396953060108,
      "grad_norm": 0.5471882224082947,
      "learning_rate": 6.458837829206197e-05,
      "loss": 1.0412,
      "step": 45570
    },
    {
      "epoch": 1.8468021312372116,
      "grad_norm": 0.6495167016983032,
      "learning_rate": 6.454867312031481e-05,
      "loss": 1.0397,
      "step": 45580
    },
    {
      "epoch": 1.847207309414315,
      "grad_norm": 0.6566828489303589,
      "learning_rate": 6.450897433982316e-05,
      "loss": 1.0631,
      "step": 45590
    },
    {
      "epoch": 1.8476124875914182,
      "grad_norm": 0.5451197028160095,
      "learning_rate": 6.4469281957744e-05,
      "loss": 1.0647,
      "step": 45600
    },
    {
      "epoch": 1.8480176657685217,
      "grad_norm": 0.633009672164917,
      "learning_rate": 6.442959598123313e-05,
      "loss": 1.0908,
      "step": 45610
    },
    {
      "epoch": 1.8484228439456252,
      "grad_norm": 0.5575665235519409,
      "learning_rate": 6.43899164174453e-05,
      "loss": 1.0903,
      "step": 45620
    },
    {
      "epoch": 1.8488280221227285,
      "grad_norm": 0.6396011710166931,
      "learning_rate": 6.435024327353405e-05,
      "loss": 1.0766,
      "step": 45630
    },
    {
      "epoch": 1.8492332002998317,
      "grad_norm": 0.5785771608352661,
      "learning_rate": 6.431057655665172e-05,
      "loss": 1.0606,
      "step": 45640
    },
    {
      "epoch": 1.8496383784769352,
      "grad_norm": 0.6331334710121155,
      "learning_rate": 6.427091627394952e-05,
      "loss": 1.104,
      "step": 45650
    },
    {
      "epoch": 1.8500435566540387,
      "grad_norm": 0.5797024369239807,
      "learning_rate": 6.423126243257753e-05,
      "loss": 1.093,
      "step": 45660
    },
    {
      "epoch": 1.850448734831142,
      "grad_norm": 0.6205481290817261,
      "learning_rate": 6.419161503968464e-05,
      "loss": 1.1003,
      "step": 45670
    },
    {
      "epoch": 1.8508539130082453,
      "grad_norm": 0.6995851397514343,
      "learning_rate": 6.41519741024186e-05,
      "loss": 1.0729,
      "step": 45680
    },
    {
      "epoch": 1.8512590911853488,
      "grad_norm": 0.6271584630012512,
      "learning_rate": 6.411233962792593e-05,
      "loss": 1.0293,
      "step": 45690
    },
    {
      "epoch": 1.851664269362452,
      "grad_norm": 0.5488691329956055,
      "learning_rate": 6.407271162335209e-05,
      "loss": 1.0565,
      "step": 45700
    },
    {
      "epoch": 1.8520694475395554,
      "grad_norm": 0.5814899206161499,
      "learning_rate": 6.403309009584128e-05,
      "loss": 1.0989,
      "step": 45710
    },
    {
      "epoch": 1.8524746257166589,
      "grad_norm": 0.6157227754592896,
      "learning_rate": 6.39934750525366e-05,
      "loss": 1.0431,
      "step": 45720
    },
    {
      "epoch": 1.8528798038937624,
      "grad_norm": 0.7448264360427856,
      "learning_rate": 6.395386650057988e-05,
      "loss": 1.0568,
      "step": 45730
    },
    {
      "epoch": 1.8532849820708657,
      "grad_norm": 0.5336378216743469,
      "learning_rate": 6.391426444711195e-05,
      "loss": 1.0842,
      "step": 45740
    },
    {
      "epoch": 1.853690160247969,
      "grad_norm": 0.559129536151886,
      "learning_rate": 6.387466889927232e-05,
      "loss": 1.0567,
      "step": 45750
    },
    {
      "epoch": 1.8540953384250725,
      "grad_norm": 0.5762572884559631,
      "learning_rate": 6.383507986419939e-05,
      "loss": 1.0492,
      "step": 45760
    },
    {
      "epoch": 1.854500516602176,
      "grad_norm": 0.6198599338531494,
      "learning_rate": 6.379549734903036e-05,
      "loss": 1.0696,
      "step": 45770
    },
    {
      "epoch": 1.854905694779279,
      "grad_norm": 0.6085919737815857,
      "learning_rate": 6.375592136090122e-05,
      "loss": 1.0784,
      "step": 45780
    },
    {
      "epoch": 1.8553108729563825,
      "grad_norm": 0.6155360341072083,
      "learning_rate": 6.371635190694691e-05,
      "loss": 1.1063,
      "step": 45790
    },
    {
      "epoch": 1.855716051133486,
      "grad_norm": 0.6900031566619873,
      "learning_rate": 6.367678899430109e-05,
      "loss": 1.0523,
      "step": 45800
    },
    {
      "epoch": 1.8561212293105893,
      "grad_norm": 0.5811262726783752,
      "learning_rate": 6.363723263009627e-05,
      "loss": 1.0837,
      "step": 45810
    },
    {
      "epoch": 1.8565264074876926,
      "grad_norm": 0.6132071614265442,
      "learning_rate": 6.359768282146373e-05,
      "loss": 1.0771,
      "step": 45820
    },
    {
      "epoch": 1.856931585664796,
      "grad_norm": 0.6044891476631165,
      "learning_rate": 6.355813957553364e-05,
      "loss": 1.0499,
      "step": 45830
    },
    {
      "epoch": 1.8573367638418996,
      "grad_norm": 0.6300112009048462,
      "learning_rate": 6.351860289943491e-05,
      "loss": 1.0906,
      "step": 45840
    },
    {
      "epoch": 1.857741942019003,
      "grad_norm": 0.655570924282074,
      "learning_rate": 6.34790728002954e-05,
      "loss": 1.0858,
      "step": 45850
    },
    {
      "epoch": 1.8581471201961062,
      "grad_norm": 0.5960208773612976,
      "learning_rate": 6.343954928524165e-05,
      "loss": 1.0723,
      "step": 45860
    },
    {
      "epoch": 1.8585522983732097,
      "grad_norm": 0.5926869511604309,
      "learning_rate": 6.340003236139907e-05,
      "loss": 1.0795,
      "step": 45870
    },
    {
      "epoch": 1.858957476550313,
      "grad_norm": 0.6827091574668884,
      "learning_rate": 6.336052203589184e-05,
      "loss": 1.0968,
      "step": 45880
    },
    {
      "epoch": 1.8593626547274162,
      "grad_norm": 0.5719303488731384,
      "learning_rate": 6.332101831584302e-05,
      "loss": 1.0579,
      "step": 45890
    },
    {
      "epoch": 1.8597678329045197,
      "grad_norm": 0.6661895513534546,
      "learning_rate": 6.328152120837439e-05,
      "loss": 1.0644,
      "step": 45900
    },
    {
      "epoch": 1.8601730110816233,
      "grad_norm": 0.6107373833656311,
      "learning_rate": 6.324203072060666e-05,
      "loss": 1.0877,
      "step": 45910
    },
    {
      "epoch": 1.8605781892587265,
      "grad_norm": 0.6682655811309814,
      "learning_rate": 6.320254685965922e-05,
      "loss": 1.0508,
      "step": 45920
    },
    {
      "epoch": 1.8609833674358298,
      "grad_norm": 0.6137130260467529,
      "learning_rate": 6.316306963265037e-05,
      "loss": 1.076,
      "step": 45930
    },
    {
      "epoch": 1.8613885456129333,
      "grad_norm": 0.6476976871490479,
      "learning_rate": 6.312359904669713e-05,
      "loss": 1.0973,
      "step": 45940
    },
    {
      "epoch": 1.8617937237900368,
      "grad_norm": 0.577832818031311,
      "learning_rate": 6.308413510891533e-05,
      "loss": 1.0805,
      "step": 45950
    },
    {
      "epoch": 1.8621989019671399,
      "grad_norm": 0.5513951778411865,
      "learning_rate": 6.30446778264197e-05,
      "loss": 1.0983,
      "step": 45960
    },
    {
      "epoch": 1.8626040801442434,
      "grad_norm": 0.6097855567932129,
      "learning_rate": 6.300522720632367e-05,
      "loss": 1.0685,
      "step": 45970
    },
    {
      "epoch": 1.863009258321347,
      "grad_norm": 0.6060397028923035,
      "learning_rate": 6.296578325573947e-05,
      "loss": 1.1093,
      "step": 45980
    },
    {
      "epoch": 1.8634144364984502,
      "grad_norm": 0.6573516726493835,
      "learning_rate": 6.292634598177818e-05,
      "loss": 1.0917,
      "step": 45990
    },
    {
      "epoch": 1.8638196146755535,
      "grad_norm": 0.6973974108695984,
      "learning_rate": 6.288691539154968e-05,
      "loss": 1.0716,
      "step": 46000
    },
    {
      "epoch": 1.864224792852657,
      "grad_norm": 0.5905716419219971,
      "learning_rate": 6.284749149216254e-05,
      "loss": 1.1057,
      "step": 46010
    },
    {
      "epoch": 1.8646299710297605,
      "grad_norm": 0.5916700959205627,
      "learning_rate": 6.280807429072427e-05,
      "loss": 1.1116,
      "step": 46020
    },
    {
      "epoch": 1.8650351492068638,
      "grad_norm": 0.6381468772888184,
      "learning_rate": 6.276866379434107e-05,
      "loss": 1.1087,
      "step": 46030
    },
    {
      "epoch": 1.865440327383967,
      "grad_norm": 0.5983758568763733,
      "learning_rate": 6.2729260010118e-05,
      "loss": 1.098,
      "step": 46040
    },
    {
      "epoch": 1.8658455055610705,
      "grad_norm": 0.6433835029602051,
      "learning_rate": 6.268986294515883e-05,
      "loss": 1.0972,
      "step": 46050
    },
    {
      "epoch": 1.8662506837381738,
      "grad_norm": 0.6390684247016907,
      "learning_rate": 6.265047260656622e-05,
      "loss": 1.0837,
      "step": 46060
    },
    {
      "epoch": 1.866655861915277,
      "grad_norm": 0.68320631980896,
      "learning_rate": 6.261108900144146e-05,
      "loss": 1.0777,
      "step": 46070
    },
    {
      "epoch": 1.8670610400923806,
      "grad_norm": 0.6742766499519348,
      "learning_rate": 6.257171213688483e-05,
      "loss": 1.0648,
      "step": 46080
    },
    {
      "epoch": 1.8674662182694841,
      "grad_norm": 0.6414894461631775,
      "learning_rate": 6.253234201999526e-05,
      "loss": 1.05,
      "step": 46090
    },
    {
      "epoch": 1.8678713964465874,
      "grad_norm": 0.6755492687225342,
      "learning_rate": 6.249297865787048e-05,
      "loss": 1.1041,
      "step": 46100
    },
    {
      "epoch": 1.8682765746236907,
      "grad_norm": 0.5925762057304382,
      "learning_rate": 6.245362205760704e-05,
      "loss": 1.0619,
      "step": 46110
    },
    {
      "epoch": 1.8686817528007942,
      "grad_norm": 0.6959025859832764,
      "learning_rate": 6.241427222630017e-05,
      "loss": 1.0933,
      "step": 46120
    },
    {
      "epoch": 1.8690869309778977,
      "grad_norm": 0.6207369565963745,
      "learning_rate": 6.237492917104407e-05,
      "loss": 1.1094,
      "step": 46130
    },
    {
      "epoch": 1.8694921091550007,
      "grad_norm": 0.6515803933143616,
      "learning_rate": 6.233559289893156e-05,
      "loss": 1.0793,
      "step": 46140
    },
    {
      "epoch": 1.8698972873321043,
      "grad_norm": 0.5986413955688477,
      "learning_rate": 6.229626341705428e-05,
      "loss": 1.0686,
      "step": 46150
    },
    {
      "epoch": 1.8703024655092078,
      "grad_norm": 0.5664257407188416,
      "learning_rate": 6.225694073250266e-05,
      "loss": 1.0463,
      "step": 46160
    },
    {
      "epoch": 1.870707643686311,
      "grad_norm": 0.7484499216079712,
      "learning_rate": 6.221762485236586e-05,
      "loss": 1.1032,
      "step": 46170
    },
    {
      "epoch": 1.8711128218634143,
      "grad_norm": 0.5803465843200684,
      "learning_rate": 6.217831578373185e-05,
      "loss": 1.0893,
      "step": 46180
    },
    {
      "epoch": 1.8715180000405178,
      "grad_norm": 0.6179912686347961,
      "learning_rate": 6.21390135336874e-05,
      "loss": 1.0935,
      "step": 46190
    },
    {
      "epoch": 1.8719231782176213,
      "grad_norm": 0.5833296179771423,
      "learning_rate": 6.209971810931804e-05,
      "loss": 1.0662,
      "step": 46200
    },
    {
      "epoch": 1.8723283563947246,
      "grad_norm": 0.6255828738212585,
      "learning_rate": 6.2060429517708e-05,
      "loss": 1.0554,
      "step": 46210
    },
    {
      "epoch": 1.872733534571828,
      "grad_norm": 0.6271907091140747,
      "learning_rate": 6.202114776594036e-05,
      "loss": 1.1014,
      "step": 46220
    },
    {
      "epoch": 1.8731387127489314,
      "grad_norm": 0.6882660984992981,
      "learning_rate": 6.19818728610969e-05,
      "loss": 1.0866,
      "step": 46230
    },
    {
      "epoch": 1.8735438909260347,
      "grad_norm": 0.614966630935669,
      "learning_rate": 6.19426048102582e-05,
      "loss": 1.092,
      "step": 46240
    },
    {
      "epoch": 1.873949069103138,
      "grad_norm": 0.5487273335456848,
      "learning_rate": 6.190334362050365e-05,
      "loss": 1.0636,
      "step": 46250
    },
    {
      "epoch": 1.8743542472802415,
      "grad_norm": 0.6368963122367859,
      "learning_rate": 6.186408929891134e-05,
      "loss": 1.0993,
      "step": 46260
    },
    {
      "epoch": 1.874759425457345,
      "grad_norm": 0.7144164443016052,
      "learning_rate": 6.182484185255812e-05,
      "loss": 1.078,
      "step": 46270
    },
    {
      "epoch": 1.8751646036344483,
      "grad_norm": 0.6494319438934326,
      "learning_rate": 6.178560128851963e-05,
      "loss": 1.0797,
      "step": 46280
    },
    {
      "epoch": 1.8755697818115515,
      "grad_norm": 0.609017550945282,
      "learning_rate": 6.174636761387023e-05,
      "loss": 1.0694,
      "step": 46290
    },
    {
      "epoch": 1.875974959988655,
      "grad_norm": 0.6556321382522583,
      "learning_rate": 6.170714083568312e-05,
      "loss": 1.0638,
      "step": 46300
    },
    {
      "epoch": 1.8763801381657585,
      "grad_norm": 0.6161240339279175,
      "learning_rate": 6.166792096103019e-05,
      "loss": 1.0748,
      "step": 46310
    },
    {
      "epoch": 1.8767853163428616,
      "grad_norm": 0.649479866027832,
      "learning_rate": 6.162870799698209e-05,
      "loss": 1.0503,
      "step": 46320
    },
    {
      "epoch": 1.8771904945199651,
      "grad_norm": 0.5932403802871704,
      "learning_rate": 6.15895019506082e-05,
      "loss": 1.0999,
      "step": 46330
    },
    {
      "epoch": 1.8775956726970686,
      "grad_norm": 0.6465829014778137,
      "learning_rate": 6.155030282897673e-05,
      "loss": 1.0759,
      "step": 46340
    },
    {
      "epoch": 1.878000850874172,
      "grad_norm": 0.6726570725440979,
      "learning_rate": 6.151111063915458e-05,
      "loss": 1.0988,
      "step": 46350
    },
    {
      "epoch": 1.8784060290512752,
      "grad_norm": 0.6804061532020569,
      "learning_rate": 6.147192538820744e-05,
      "loss": 1.0587,
      "step": 46360
    },
    {
      "epoch": 1.8788112072283787,
      "grad_norm": 0.5957586765289307,
      "learning_rate": 6.143274708319967e-05,
      "loss": 1.0479,
      "step": 46370
    },
    {
      "epoch": 1.8792163854054822,
      "grad_norm": 0.6571007370948792,
      "learning_rate": 6.13935757311945e-05,
      "loss": 1.0843,
      "step": 46380
    },
    {
      "epoch": 1.8796215635825855,
      "grad_norm": 0.5765702724456787,
      "learning_rate": 6.135441133925382e-05,
      "loss": 1.0906,
      "step": 46390
    },
    {
      "epoch": 1.8800267417596888,
      "grad_norm": 0.6416252255439758,
      "learning_rate": 6.131525391443827e-05,
      "loss": 1.0333,
      "step": 46400
    },
    {
      "epoch": 1.8804319199367923,
      "grad_norm": 0.6067430973052979,
      "learning_rate": 6.127610346380724e-05,
      "loss": 1.1058,
      "step": 46410
    },
    {
      "epoch": 1.8808370981138955,
      "grad_norm": 0.6090953350067139,
      "learning_rate": 6.123695999441891e-05,
      "loss": 1.0851,
      "step": 46420
    },
    {
      "epoch": 1.8812422762909988,
      "grad_norm": 0.724753737449646,
      "learning_rate": 6.119782351333018e-05,
      "loss": 1.1163,
      "step": 46430
    },
    {
      "epoch": 1.8816474544681023,
      "grad_norm": 0.6418900489807129,
      "learning_rate": 6.115869402759662e-05,
      "loss": 1.1038,
      "step": 46440
    },
    {
      "epoch": 1.8820526326452058,
      "grad_norm": 0.6722566485404968,
      "learning_rate": 6.111957154427261e-05,
      "loss": 1.0904,
      "step": 46450
    },
    {
      "epoch": 1.8824578108223091,
      "grad_norm": 0.5538800954818726,
      "learning_rate": 6.108045607041125e-05,
      "loss": 1.083,
      "step": 46460
    },
    {
      "epoch": 1.8828629889994124,
      "grad_norm": 0.6468640565872192,
      "learning_rate": 6.104134761306439e-05,
      "loss": 1.0575,
      "step": 46470
    },
    {
      "epoch": 1.883268167176516,
      "grad_norm": 0.6603781580924988,
      "learning_rate": 6.100224617928261e-05,
      "loss": 1.0467,
      "step": 46480
    },
    {
      "epoch": 1.8836733453536194,
      "grad_norm": 0.6412304043769836,
      "learning_rate": 6.0963151776115194e-05,
      "loss": 1.0674,
      "step": 46490
    },
    {
      "epoch": 1.8840785235307225,
      "grad_norm": 0.6772738099098206,
      "learning_rate": 6.092406441061018e-05,
      "loss": 1.1044,
      "step": 46500
    },
    {
      "epoch": 1.884483701707826,
      "grad_norm": 0.6470524668693542,
      "learning_rate": 6.0884984089814366e-05,
      "loss": 1.1033,
      "step": 46510
    },
    {
      "epoch": 1.8848888798849295,
      "grad_norm": 0.6876247525215149,
      "learning_rate": 6.084591082077319e-05,
      "loss": 1.1029,
      "step": 46520
    },
    {
      "epoch": 1.8852940580620328,
      "grad_norm": 0.5797191262245178,
      "learning_rate": 6.0806844610530956e-05,
      "loss": 1.0751,
      "step": 46530
    },
    {
      "epoch": 1.885699236239136,
      "grad_norm": 0.5976259112358093,
      "learning_rate": 6.076778546613059e-05,
      "loss": 1.0878,
      "step": 46540
    },
    {
      "epoch": 1.8861044144162395,
      "grad_norm": 0.6276453137397766,
      "learning_rate": 6.072873339461377e-05,
      "loss": 1.1301,
      "step": 46550
    },
    {
      "epoch": 1.886509592593343,
      "grad_norm": 0.645881175994873,
      "learning_rate": 6.068968840302091e-05,
      "loss": 1.051,
      "step": 46560
    },
    {
      "epoch": 1.8869147707704463,
      "grad_norm": 0.646101176738739,
      "learning_rate": 6.0650650498391084e-05,
      "loss": 1.0754,
      "step": 46570
    },
    {
      "epoch": 1.8873199489475496,
      "grad_norm": 0.60142982006073,
      "learning_rate": 6.061161968776223e-05,
      "loss": 1.0374,
      "step": 46580
    },
    {
      "epoch": 1.8877251271246531,
      "grad_norm": 0.6532179117202759,
      "learning_rate": 6.0572595978170896e-05,
      "loss": 1.0653,
      "step": 46590
    },
    {
      "epoch": 1.8881303053017564,
      "grad_norm": 0.5591765642166138,
      "learning_rate": 6.053357937665237e-05,
      "loss": 1.037,
      "step": 46600
    },
    {
      "epoch": 1.8885354834788597,
      "grad_norm": 0.593299388885498,
      "learning_rate": 6.049456989024066e-05,
      "loss": 1.0842,
      "step": 46610
    },
    {
      "epoch": 1.8889406616559632,
      "grad_norm": 0.6301593780517578,
      "learning_rate": 6.045556752596847e-05,
      "loss": 1.0712,
      "step": 46620
    },
    {
      "epoch": 1.8893458398330667,
      "grad_norm": 0.5782089829444885,
      "learning_rate": 6.041657229086728e-05,
      "loss": 1.064,
      "step": 46630
    },
    {
      "epoch": 1.88975101801017,
      "grad_norm": 0.6314309239387512,
      "learning_rate": 6.037758419196725e-05,
      "loss": 1.0989,
      "step": 46640
    },
    {
      "epoch": 1.8901561961872733,
      "grad_norm": 0.5306643843650818,
      "learning_rate": 6.0338603236297256e-05,
      "loss": 1.0789,
      "step": 46650
    },
    {
      "epoch": 1.8905613743643768,
      "grad_norm": 0.5471155047416687,
      "learning_rate": 6.0299629430884854e-05,
      "loss": 1.0879,
      "step": 46660
    },
    {
      "epoch": 1.8909665525414803,
      "grad_norm": 0.6019802689552307,
      "learning_rate": 6.0260662782756374e-05,
      "loss": 1.0513,
      "step": 46670
    },
    {
      "epoch": 1.8913717307185833,
      "grad_norm": 0.6878660917282104,
      "learning_rate": 6.0221703298936814e-05,
      "loss": 1.0839,
      "step": 46680
    },
    {
      "epoch": 1.8917769088956868,
      "grad_norm": 0.6490032076835632,
      "learning_rate": 6.018275098644987e-05,
      "loss": 1.0825,
      "step": 46690
    },
    {
      "epoch": 1.8921820870727903,
      "grad_norm": 0.610815703868866,
      "learning_rate": 6.014380585231797e-05,
      "loss": 1.112,
      "step": 46700
    },
    {
      "epoch": 1.8925872652498936,
      "grad_norm": 0.5904674530029297,
      "learning_rate": 6.010486790356227e-05,
      "loss": 1.0801,
      "step": 46710
    },
    {
      "epoch": 1.892992443426997,
      "grad_norm": 0.610196053981781,
      "learning_rate": 6.006593714720259e-05,
      "loss": 1.0659,
      "step": 46720
    },
    {
      "epoch": 1.8933976216041004,
      "grad_norm": 0.6438460350036621,
      "learning_rate": 6.002701359025745e-05,
      "loss": 1.0457,
      "step": 46730
    },
    {
      "epoch": 1.893802799781204,
      "grad_norm": 0.5706272721290588,
      "learning_rate": 5.998809723974407e-05,
      "loss": 1.085,
      "step": 46740
    },
    {
      "epoch": 1.8942079779583072,
      "grad_norm": 0.5950167775154114,
      "learning_rate": 5.9949188102678444e-05,
      "loss": 1.0991,
      "step": 46750
    },
    {
      "epoch": 1.8946131561354105,
      "grad_norm": 0.5476616621017456,
      "learning_rate": 5.991028618607519e-05,
      "loss": 1.0625,
      "step": 46760
    },
    {
      "epoch": 1.895018334312514,
      "grad_norm": 0.6264427900314331,
      "learning_rate": 5.987139149694763e-05,
      "loss": 1.0888,
      "step": 46770
    },
    {
      "epoch": 1.8954235124896173,
      "grad_norm": 0.6017371416091919,
      "learning_rate": 5.98325040423078e-05,
      "loss": 1.1068,
      "step": 46780
    },
    {
      "epoch": 1.8958286906667206,
      "grad_norm": 0.6038373112678528,
      "learning_rate": 5.9793623829166447e-05,
      "loss": 1.0733,
      "step": 46790
    },
    {
      "epoch": 1.896233868843824,
      "grad_norm": 0.6217964887619019,
      "learning_rate": 5.975475086453293e-05,
      "loss": 1.1194,
      "step": 46800
    },
    {
      "epoch": 1.8966390470209276,
      "grad_norm": 0.5557763576507568,
      "learning_rate": 5.971588515541546e-05,
      "loss": 1.07,
      "step": 46810
    },
    {
      "epoch": 1.8970442251980308,
      "grad_norm": 0.6526584029197693,
      "learning_rate": 5.96770267088208e-05,
      "loss": 1.0878,
      "step": 46820
    },
    {
      "epoch": 1.8974494033751341,
      "grad_norm": 0.6277955174446106,
      "learning_rate": 5.963817553175447e-05,
      "loss": 1.1048,
      "step": 46830
    },
    {
      "epoch": 1.8978545815522376,
      "grad_norm": 0.6512948274612427,
      "learning_rate": 5.9599331631220636e-05,
      "loss": 1.0808,
      "step": 46840
    },
    {
      "epoch": 1.8982597597293411,
      "grad_norm": 0.6480260491371155,
      "learning_rate": 5.956049501422217e-05,
      "loss": 1.0578,
      "step": 46850
    },
    {
      "epoch": 1.8986649379064442,
      "grad_norm": 0.6487553715705872,
      "learning_rate": 5.952166568776062e-05,
      "loss": 1.104,
      "step": 46860
    },
    {
      "epoch": 1.8990701160835477,
      "grad_norm": 0.6582857370376587,
      "learning_rate": 5.94828436588363e-05,
      "loss": 1.0956,
      "step": 46870
    },
    {
      "epoch": 1.8994752942606512,
      "grad_norm": 0.681060791015625,
      "learning_rate": 5.94440289344481e-05,
      "loss": 1.0776,
      "step": 46880
    },
    {
      "epoch": 1.8998804724377545,
      "grad_norm": 0.5586597919464111,
      "learning_rate": 5.940522152159365e-05,
      "loss": 1.0805,
      "step": 46890
    },
    {
      "epoch": 1.9002856506148578,
      "grad_norm": 0.608357846736908,
      "learning_rate": 5.936642142726925e-05,
      "loss": 1.0808,
      "step": 46900
    },
    {
      "epoch": 1.9006908287919613,
      "grad_norm": 0.5926377177238464,
      "learning_rate": 5.9327628658469834e-05,
      "loss": 1.0525,
      "step": 46910
    },
    {
      "epoch": 1.9010960069690648,
      "grad_norm": 0.6952172517776489,
      "learning_rate": 5.928884322218913e-05,
      "loss": 1.0999,
      "step": 46920
    },
    {
      "epoch": 1.901501185146168,
      "grad_norm": 0.6419894099235535,
      "learning_rate": 5.9250065125419465e-05,
      "loss": 1.0993,
      "step": 46930
    },
    {
      "epoch": 1.9019063633232713,
      "grad_norm": 0.5802138447761536,
      "learning_rate": 5.921129437515182e-05,
      "loss": 1.0521,
      "step": 46940
    },
    {
      "epoch": 1.9023115415003748,
      "grad_norm": 0.6033638119697571,
      "learning_rate": 5.9172530978375894e-05,
      "loss": 1.0644,
      "step": 46950
    },
    {
      "epoch": 1.9027167196774781,
      "grad_norm": 0.6738817095756531,
      "learning_rate": 5.9133774942080054e-05,
      "loss": 1.0364,
      "step": 46960
    },
    {
      "epoch": 1.9031218978545814,
      "grad_norm": 0.5898933410644531,
      "learning_rate": 5.909502627325133e-05,
      "loss": 1.0731,
      "step": 46970
    },
    {
      "epoch": 1.903527076031685,
      "grad_norm": 0.6303744912147522,
      "learning_rate": 5.9056284978875464e-05,
      "loss": 1.0646,
      "step": 46980
    },
    {
      "epoch": 1.9039322542087884,
      "grad_norm": 0.7645599246025085,
      "learning_rate": 5.901755106593681e-05,
      "loss": 1.063,
      "step": 46990
    },
    {
      "epoch": 1.9043374323858917,
      "grad_norm": 0.6548373103141785,
      "learning_rate": 5.897882454141839e-05,
      "loss": 1.0871,
      "step": 47000
    },
    {
      "epoch": 1.904742610562995,
      "grad_norm": 0.6582369208335876,
      "learning_rate": 5.894010541230197e-05,
      "loss": 1.0872,
      "step": 47010
    },
    {
      "epoch": 1.9051477887400985,
      "grad_norm": 0.6742396950721741,
      "learning_rate": 5.890139368556791e-05,
      "loss": 1.0956,
      "step": 47020
    },
    {
      "epoch": 1.905552966917202,
      "grad_norm": 0.664413571357727,
      "learning_rate": 5.886268936819526e-05,
      "loss": 1.0707,
      "step": 47030
    },
    {
      "epoch": 1.905958145094305,
      "grad_norm": 0.5554273128509521,
      "learning_rate": 5.882399246716171e-05,
      "loss": 1.0966,
      "step": 47040
    },
    {
      "epoch": 1.9063633232714086,
      "grad_norm": 0.590790867805481,
      "learning_rate": 5.878530298944369e-05,
      "loss": 1.1087,
      "step": 47050
    },
    {
      "epoch": 1.906768501448512,
      "grad_norm": 0.653457522392273,
      "learning_rate": 5.874662094201619e-05,
      "loss": 1.0865,
      "step": 47060
    },
    {
      "epoch": 1.9071736796256153,
      "grad_norm": 0.6440012454986572,
      "learning_rate": 5.8707946331852946e-05,
      "loss": 1.0634,
      "step": 47070
    },
    {
      "epoch": 1.9075788578027186,
      "grad_norm": 0.5927245020866394,
      "learning_rate": 5.866927916592624e-05,
      "loss": 1.099,
      "step": 47080
    },
    {
      "epoch": 1.9079840359798221,
      "grad_norm": 0.6505098938941956,
      "learning_rate": 5.863061945120719e-05,
      "loss": 1.0689,
      "step": 47090
    },
    {
      "epoch": 1.9083892141569256,
      "grad_norm": 0.6206384301185608,
      "learning_rate": 5.859196719466542e-05,
      "loss": 1.1056,
      "step": 47100
    },
    {
      "epoch": 1.908794392334029,
      "grad_norm": 0.5619335770606995,
      "learning_rate": 5.855332240326925e-05,
      "loss": 1.1004,
      "step": 47110
    },
    {
      "epoch": 1.9091995705111322,
      "grad_norm": 0.6237898468971252,
      "learning_rate": 5.851468508398566e-05,
      "loss": 1.0547,
      "step": 47120
    },
    {
      "epoch": 1.9096047486882357,
      "grad_norm": 0.6375855803489685,
      "learning_rate": 5.84760552437803e-05,
      "loss": 1.0679,
      "step": 47130
    },
    {
      "epoch": 1.910009926865339,
      "grad_norm": 0.6144484281539917,
      "learning_rate": 5.8437432889617405e-05,
      "loss": 1.0706,
      "step": 47140
    },
    {
      "epoch": 1.9104151050424423,
      "grad_norm": 0.454733282327652,
      "learning_rate": 5.839881802845999e-05,
      "loss": 1.1008,
      "step": 47150
    },
    {
      "epoch": 1.9108202832195458,
      "grad_norm": 0.6244305968284607,
      "learning_rate": 5.836021066726962e-05,
      "loss": 1.0879,
      "step": 47160
    },
    {
      "epoch": 1.9112254613966493,
      "grad_norm": 0.6266106367111206,
      "learning_rate": 5.832161081300649e-05,
      "loss": 1.0993,
      "step": 47170
    },
    {
      "epoch": 1.9116306395737526,
      "grad_norm": 0.6272768378257751,
      "learning_rate": 5.828301847262949e-05,
      "loss": 1.0903,
      "step": 47180
    },
    {
      "epoch": 1.9120358177508558,
      "grad_norm": 0.6488293409347534,
      "learning_rate": 5.824443365309615e-05,
      "loss": 1.0649,
      "step": 47190
    },
    {
      "epoch": 1.9124409959279594,
      "grad_norm": 0.6154455542564392,
      "learning_rate": 5.820585636136271e-05,
      "loss": 1.099,
      "step": 47200
    },
    {
      "epoch": 1.9128461741050629,
      "grad_norm": 0.6163362860679626,
      "learning_rate": 5.816728660438386e-05,
      "loss": 1.0971,
      "step": 47210
    },
    {
      "epoch": 1.913251352282166,
      "grad_norm": 0.5960733294487,
      "learning_rate": 5.812872438911313e-05,
      "loss": 1.0865,
      "step": 47220
    },
    {
      "epoch": 1.9136565304592694,
      "grad_norm": 0.6277028918266296,
      "learning_rate": 5.809016972250263e-05,
      "loss": 1.0712,
      "step": 47230
    },
    {
      "epoch": 1.914061708636373,
      "grad_norm": 0.641160249710083,
      "learning_rate": 5.8051622611503034e-05,
      "loss": 1.0646,
      "step": 47240
    },
    {
      "epoch": 1.9144668868134762,
      "grad_norm": 0.6346921920776367,
      "learning_rate": 5.801308306306372e-05,
      "loss": 1.0776,
      "step": 47250
    },
    {
      "epoch": 1.9148720649905795,
      "grad_norm": 0.5990679264068604,
      "learning_rate": 5.7974551084132725e-05,
      "loss": 1.0642,
      "step": 47260
    },
    {
      "epoch": 1.915277243167683,
      "grad_norm": 0.6319941878318787,
      "learning_rate": 5.793602668165673e-05,
      "loss": 1.0775,
      "step": 47270
    },
    {
      "epoch": 1.9156824213447865,
      "grad_norm": 0.6946179270744324,
      "learning_rate": 5.7897509862580925e-05,
      "loss": 1.0815,
      "step": 47280
    },
    {
      "epoch": 1.9160875995218898,
      "grad_norm": 0.6597762703895569,
      "learning_rate": 5.7859000633849306e-05,
      "loss": 1.1016,
      "step": 47290
    },
    {
      "epoch": 1.916492777698993,
      "grad_norm": 0.5943478941917419,
      "learning_rate": 5.782049900240432e-05,
      "loss": 1.0683,
      "step": 47300
    },
    {
      "epoch": 1.9168979558760966,
      "grad_norm": 0.6144667863845825,
      "learning_rate": 5.778200497518719e-05,
      "loss": 1.0725,
      "step": 47310
    },
    {
      "epoch": 1.9173031340531999,
      "grad_norm": 0.6507278680801392,
      "learning_rate": 5.774351855913772e-05,
      "loss": 1.0387,
      "step": 47320
    },
    {
      "epoch": 1.9177083122303031,
      "grad_norm": 0.5961483120918274,
      "learning_rate": 5.770503976119438e-05,
      "loss": 1.1414,
      "step": 47330
    },
    {
      "epoch": 1.9181134904074066,
      "grad_norm": 0.690041720867157,
      "learning_rate": 5.766656858829414e-05,
      "loss": 1.0938,
      "step": 47340
    },
    {
      "epoch": 1.9185186685845101,
      "grad_norm": 0.6624987125396729,
      "learning_rate": 5.762810504737277e-05,
      "loss": 1.0749,
      "step": 47350
    },
    {
      "epoch": 1.9189238467616134,
      "grad_norm": 0.5962397456169128,
      "learning_rate": 5.758964914536444e-05,
      "loss": 1.0895,
      "step": 47360
    },
    {
      "epoch": 1.9193290249387167,
      "grad_norm": 0.5972073674201965,
      "learning_rate": 5.755120088920225e-05,
      "loss": 1.0599,
      "step": 47370
    },
    {
      "epoch": 1.9197342031158202,
      "grad_norm": 0.6265875101089478,
      "learning_rate": 5.751276028581763e-05,
      "loss": 1.074,
      "step": 47380
    },
    {
      "epoch": 1.9201393812929237,
      "grad_norm": 0.5631691217422485,
      "learning_rate": 5.747432734214081e-05,
      "loss": 1.0776,
      "step": 47390
    },
    {
      "epoch": 1.9205445594700268,
      "grad_norm": 0.6515172123908997,
      "learning_rate": 5.743590206510052e-05,
      "loss": 1.1004,
      "step": 47400
    },
    {
      "epoch": 1.9209497376471303,
      "grad_norm": 0.6236814260482788,
      "learning_rate": 5.739748446162425e-05,
      "loss": 1.0746,
      "step": 47410
    },
    {
      "epoch": 1.9213549158242338,
      "grad_norm": 0.6009342074394226,
      "learning_rate": 5.7359074538637894e-05,
      "loss": 1.0589,
      "step": 47420
    },
    {
      "epoch": 1.921760094001337,
      "grad_norm": 0.6462274193763733,
      "learning_rate": 5.732067230306624e-05,
      "loss": 1.0793,
      "step": 47430
    },
    {
      "epoch": 1.9221652721784404,
      "grad_norm": 0.627143383026123,
      "learning_rate": 5.728227776183244e-05,
      "loss": 1.075,
      "step": 47440
    },
    {
      "epoch": 1.9225704503555439,
      "grad_norm": 0.6654924750328064,
      "learning_rate": 5.724389092185844e-05,
      "loss": 1.063,
      "step": 47450
    },
    {
      "epoch": 1.9229756285326474,
      "grad_norm": 0.6190140843391418,
      "learning_rate": 5.720551179006461e-05,
      "loss": 1.053,
      "step": 47460
    },
    {
      "epoch": 1.9233808067097506,
      "grad_norm": 0.5942806601524353,
      "learning_rate": 5.716714037337011e-05,
      "loss": 1.078,
      "step": 47470
    },
    {
      "epoch": 1.923785984886854,
      "grad_norm": 0.6618942618370056,
      "learning_rate": 5.712877667869262e-05,
      "loss": 1.0521,
      "step": 47480
    },
    {
      "epoch": 1.9241911630639574,
      "grad_norm": 0.6548323035240173,
      "learning_rate": 5.709042071294849e-05,
      "loss": 1.0808,
      "step": 47490
    },
    {
      "epoch": 1.9245963412410607,
      "grad_norm": 0.571850597858429,
      "learning_rate": 5.7052072483052555e-05,
      "loss": 1.0994,
      "step": 47500
    },
    {
      "epoch": 1.925001519418164,
      "grad_norm": 0.64261394739151,
      "learning_rate": 5.701373199591835e-05,
      "loss": 1.0782,
      "step": 47510
    },
    {
      "epoch": 1.9254066975952675,
      "grad_norm": 0.6266999244689941,
      "learning_rate": 5.6975399258458064e-05,
      "loss": 1.0791,
      "step": 47520
    },
    {
      "epoch": 1.925811875772371,
      "grad_norm": 0.5923541188240051,
      "learning_rate": 5.693707427758232e-05,
      "loss": 1.0842,
      "step": 47530
    },
    {
      "epoch": 1.9262170539494743,
      "grad_norm": 0.6677608489990234,
      "learning_rate": 5.689875706020049e-05,
      "loss": 1.0384,
      "step": 47540
    },
    {
      "epoch": 1.9266222321265776,
      "grad_norm": 0.5871165990829468,
      "learning_rate": 5.686044761322051e-05,
      "loss": 1.0828,
      "step": 47550
    },
    {
      "epoch": 1.927027410303681,
      "grad_norm": 0.5989214181900024,
      "learning_rate": 5.682214594354893e-05,
      "loss": 1.0615,
      "step": 47560
    },
    {
      "epoch": 1.9274325884807846,
      "grad_norm": 0.6148975491523743,
      "learning_rate": 5.6783852058090806e-05,
      "loss": 1.0262,
      "step": 47570
    },
    {
      "epoch": 1.9278377666578876,
      "grad_norm": 0.5759084224700928,
      "learning_rate": 5.6745565963749925e-05,
      "loss": 1.062,
      "step": 47580
    },
    {
      "epoch": 1.9282429448349911,
      "grad_norm": 0.5949957370758057,
      "learning_rate": 5.670728766742848e-05,
      "loss": 1.0699,
      "step": 47590
    },
    {
      "epoch": 1.9286481230120947,
      "grad_norm": 0.630165159702301,
      "learning_rate": 5.666901717602755e-05,
      "loss": 1.0931,
      "step": 47600
    },
    {
      "epoch": 1.929053301189198,
      "grad_norm": 0.560485303401947,
      "learning_rate": 5.663075449644651e-05,
      "loss": 1.0647,
      "step": 47610
    },
    {
      "epoch": 1.9294584793663012,
      "grad_norm": 0.5932311415672302,
      "learning_rate": 5.659249963558354e-05,
      "loss": 1.1188,
      "step": 47620
    },
    {
      "epoch": 1.9298636575434047,
      "grad_norm": 0.6876125931739807,
      "learning_rate": 5.6554252600335235e-05,
      "loss": 1.1006,
      "step": 47630
    },
    {
      "epoch": 1.9302688357205082,
      "grad_norm": 0.7097763419151306,
      "learning_rate": 5.6516013397596914e-05,
      "loss": 1.0629,
      "step": 47640
    },
    {
      "epoch": 1.9306740138976115,
      "grad_norm": 0.6650599241256714,
      "learning_rate": 5.6477782034262436e-05,
      "loss": 1.0578,
      "step": 47650
    },
    {
      "epoch": 1.9310791920747148,
      "grad_norm": 0.6409139037132263,
      "learning_rate": 5.643955851722429e-05,
      "loss": 1.0425,
      "step": 47660
    },
    {
      "epoch": 1.9314843702518183,
      "grad_norm": 0.5485729575157166,
      "learning_rate": 5.6401342853373416e-05,
      "loss": 1.0547,
      "step": 47670
    },
    {
      "epoch": 1.9318895484289216,
      "grad_norm": 0.7064619064331055,
      "learning_rate": 5.636313504959954e-05,
      "loss": 1.0678,
      "step": 47680
    },
    {
      "epoch": 1.9322947266060249,
      "grad_norm": 0.5436981320381165,
      "learning_rate": 5.6324935112790754e-05,
      "loss": 1.0765,
      "step": 47690
    },
    {
      "epoch": 1.9326999047831284,
      "grad_norm": 0.6042105555534363,
      "learning_rate": 5.628674304983389e-05,
      "loss": 1.082,
      "step": 47700
    },
    {
      "epoch": 1.9331050829602319,
      "grad_norm": 0.6834929585456848,
      "learning_rate": 5.624855886761431e-05,
      "loss": 1.092,
      "step": 47710
    },
    {
      "epoch": 1.9335102611373352,
      "grad_norm": 0.6566229462623596,
      "learning_rate": 5.621038257301601e-05,
      "loss": 1.0902,
      "step": 47720
    },
    {
      "epoch": 1.9339154393144384,
      "grad_norm": 0.6478490233421326,
      "learning_rate": 5.61722141729214e-05,
      "loss": 1.0543,
      "step": 47730
    },
    {
      "epoch": 1.934320617491542,
      "grad_norm": 0.5554546117782593,
      "learning_rate": 5.613405367421167e-05,
      "loss": 1.0739,
      "step": 47740
    },
    {
      "epoch": 1.9347257956686454,
      "grad_norm": 0.5679744482040405,
      "learning_rate": 5.609590108376643e-05,
      "loss": 1.0668,
      "step": 47750
    },
    {
      "epoch": 1.9351309738457485,
      "grad_norm": 0.5669547915458679,
      "learning_rate": 5.6057756408463956e-05,
      "loss": 1.0439,
      "step": 47760
    },
    {
      "epoch": 1.935536152022852,
      "grad_norm": 0.6358092427253723,
      "learning_rate": 5.601961965518104e-05,
      "loss": 1.0456,
      "step": 47770
    },
    {
      "epoch": 1.9359413301999555,
      "grad_norm": 0.6135358810424805,
      "learning_rate": 5.598149083079315e-05,
      "loss": 1.1234,
      "step": 47780
    },
    {
      "epoch": 1.9363465083770588,
      "grad_norm": 0.652281641960144,
      "learning_rate": 5.594336994217415e-05,
      "loss": 1.0456,
      "step": 47790
    },
    {
      "epoch": 1.936751686554162,
      "grad_norm": 0.62815260887146,
      "learning_rate": 5.590525699619661e-05,
      "loss": 1.0917,
      "step": 47800
    },
    {
      "epoch": 1.9371568647312656,
      "grad_norm": 0.6401068568229675,
      "learning_rate": 5.586715199973163e-05,
      "loss": 1.0659,
      "step": 47810
    },
    {
      "epoch": 1.937562042908369,
      "grad_norm": 0.5857486724853516,
      "learning_rate": 5.5829054959648916e-05,
      "loss": 1.0896,
      "step": 47820
    },
    {
      "epoch": 1.9379672210854724,
      "grad_norm": 0.6683955788612366,
      "learning_rate": 5.579096588281664e-05,
      "loss": 1.0874,
      "step": 47830
    },
    {
      "epoch": 1.9383723992625757,
      "grad_norm": 0.7160133123397827,
      "learning_rate": 5.575288477610161e-05,
      "loss": 1.1085,
      "step": 47840
    },
    {
      "epoch": 1.9387775774396792,
      "grad_norm": 0.6096618175506592,
      "learning_rate": 5.571481164636922e-05,
      "loss": 1.1017,
      "step": 47850
    },
    {
      "epoch": 1.9391827556167824,
      "grad_norm": 0.5850949883460999,
      "learning_rate": 5.5676746500483336e-05,
      "loss": 1.0402,
      "step": 47860
    },
    {
      "epoch": 1.9395879337938857,
      "grad_norm": 0.5975541472434998,
      "learning_rate": 5.563868934530647e-05,
      "loss": 1.0541,
      "step": 47870
    },
    {
      "epoch": 1.9399931119709892,
      "grad_norm": 0.520793616771698,
      "learning_rate": 5.560064018769966e-05,
      "loss": 1.0718,
      "step": 47880
    },
    {
      "epoch": 1.9403982901480927,
      "grad_norm": 0.5867084264755249,
      "learning_rate": 5.556259903452254e-05,
      "loss": 1.0578,
      "step": 47890
    },
    {
      "epoch": 1.940803468325196,
      "grad_norm": 0.6344953775405884,
      "learning_rate": 5.552456589263321e-05,
      "loss": 1.1049,
      "step": 47900
    },
    {
      "epoch": 1.9412086465022993,
      "grad_norm": 0.6288391947746277,
      "learning_rate": 5.5486540768888426e-05,
      "loss": 1.1062,
      "step": 47910
    },
    {
      "epoch": 1.9416138246794028,
      "grad_norm": 0.6128796339035034,
      "learning_rate": 5.5448523670143403e-05,
      "loss": 1.0772,
      "step": 47920
    },
    {
      "epoch": 1.9420190028565063,
      "grad_norm": 0.642906665802002,
      "learning_rate": 5.5410514603251985e-05,
      "loss": 1.0744,
      "step": 47930
    },
    {
      "epoch": 1.9424241810336094,
      "grad_norm": 0.6099886894226074,
      "learning_rate": 5.5372513575066546e-05,
      "loss": 1.0674,
      "step": 47940
    },
    {
      "epoch": 1.9428293592107129,
      "grad_norm": 0.6381770968437195,
      "learning_rate": 5.533452059243806e-05,
      "loss": 1.0734,
      "step": 47950
    },
    {
      "epoch": 1.9432345373878164,
      "grad_norm": 0.6606588363647461,
      "learning_rate": 5.5296535662215885e-05,
      "loss": 1.0685,
      "step": 47960
    },
    {
      "epoch": 1.9436397155649197,
      "grad_norm": 0.7084717750549316,
      "learning_rate": 5.525855879124817e-05,
      "loss": 1.1149,
      "step": 47970
    },
    {
      "epoch": 1.944044893742023,
      "grad_norm": 0.6479068994522095,
      "learning_rate": 5.5220589986381334e-05,
      "loss": 1.0727,
      "step": 47980
    },
    {
      "epoch": 1.9444500719191264,
      "grad_norm": 0.6341657042503357,
      "learning_rate": 5.518262925446064e-05,
      "loss": 1.104,
      "step": 47990
    },
    {
      "epoch": 1.94485525009623,
      "grad_norm": 0.6789267063140869,
      "learning_rate": 5.514467660232965e-05,
      "loss": 1.1045,
      "step": 48000
    },
    {
      "epoch": 1.9452604282733332,
      "grad_norm": 0.6281228065490723,
      "learning_rate": 5.510673203683061e-05,
      "loss": 1.0819,
      "step": 48010
    },
    {
      "epoch": 1.9456656064504365,
      "grad_norm": 0.6865042448043823,
      "learning_rate": 5.506879556480422e-05,
      "loss": 1.0721,
      "step": 48020
    },
    {
      "epoch": 1.94607078462754,
      "grad_norm": 0.6555408239364624,
      "learning_rate": 5.503086719308983e-05,
      "loss": 1.0572,
      "step": 48030
    },
    {
      "epoch": 1.9464759628046433,
      "grad_norm": 0.6094871759414673,
      "learning_rate": 5.4992946928525146e-05,
      "loss": 1.0494,
      "step": 48040
    },
    {
      "epoch": 1.9468811409817466,
      "grad_norm": 0.5591763854026794,
      "learning_rate": 5.495503477794669e-05,
      "loss": 1.1038,
      "step": 48050
    },
    {
      "epoch": 1.94728631915885,
      "grad_norm": 0.7272982597351074,
      "learning_rate": 5.491713074818923e-05,
      "loss": 1.075,
      "step": 48060
    },
    {
      "epoch": 1.9476914973359536,
      "grad_norm": 0.5759680867195129,
      "learning_rate": 5.487923484608629e-05,
      "loss": 1.0308,
      "step": 48070
    },
    {
      "epoch": 1.9480966755130569,
      "grad_norm": 0.6414573192596436,
      "learning_rate": 5.4841347078469764e-05,
      "loss": 1.0285,
      "step": 48080
    },
    {
      "epoch": 1.9485018536901602,
      "grad_norm": 0.7002025842666626,
      "learning_rate": 5.4803467452170176e-05,
      "loss": 1.1275,
      "step": 48090
    },
    {
      "epoch": 1.9489070318672637,
      "grad_norm": 0.6476113200187683,
      "learning_rate": 5.476559597401659e-05,
      "loss": 1.1307,
      "step": 48100
    },
    {
      "epoch": 1.9493122100443672,
      "grad_norm": 0.6242934465408325,
      "learning_rate": 5.472773265083659e-05,
      "loss": 1.0458,
      "step": 48110
    },
    {
      "epoch": 1.9497173882214702,
      "grad_norm": 0.6268385648727417,
      "learning_rate": 5.4689877489456196e-05,
      "loss": 1.0672,
      "step": 48120
    },
    {
      "epoch": 1.9501225663985737,
      "grad_norm": 0.7166338562965393,
      "learning_rate": 5.4652030496700134e-05,
      "loss": 1.0747,
      "step": 48130
    },
    {
      "epoch": 1.9505277445756772,
      "grad_norm": 0.6183237433433533,
      "learning_rate": 5.4614191679391444e-05,
      "loss": 1.0873,
      "step": 48140
    },
    {
      "epoch": 1.9509329227527805,
      "grad_norm": 0.5798344612121582,
      "learning_rate": 5.4576361044351866e-05,
      "loss": 1.0671,
      "step": 48150
    },
    {
      "epoch": 1.9513381009298838,
      "grad_norm": 0.6782681941986084,
      "learning_rate": 5.453853859840159e-05,
      "loss": 1.0738,
      "step": 48160
    },
    {
      "epoch": 1.9517432791069873,
      "grad_norm": 0.5310600996017456,
      "learning_rate": 5.45007243483594e-05,
      "loss": 1.0768,
      "step": 48170
    },
    {
      "epoch": 1.9521484572840908,
      "grad_norm": 0.6469187140464783,
      "learning_rate": 5.446291830104244e-05,
      "loss": 1.0684,
      "step": 48180
    },
    {
      "epoch": 1.952553635461194,
      "grad_norm": 0.6146342754364014,
      "learning_rate": 5.4425120463266554e-05,
      "loss": 1.0612,
      "step": 48190
    },
    {
      "epoch": 1.9529588136382974,
      "grad_norm": 0.5984562039375305,
      "learning_rate": 5.438733084184604e-05,
      "loss": 1.0638,
      "step": 48200
    },
    {
      "epoch": 1.9533639918154009,
      "grad_norm": 0.633702278137207,
      "learning_rate": 5.434954944359365e-05,
      "loss": 1.0479,
      "step": 48210
    },
    {
      "epoch": 1.9537691699925042,
      "grad_norm": 0.6160995364189148,
      "learning_rate": 5.431177627532073e-05,
      "loss": 1.0757,
      "step": 48220
    },
    {
      "epoch": 1.9541743481696074,
      "grad_norm": 0.6334740519523621,
      "learning_rate": 5.427401134383714e-05,
      "loss": 1.0818,
      "step": 48230
    },
    {
      "epoch": 1.954579526346711,
      "grad_norm": 0.6614764928817749,
      "learning_rate": 5.423625465595128e-05,
      "loss": 1.1039,
      "step": 48240
    },
    {
      "epoch": 1.9549847045238145,
      "grad_norm": 0.5700522065162659,
      "learning_rate": 5.4198506218469925e-05,
      "loss": 1.0843,
      "step": 48250
    },
    {
      "epoch": 1.9553898827009177,
      "grad_norm": 0.6780952215194702,
      "learning_rate": 5.416076603819851e-05,
      "loss": 1.1062,
      "step": 48260
    },
    {
      "epoch": 1.955795060878021,
      "grad_norm": 0.656966507434845,
      "learning_rate": 5.412303412194093e-05,
      "loss": 1.0892,
      "step": 48270
    },
    {
      "epoch": 1.9562002390551245,
      "grad_norm": 0.6640673875808716,
      "learning_rate": 5.408531047649964e-05,
      "loss": 1.1251,
      "step": 48280
    },
    {
      "epoch": 1.956605417232228,
      "grad_norm": 0.6117696166038513,
      "learning_rate": 5.404759510867545e-05,
      "loss": 1.0781,
      "step": 48290
    },
    {
      "epoch": 1.957010595409331,
      "grad_norm": 0.6202933192253113,
      "learning_rate": 5.4009888025267894e-05,
      "loss": 1.088,
      "step": 48300
    },
    {
      "epoch": 1.9574157735864346,
      "grad_norm": 0.5914474725723267,
      "learning_rate": 5.39721892330748e-05,
      "loss": 1.0608,
      "step": 48310
    },
    {
      "epoch": 1.957820951763538,
      "grad_norm": 0.6325544118881226,
      "learning_rate": 5.393449873889265e-05,
      "loss": 1.1311,
      "step": 48320
    },
    {
      "epoch": 1.9582261299406414,
      "grad_norm": 0.6088436841964722,
      "learning_rate": 5.389681654951637e-05,
      "loss": 1.0636,
      "step": 48330
    },
    {
      "epoch": 1.9586313081177447,
      "grad_norm": 0.5912893414497375,
      "learning_rate": 5.385914267173946e-05,
      "loss": 1.0716,
      "step": 48340
    },
    {
      "epoch": 1.9590364862948482,
      "grad_norm": 0.6698079109191895,
      "learning_rate": 5.382147711235377e-05,
      "loss": 1.0754,
      "step": 48350
    },
    {
      "epoch": 1.9594416644719517,
      "grad_norm": 0.6249427199363708,
      "learning_rate": 5.3783819878149824e-05,
      "loss": 1.0739,
      "step": 48360
    },
    {
      "epoch": 1.959846842649055,
      "grad_norm": 0.6173605918884277,
      "learning_rate": 5.37461709759165e-05,
      "loss": 1.0749,
      "step": 48370
    },
    {
      "epoch": 1.9602520208261582,
      "grad_norm": 0.6668561100959778,
      "learning_rate": 5.370853041244125e-05,
      "loss": 1.0589,
      "step": 48380
    },
    {
      "epoch": 1.9606571990032617,
      "grad_norm": 0.6392799019813538,
      "learning_rate": 5.367089819451003e-05,
      "loss": 1.094,
      "step": 48390
    },
    {
      "epoch": 1.961062377180365,
      "grad_norm": 0.6658574938774109,
      "learning_rate": 5.36332743289073e-05,
      "loss": 1.0191,
      "step": 48400
    },
    {
      "epoch": 1.9614675553574683,
      "grad_norm": 0.6633474230766296,
      "learning_rate": 5.359565882241593e-05,
      "loss": 1.1103,
      "step": 48410
    },
    {
      "epoch": 1.9618727335345718,
      "grad_norm": 0.5923685431480408,
      "learning_rate": 5.355805168181738e-05,
      "loss": 1.081,
      "step": 48420
    },
    {
      "epoch": 1.9622779117116753,
      "grad_norm": 0.615236759185791,
      "learning_rate": 5.3520452913891486e-05,
      "loss": 1.0618,
      "step": 48430
    },
    {
      "epoch": 1.9626830898887786,
      "grad_norm": 0.6727345585823059,
      "learning_rate": 5.3482862525416765e-05,
      "loss": 1.0569,
      "step": 48440
    },
    {
      "epoch": 1.9630882680658819,
      "grad_norm": 0.6584258079528809,
      "learning_rate": 5.3445280523170035e-05,
      "loss": 1.0709,
      "step": 48450
    },
    {
      "epoch": 1.9634934462429854,
      "grad_norm": 0.5608727335929871,
      "learning_rate": 5.3407706913926714e-05,
      "loss": 1.0508,
      "step": 48460
    },
    {
      "epoch": 1.963898624420089,
      "grad_norm": 0.5571438074111938,
      "learning_rate": 5.337014170446061e-05,
      "loss": 1.0749,
      "step": 48470
    },
    {
      "epoch": 1.964303802597192,
      "grad_norm": 0.6441141366958618,
      "learning_rate": 5.333258490154411e-05,
      "loss": 1.0554,
      "step": 48480
    },
    {
      "epoch": 1.9647089807742955,
      "grad_norm": 0.6543720960617065,
      "learning_rate": 5.329503651194805e-05,
      "loss": 1.0678,
      "step": 48490
    },
    {
      "epoch": 1.965114158951399,
      "grad_norm": 0.6999475359916687,
      "learning_rate": 5.325749654244179e-05,
      "loss": 1.087,
      "step": 48500
    },
    {
      "epoch": 1.9655193371285022,
      "grad_norm": 0.6370114684104919,
      "learning_rate": 5.321996499979304e-05,
      "loss": 1.0677,
      "step": 48510
    },
    {
      "epoch": 1.9659245153056055,
      "grad_norm": 0.6438090205192566,
      "learning_rate": 5.318244189076813e-05,
      "loss": 1.1055,
      "step": 48520
    },
    {
      "epoch": 1.966329693482709,
      "grad_norm": 0.642512321472168,
      "learning_rate": 5.314492722213188e-05,
      "loss": 1.0601,
      "step": 48530
    },
    {
      "epoch": 1.9667348716598125,
      "grad_norm": 0.5581594109535217,
      "learning_rate": 5.310742100064743e-05,
      "loss": 1.0584,
      "step": 48540
    },
    {
      "epoch": 1.9671400498369158,
      "grad_norm": 0.6130458116531372,
      "learning_rate": 5.306992323307653e-05,
      "loss": 1.0702,
      "step": 48550
    },
    {
      "epoch": 1.967545228014019,
      "grad_norm": 0.6393803358078003,
      "learning_rate": 5.3032433926179395e-05,
      "loss": 1.0868,
      "step": 48560
    },
    {
      "epoch": 1.9679504061911226,
      "grad_norm": 0.6790395975112915,
      "learning_rate": 5.2994953086714715e-05,
      "loss": 1.0958,
      "step": 48570
    },
    {
      "epoch": 1.9683555843682259,
      "grad_norm": 0.6102782487869263,
      "learning_rate": 5.295748072143956e-05,
      "loss": 1.0456,
      "step": 48580
    },
    {
      "epoch": 1.9687607625453292,
      "grad_norm": 0.7626858353614807,
      "learning_rate": 5.292001683710961e-05,
      "loss": 1.0335,
      "step": 48590
    },
    {
      "epoch": 1.9691659407224327,
      "grad_norm": 0.5979097485542297,
      "learning_rate": 5.288256144047885e-05,
      "loss": 1.1065,
      "step": 48600
    },
    {
      "epoch": 1.9695711188995362,
      "grad_norm": 0.592881977558136,
      "learning_rate": 5.284511453829999e-05,
      "loss": 1.0965,
      "step": 48610
    },
    {
      "epoch": 1.9699762970766395,
      "grad_norm": 0.6243888139724731,
      "learning_rate": 5.280767613732392e-05,
      "loss": 1.0837,
      "step": 48620
    },
    {
      "epoch": 1.9703814752537427,
      "grad_norm": 0.5852140188217163,
      "learning_rate": 5.2770246244300224e-05,
      "loss": 1.068,
      "step": 48630
    },
    {
      "epoch": 1.9707866534308462,
      "grad_norm": 0.5958802103996277,
      "learning_rate": 5.273282486597677e-05,
      "loss": 1.0966,
      "step": 48640
    },
    {
      "epoch": 1.9711918316079498,
      "grad_norm": 0.6603156924247742,
      "learning_rate": 5.2695412009100064e-05,
      "loss": 1.0727,
      "step": 48650
    },
    {
      "epoch": 1.9715970097850528,
      "grad_norm": 0.6363398432731628,
      "learning_rate": 5.265800768041487e-05,
      "loss": 1.0624,
      "step": 48660
    },
    {
      "epoch": 1.9720021879621563,
      "grad_norm": 0.6622248291969299,
      "learning_rate": 5.2620611886664696e-05,
      "loss": 1.085,
      "step": 48670
    },
    {
      "epoch": 1.9724073661392598,
      "grad_norm": 0.5624070763587952,
      "learning_rate": 5.258322463459122e-05,
      "loss": 1.0602,
      "step": 48680
    },
    {
      "epoch": 1.972812544316363,
      "grad_norm": 0.664053738117218,
      "learning_rate": 5.2545845930934804e-05,
      "loss": 1.0948,
      "step": 48690
    },
    {
      "epoch": 1.9732177224934664,
      "grad_norm": 0.5901891589164734,
      "learning_rate": 5.2508475782434093e-05,
      "loss": 1.0689,
      "step": 48700
    },
    {
      "epoch": 1.97362290067057,
      "grad_norm": 0.6587003469467163,
      "learning_rate": 5.247111419582631e-05,
      "loss": 1.0663,
      "step": 48710
    },
    {
      "epoch": 1.9740280788476734,
      "grad_norm": 0.5877283215522766,
      "learning_rate": 5.243376117784711e-05,
      "loss": 1.0943,
      "step": 48720
    },
    {
      "epoch": 1.9744332570247767,
      "grad_norm": 0.6027305722236633,
      "learning_rate": 5.2396416735230604e-05,
      "loss": 1.0942,
      "step": 48730
    },
    {
      "epoch": 1.97483843520188,
      "grad_norm": 0.7264521718025208,
      "learning_rate": 5.2359080874709285e-05,
      "loss": 1.0739,
      "step": 48740
    },
    {
      "epoch": 1.9752436133789835,
      "grad_norm": 0.6402539610862732,
      "learning_rate": 5.232175360301424e-05,
      "loss": 1.0698,
      "step": 48750
    },
    {
      "epoch": 1.9756487915560867,
      "grad_norm": 0.5913277864456177,
      "learning_rate": 5.228443492687481e-05,
      "loss": 1.0628,
      "step": 48760
    },
    {
      "epoch": 1.97605396973319,
      "grad_norm": 0.6599008440971375,
      "learning_rate": 5.224712485301898e-05,
      "loss": 1.0858,
      "step": 48770
    },
    {
      "epoch": 1.9764591479102935,
      "grad_norm": 0.6049656867980957,
      "learning_rate": 5.2209823388173095e-05,
      "loss": 1.0479,
      "step": 48780
    },
    {
      "epoch": 1.976864326087397,
      "grad_norm": 0.5526579022407532,
      "learning_rate": 5.217253053906199e-05,
      "loss": 1.0929,
      "step": 48790
    },
    {
      "epoch": 1.9772695042645003,
      "grad_norm": 0.631309986114502,
      "learning_rate": 5.2135246312408845e-05,
      "loss": 1.0699,
      "step": 48800
    },
    {
      "epoch": 1.9776746824416036,
      "grad_norm": 0.6658692359924316,
      "learning_rate": 5.209797071493539e-05,
      "loss": 1.0828,
      "step": 48810
    },
    {
      "epoch": 1.978079860618707,
      "grad_norm": 0.6394604444503784,
      "learning_rate": 5.2060703753361785e-05,
      "loss": 1.0972,
      "step": 48820
    },
    {
      "epoch": 1.9784850387958106,
      "grad_norm": 0.6210819482803345,
      "learning_rate": 5.202344543440657e-05,
      "loss": 1.0598,
      "step": 48830
    },
    {
      "epoch": 1.9788902169729137,
      "grad_norm": 0.6596384644508362,
      "learning_rate": 5.198619576478678e-05,
      "loss": 1.0939,
      "step": 48840
    },
    {
      "epoch": 1.9792953951500172,
      "grad_norm": 0.8433650732040405,
      "learning_rate": 5.19489547512179e-05,
      "loss": 1.1062,
      "step": 48850
    },
    {
      "epoch": 1.9797005733271207,
      "grad_norm": 0.6279051303863525,
      "learning_rate": 5.191172240041387e-05,
      "loss": 1.0501,
      "step": 48860
    },
    {
      "epoch": 1.980105751504224,
      "grad_norm": 0.542404294013977,
      "learning_rate": 5.187449871908694e-05,
      "loss": 1.1077,
      "step": 48870
    },
    {
      "epoch": 1.9805109296813272,
      "grad_norm": 0.6695194244384766,
      "learning_rate": 5.1837283713947936e-05,
      "loss": 1.0942,
      "step": 48880
    },
    {
      "epoch": 1.9809161078584308,
      "grad_norm": 0.6225569844245911,
      "learning_rate": 5.180007739170609e-05,
      "loss": 1.1159,
      "step": 48890
    },
    {
      "epoch": 1.9813212860355343,
      "grad_norm": 0.6230878233909607,
      "learning_rate": 5.1762879759069085e-05,
      "loss": 1.0719,
      "step": 48900
    },
    {
      "epoch": 1.9817264642126375,
      "grad_norm": 1.310440182685852,
      "learning_rate": 5.17256908227429e-05,
      "loss": 1.067,
      "step": 48910
    },
    {
      "epoch": 1.9821316423897408,
      "grad_norm": 0.5634238123893738,
      "learning_rate": 5.168851058943216e-05,
      "loss": 1.0581,
      "step": 48920
    },
    {
      "epoch": 1.9825368205668443,
      "grad_norm": 0.6761951446533203,
      "learning_rate": 5.165133906583972e-05,
      "loss": 1.0416,
      "step": 48930
    },
    {
      "epoch": 1.9829419987439476,
      "grad_norm": 0.6263960003852844,
      "learning_rate": 5.1614176258667e-05,
      "loss": 1.0916,
      "step": 48940
    },
    {
      "epoch": 1.983347176921051,
      "grad_norm": 0.6370680928230286,
      "learning_rate": 5.157702217461382e-05,
      "loss": 1.0609,
      "step": 48950
    },
    {
      "epoch": 1.9837523550981544,
      "grad_norm": 0.6852926015853882,
      "learning_rate": 5.153987682037842e-05,
      "loss": 1.0742,
      "step": 48960
    },
    {
      "epoch": 1.984157533275258,
      "grad_norm": 0.6022751927375793,
      "learning_rate": 5.150274020265741e-05,
      "loss": 1.0887,
      "step": 48970
    },
    {
      "epoch": 1.9845627114523612,
      "grad_norm": 0.5522090196609497,
      "learning_rate": 5.146561232814593e-05,
      "loss": 1.0669,
      "step": 48980
    },
    {
      "epoch": 1.9849678896294645,
      "grad_norm": 0.5529991984367371,
      "learning_rate": 5.142849320353743e-05,
      "loss": 1.0683,
      "step": 48990
    },
    {
      "epoch": 1.985373067806568,
      "grad_norm": 0.6668347716331482,
      "learning_rate": 5.139138283552387e-05,
      "loss": 1.0553,
      "step": 49000
    },
    {
      "epoch": 1.9857782459836715,
      "grad_norm": 0.6432819366455078,
      "learning_rate": 5.13542812307956e-05,
      "loss": 1.0753,
      "step": 49010
    },
    {
      "epoch": 1.9861834241607745,
      "grad_norm": 0.6244034171104431,
      "learning_rate": 5.1317188396041426e-05,
      "loss": 1.0509,
      "step": 49020
    },
    {
      "epoch": 1.986588602337878,
      "grad_norm": 0.615607500076294,
      "learning_rate": 5.128010433794846e-05,
      "loss": 1.0744,
      "step": 49030
    },
    {
      "epoch": 1.9869937805149815,
      "grad_norm": 0.6271893978118896,
      "learning_rate": 5.124302906320241e-05,
      "loss": 1.0933,
      "step": 49040
    },
    {
      "epoch": 1.9873989586920848,
      "grad_norm": 0.647770345211029,
      "learning_rate": 5.1205962578487155e-05,
      "loss": 1.1121,
      "step": 49050
    },
    {
      "epoch": 1.987804136869188,
      "grad_norm": 0.6285046339035034,
      "learning_rate": 5.116890489048531e-05,
      "loss": 1.0609,
      "step": 49060
    },
    {
      "epoch": 1.9882093150462916,
      "grad_norm": 0.7415581345558167,
      "learning_rate": 5.113185600587762e-05,
      "loss": 1.0656,
      "step": 49070
    },
    {
      "epoch": 1.9886144932233951,
      "grad_norm": 0.646701455116272,
      "learning_rate": 5.109481593134342e-05,
      "loss": 1.0825,
      "step": 49080
    },
    {
      "epoch": 1.9890196714004984,
      "grad_norm": 0.5649621486663818,
      "learning_rate": 5.1057784673560295e-05,
      "loss": 1.1068,
      "step": 49090
    },
    {
      "epoch": 1.9894248495776017,
      "grad_norm": 0.5801345109939575,
      "learning_rate": 5.102076223920439e-05,
      "loss": 1.0809,
      "step": 49100
    },
    {
      "epoch": 1.9898300277547052,
      "grad_norm": 0.6076604723930359,
      "learning_rate": 5.0983748634950204e-05,
      "loss": 1.0874,
      "step": 49110
    },
    {
      "epoch": 1.9902352059318085,
      "grad_norm": 0.6036555767059326,
      "learning_rate": 5.094674386747067e-05,
      "loss": 1.0903,
      "step": 49120
    },
    {
      "epoch": 1.9906403841089118,
      "grad_norm": 0.59818035364151,
      "learning_rate": 5.090974794343705e-05,
      "loss": 1.0441,
      "step": 49130
    },
    {
      "epoch": 1.9910455622860153,
      "grad_norm": 0.6133650541305542,
      "learning_rate": 5.087276086951908e-05,
      "loss": 1.0329,
      "step": 49140
    },
    {
      "epoch": 1.9914507404631188,
      "grad_norm": 0.6454243659973145,
      "learning_rate": 5.083578265238492e-05,
      "loss": 1.0471,
      "step": 49150
    },
    {
      "epoch": 1.991855918640222,
      "grad_norm": 0.625420331954956,
      "learning_rate": 5.079881329870104e-05,
      "loss": 1.0758,
      "step": 49160
    },
    {
      "epoch": 1.9922610968173253,
      "grad_norm": 0.6343829035758972,
      "learning_rate": 5.0761852815132394e-05,
      "loss": 1.1099,
      "step": 49170
    },
    {
      "epoch": 1.9926662749944288,
      "grad_norm": 0.5965797901153564,
      "learning_rate": 5.0724901208342304e-05,
      "loss": 1.0784,
      "step": 49180
    },
    {
      "epoch": 1.9930714531715323,
      "grad_norm": 0.5951653718948364,
      "learning_rate": 5.068795848499257e-05,
      "loss": 1.0955,
      "step": 49190
    },
    {
      "epoch": 1.9934766313486354,
      "grad_norm": 0.5910769701004028,
      "learning_rate": 5.0651024651743205e-05,
      "loss": 1.0719,
      "step": 49200
    },
    {
      "epoch": 1.993881809525739,
      "grad_norm": 0.5854232907295227,
      "learning_rate": 5.061409971525284e-05,
      "loss": 1.0335,
      "step": 49210
    },
    {
      "epoch": 1.9942869877028424,
      "grad_norm": 0.5742017030715942,
      "learning_rate": 5.057718368217825e-05,
      "loss": 1.087,
      "step": 49220
    },
    {
      "epoch": 1.9946921658799457,
      "grad_norm": 0.6384209394454956,
      "learning_rate": 5.054027655917495e-05,
      "loss": 1.052,
      "step": 49230
    },
    {
      "epoch": 1.995097344057049,
      "grad_norm": 0.5987688899040222,
      "learning_rate": 5.050337835289649e-05,
      "loss": 1.1223,
      "step": 49240
    },
    {
      "epoch": 1.9955025222341525,
      "grad_norm": 0.6527283787727356,
      "learning_rate": 5.046648906999507e-05,
      "loss": 1.0944,
      "step": 49250
    },
    {
      "epoch": 1.995907700411256,
      "grad_norm": 0.6813449263572693,
      "learning_rate": 5.042960871712112e-05,
      "loss": 1.0793,
      "step": 49260
    },
    {
      "epoch": 1.9963128785883593,
      "grad_norm": 0.5833953022956848,
      "learning_rate": 5.0392737300923576e-05,
      "loss": 1.0733,
      "step": 49270
    },
    {
      "epoch": 1.9967180567654625,
      "grad_norm": 0.5683633089065552,
      "learning_rate": 5.035587482804962e-05,
      "loss": 1.0982,
      "step": 49280
    },
    {
      "epoch": 1.997123234942566,
      "grad_norm": 0.6328427791595459,
      "learning_rate": 5.031902130514503e-05,
      "loss": 1.0852,
      "step": 49290
    },
    {
      "epoch": 1.9975284131196693,
      "grad_norm": 0.5781868696212769,
      "learning_rate": 5.028217673885378e-05,
      "loss": 1.0749,
      "step": 49300
    },
    {
      "epoch": 1.9979335912967726,
      "grad_norm": 0.6396767497062683,
      "learning_rate": 5.0245341135818355e-05,
      "loss": 1.0964,
      "step": 49310
    },
    {
      "epoch": 1.9983387694738761,
      "grad_norm": 0.5985438227653503,
      "learning_rate": 5.020851450267949e-05,
      "loss": 1.0868,
      "step": 49320
    },
    {
      "epoch": 1.9987439476509796,
      "grad_norm": 0.5922365188598633,
      "learning_rate": 5.0171696846076446e-05,
      "loss": 1.0893,
      "step": 49330
    },
    {
      "epoch": 1.999149125828083,
      "grad_norm": 0.6126964092254639,
      "learning_rate": 5.0134888172646776e-05,
      "loss": 1.0762,
      "step": 49340
    },
    {
      "epoch": 1.9995543040051862,
      "grad_norm": 0.6332033276557922,
      "learning_rate": 5.0098088489026504e-05,
      "loss": 1.0485,
      "step": 49350
    },
    {
      "epoch": 1.9999594821822897,
      "grad_norm": 0.5765609741210938,
      "learning_rate": 5.0061297801849885e-05,
      "loss": 1.1504,
      "step": 49360
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.0912619829177856,
      "eval_runtime": 952.3368,
      "eval_samples_per_second": 11.519,
      "eval_steps_per_second": 1.441,
      "step": 49361
    },
    {
      "epoch": 2.000364660359393,
      "grad_norm": 0.6282658576965332,
      "learning_rate": 5.0024516117749695e-05,
      "loss": 1.0592,
      "step": 49370
    },
    {
      "epoch": 2.0007698385364963,
      "grad_norm": 0.6609070897102356,
      "learning_rate": 4.9987743443356994e-05,
      "loss": 1.0507,
      "step": 49380
    },
    {
      "epoch": 2.0011750167135998,
      "grad_norm": 0.6283262968063354,
      "learning_rate": 4.995097978530125e-05,
      "loss": 1.0465,
      "step": 49390
    },
    {
      "epoch": 2.0015801948907033,
      "grad_norm": 0.6575890183448792,
      "learning_rate": 4.9914225150210335e-05,
      "loss": 1.0149,
      "step": 49400
    },
    {
      "epoch": 2.0019853730678068,
      "grad_norm": 0.6332368850708008,
      "learning_rate": 4.9877479544710504e-05,
      "loss": 1.0551,
      "step": 49410
    },
    {
      "epoch": 2.00239055124491,
      "grad_norm": 0.6696742177009583,
      "learning_rate": 4.9840742975426256e-05,
      "loss": 1.0442,
      "step": 49420
    },
    {
      "epoch": 2.0027957294220133,
      "grad_norm": 0.7259538769721985,
      "learning_rate": 4.980401544898064e-05,
      "loss": 1.0517,
      "step": 49430
    },
    {
      "epoch": 2.003200907599117,
      "grad_norm": 0.5558461546897888,
      "learning_rate": 4.976729697199493e-05,
      "loss": 1.0429,
      "step": 49440
    },
    {
      "epoch": 2.00360608577622,
      "grad_norm": 0.6062142848968506,
      "learning_rate": 4.973058755108882e-05,
      "loss": 1.0784,
      "step": 49450
    },
    {
      "epoch": 2.0040112639533234,
      "grad_norm": 0.631935179233551,
      "learning_rate": 4.969388719288043e-05,
      "loss": 1.0281,
      "step": 49460
    },
    {
      "epoch": 2.004416442130427,
      "grad_norm": 0.6566844582557678,
      "learning_rate": 4.9657195903986185e-05,
      "loss": 1.0813,
      "step": 49470
    },
    {
      "epoch": 2.0048216203075304,
      "grad_norm": 0.5708945989608765,
      "learning_rate": 4.962051369102084e-05,
      "loss": 1.0931,
      "step": 49480
    },
    {
      "epoch": 2.0052267984846335,
      "grad_norm": 0.6495814323425293,
      "learning_rate": 4.9583840560597596e-05,
      "loss": 1.0566,
      "step": 49490
    },
    {
      "epoch": 2.005631976661737,
      "grad_norm": 0.780982494354248,
      "learning_rate": 4.9547176519327955e-05,
      "loss": 1.028,
      "step": 49500
    },
    {
      "epoch": 2.0060371548388405,
      "grad_norm": 0.5884227752685547,
      "learning_rate": 4.951052157382186e-05,
      "loss": 1.0912,
      "step": 49510
    },
    {
      "epoch": 2.0064423330159435,
      "grad_norm": 0.6640500426292419,
      "learning_rate": 4.947387573068747e-05,
      "loss": 1.0491,
      "step": 49520
    },
    {
      "epoch": 2.006847511193047,
      "grad_norm": 0.5884891152381897,
      "learning_rate": 4.943723899653143e-05,
      "loss": 1.0697,
      "step": 49530
    },
    {
      "epoch": 2.0072526893701506,
      "grad_norm": 0.6185877919197083,
      "learning_rate": 4.940061137795876e-05,
      "loss": 1.0594,
      "step": 49540
    },
    {
      "epoch": 2.007657867547254,
      "grad_norm": 0.6371936798095703,
      "learning_rate": 4.936399288157268e-05,
      "loss": 1.017,
      "step": 49550
    },
    {
      "epoch": 2.008063045724357,
      "grad_norm": 0.6459710597991943,
      "learning_rate": 4.932738351397492e-05,
      "loss": 1.0696,
      "step": 49560
    },
    {
      "epoch": 2.0084682239014606,
      "grad_norm": 0.6290584206581116,
      "learning_rate": 4.9290783281765516e-05,
      "loss": 1.0566,
      "step": 49570
    },
    {
      "epoch": 2.008873402078564,
      "grad_norm": 0.6040071845054626,
      "learning_rate": 4.925419219154287e-05,
      "loss": 1.0635,
      "step": 49580
    },
    {
      "epoch": 2.0092785802556676,
      "grad_norm": 0.5669220685958862,
      "learning_rate": 4.921761024990364e-05,
      "loss": 1.0181,
      "step": 49590
    },
    {
      "epoch": 2.0096837584327707,
      "grad_norm": 0.6093472838401794,
      "learning_rate": 4.9181037463443015e-05,
      "loss": 1.0718,
      "step": 49600
    },
    {
      "epoch": 2.010088936609874,
      "grad_norm": 0.6833994388580322,
      "learning_rate": 4.914447383875432e-05,
      "loss": 1.098,
      "step": 49610
    },
    {
      "epoch": 2.0104941147869777,
      "grad_norm": 0.5997800230979919,
      "learning_rate": 4.91079193824294e-05,
      "loss": 1.073,
      "step": 49620
    },
    {
      "epoch": 2.0108992929640808,
      "grad_norm": 0.6723470687866211,
      "learning_rate": 4.907137410105838e-05,
      "loss": 1.0458,
      "step": 49630
    },
    {
      "epoch": 2.0113044711411843,
      "grad_norm": 0.600959062576294,
      "learning_rate": 4.903483800122977e-05,
      "loss": 1.0994,
      "step": 49640
    },
    {
      "epoch": 2.0117096493182878,
      "grad_norm": 0.647742509841919,
      "learning_rate": 4.899831108953031e-05,
      "loss": 1.0439,
      "step": 49650
    },
    {
      "epoch": 2.0121148274953913,
      "grad_norm": 0.6461215019226074,
      "learning_rate": 4.896179337254525e-05,
      "loss": 1.0679,
      "step": 49660
    },
    {
      "epoch": 2.0125200056724943,
      "grad_norm": 0.7219367623329163,
      "learning_rate": 4.892528485685799e-05,
      "loss": 1.0394,
      "step": 49670
    },
    {
      "epoch": 2.012925183849598,
      "grad_norm": 0.6536914706230164,
      "learning_rate": 4.888878554905051e-05,
      "loss": 1.0544,
      "step": 49680
    },
    {
      "epoch": 2.0133303620267013,
      "grad_norm": 0.7232722640037537,
      "learning_rate": 4.88522954557029e-05,
      "loss": 1.0368,
      "step": 49690
    },
    {
      "epoch": 2.0137355402038044,
      "grad_norm": 0.7453382015228271,
      "learning_rate": 4.881581458339376e-05,
      "loss": 1.0362,
      "step": 49700
    },
    {
      "epoch": 2.014140718380908,
      "grad_norm": 0.7703766822814941,
      "learning_rate": 4.877934293869988e-05,
      "loss": 1.0433,
      "step": 49710
    },
    {
      "epoch": 2.0145458965580114,
      "grad_norm": 0.6077163815498352,
      "learning_rate": 4.8742880528196535e-05,
      "loss": 1.0459,
      "step": 49720
    },
    {
      "epoch": 2.014951074735115,
      "grad_norm": 0.6270671486854553,
      "learning_rate": 4.870642735845713e-05,
      "loss": 1.0282,
      "step": 49730
    },
    {
      "epoch": 2.015356252912218,
      "grad_norm": 0.597611665725708,
      "learning_rate": 4.866998343605371e-05,
      "loss": 1.0925,
      "step": 49740
    },
    {
      "epoch": 2.0157614310893215,
      "grad_norm": 0.7107589840888977,
      "learning_rate": 4.863354876755637e-05,
      "loss": 1.0264,
      "step": 49750
    },
    {
      "epoch": 2.016166609266425,
      "grad_norm": 0.63604736328125,
      "learning_rate": 4.8597123359533704e-05,
      "loss": 1.0605,
      "step": 49760
    },
    {
      "epoch": 2.0165717874435285,
      "grad_norm": 0.6257737278938293,
      "learning_rate": 4.8560707218552495e-05,
      "loss": 0.9951,
      "step": 49770
    },
    {
      "epoch": 2.0169769656206316,
      "grad_norm": 0.6092990636825562,
      "learning_rate": 4.852430035117799e-05,
      "loss": 1.0714,
      "step": 49780
    },
    {
      "epoch": 2.017382143797735,
      "grad_norm": 0.6728988289833069,
      "learning_rate": 4.8487902763973716e-05,
      "loss": 1.0647,
      "step": 49790
    },
    {
      "epoch": 2.0177873219748386,
      "grad_norm": 0.7449051141738892,
      "learning_rate": 4.8451514463501545e-05,
      "loss": 1.0582,
      "step": 49800
    },
    {
      "epoch": 2.0181925001519416,
      "grad_norm": 0.6695595383644104,
      "learning_rate": 4.841513545632158e-05,
      "loss": 1.0307,
      "step": 49810
    },
    {
      "epoch": 2.018597678329045,
      "grad_norm": 0.6329771876335144,
      "learning_rate": 4.837876574899237e-05,
      "loss": 1.0414,
      "step": 49820
    },
    {
      "epoch": 2.0190028565061486,
      "grad_norm": 0.6883831024169922,
      "learning_rate": 4.8342405348070775e-05,
      "loss": 1.0254,
      "step": 49830
    },
    {
      "epoch": 2.019408034683252,
      "grad_norm": 0.6043116450309753,
      "learning_rate": 4.8306054260111854e-05,
      "loss": 1.0177,
      "step": 49840
    },
    {
      "epoch": 2.019813212860355,
      "grad_norm": 0.5742360949516296,
      "learning_rate": 4.826971249166912e-05,
      "loss": 1.0572,
      "step": 49850
    },
    {
      "epoch": 2.0202183910374587,
      "grad_norm": 0.6723132729530334,
      "learning_rate": 4.823338004929434e-05,
      "loss": 1.0834,
      "step": 49860
    },
    {
      "epoch": 2.020623569214562,
      "grad_norm": 0.615462064743042,
      "learning_rate": 4.8197056939537686e-05,
      "loss": 1.031,
      "step": 49870
    },
    {
      "epoch": 2.0210287473916653,
      "grad_norm": 0.6577534079551697,
      "learning_rate": 4.8160743168947496e-05,
      "loss": 1.084,
      "step": 49880
    },
    {
      "epoch": 2.0214339255687688,
      "grad_norm": 0.6892037391662598,
      "learning_rate": 4.812443874407059e-05,
      "loss": 1.032,
      "step": 49890
    },
    {
      "epoch": 2.0218391037458723,
      "grad_norm": 0.590609073638916,
      "learning_rate": 4.808814367145188e-05,
      "loss": 1.0566,
      "step": 49900
    },
    {
      "epoch": 2.022244281922976,
      "grad_norm": 0.6360458731651306,
      "learning_rate": 4.8051857957634914e-05,
      "loss": 1.0182,
      "step": 49910
    },
    {
      "epoch": 2.022649460100079,
      "grad_norm": 0.5844234228134155,
      "learning_rate": 4.801558160916125e-05,
      "loss": 1.0363,
      "step": 49920
    },
    {
      "epoch": 2.0230546382771823,
      "grad_norm": 0.6259726285934448,
      "learning_rate": 4.797931463257096e-05,
      "loss": 1.05,
      "step": 49930
    },
    {
      "epoch": 2.023459816454286,
      "grad_norm": 0.5752435922622681,
      "learning_rate": 4.794305703440227e-05,
      "loss": 1.071,
      "step": 49940
    },
    {
      "epoch": 2.0238649946313894,
      "grad_norm": 0.6056144833564758,
      "learning_rate": 4.790680882119183e-05,
      "loss": 1.0687,
      "step": 49950
    },
    {
      "epoch": 2.0242701728084924,
      "grad_norm": 0.6565794944763184,
      "learning_rate": 4.787056999947455e-05,
      "loss": 1.0465,
      "step": 49960
    },
    {
      "epoch": 2.024675350985596,
      "grad_norm": 0.6501371264457703,
      "learning_rate": 4.783434057578371e-05,
      "loss": 1.0501,
      "step": 49970
    },
    {
      "epoch": 2.0250805291626994,
      "grad_norm": 0.6623795628547668,
      "learning_rate": 4.779812055665076e-05,
      "loss": 1.0849,
      "step": 49980
    },
    {
      "epoch": 2.0254857073398025,
      "grad_norm": 0.6123794913291931,
      "learning_rate": 4.776190994860561e-05,
      "loss": 1.066,
      "step": 49990
    },
    {
      "epoch": 2.025890885516906,
      "grad_norm": 0.5747687816619873,
      "learning_rate": 4.772570875817632e-05,
      "loss": 1.064,
      "step": 50000
    },
    {
      "epoch": 2.0262960636940095,
      "grad_norm": 0.6510768532752991,
      "learning_rate": 4.768951699188939e-05,
      "loss": 1.0477,
      "step": 50010
    },
    {
      "epoch": 2.026701241871113,
      "grad_norm": 0.5953274965286255,
      "learning_rate": 4.765333465626955e-05,
      "loss": 1.0545,
      "step": 50020
    },
    {
      "epoch": 2.027106420048216,
      "grad_norm": 0.6724552512168884,
      "learning_rate": 4.761716175783989e-05,
      "loss": 1.0179,
      "step": 50030
    },
    {
      "epoch": 2.0275115982253196,
      "grad_norm": 0.6233845949172974,
      "learning_rate": 4.758099830312167e-05,
      "loss": 1.0646,
      "step": 50040
    },
    {
      "epoch": 2.027916776402423,
      "grad_norm": 0.6870898604393005,
      "learning_rate": 4.7544844298634604e-05,
      "loss": 1.0743,
      "step": 50050
    },
    {
      "epoch": 2.028321954579526,
      "grad_norm": 0.678706169128418,
      "learning_rate": 4.7508699750896556e-05,
      "loss": 1.0565,
      "step": 50060
    },
    {
      "epoch": 2.0287271327566296,
      "grad_norm": 0.6771438121795654,
      "learning_rate": 4.747256466642379e-05,
      "loss": 1.0564,
      "step": 50070
    },
    {
      "epoch": 2.029132310933733,
      "grad_norm": 0.6359066367149353,
      "learning_rate": 4.743643905173084e-05,
      "loss": 1.0655,
      "step": 50080
    },
    {
      "epoch": 2.0295374891108366,
      "grad_norm": 0.6083452105522156,
      "learning_rate": 4.7400322913330556e-05,
      "loss": 1.0467,
      "step": 50090
    },
    {
      "epoch": 2.0299426672879397,
      "grad_norm": 0.6345486640930176,
      "learning_rate": 4.736421625773396e-05,
      "loss": 0.9973,
      "step": 50100
    },
    {
      "epoch": 2.030347845465043,
      "grad_norm": 0.7213921546936035,
      "learning_rate": 4.7328119091450505e-05,
      "loss": 1.0832,
      "step": 50110
    },
    {
      "epoch": 2.0307530236421467,
      "grad_norm": 0.6503602266311646,
      "learning_rate": 4.729203142098789e-05,
      "loss": 1.0125,
      "step": 50120
    },
    {
      "epoch": 2.03115820181925,
      "grad_norm": 0.5897327065467834,
      "learning_rate": 4.72559532528521e-05,
      "loss": 1.0564,
      "step": 50130
    },
    {
      "epoch": 2.0315633799963533,
      "grad_norm": 0.7146672606468201,
      "learning_rate": 4.721988459354736e-05,
      "loss": 1.0306,
      "step": 50140
    },
    {
      "epoch": 2.031968558173457,
      "grad_norm": 0.6283195614814758,
      "learning_rate": 4.7183825449576224e-05,
      "loss": 1.048,
      "step": 50150
    },
    {
      "epoch": 2.0323737363505603,
      "grad_norm": 0.6608067154884338,
      "learning_rate": 4.714777582743958e-05,
      "loss": 1.0749,
      "step": 50160
    },
    {
      "epoch": 2.0327789145276633,
      "grad_norm": 0.7020040154457092,
      "learning_rate": 4.7111735733636466e-05,
      "loss": 1.0386,
      "step": 50170
    },
    {
      "epoch": 2.033184092704767,
      "grad_norm": 0.6634446978569031,
      "learning_rate": 4.707570517466432e-05,
      "loss": 1.0199,
      "step": 50180
    },
    {
      "epoch": 2.0335892708818704,
      "grad_norm": 0.642899751663208,
      "learning_rate": 4.7039684157018827e-05,
      "loss": 1.0328,
      "step": 50190
    },
    {
      "epoch": 2.033994449058974,
      "grad_norm": 0.6114769577980042,
      "learning_rate": 4.700367268719399e-05,
      "loss": 1.0552,
      "step": 50200
    },
    {
      "epoch": 2.034399627236077,
      "grad_norm": 0.5710752010345459,
      "learning_rate": 4.696767077168196e-05,
      "loss": 1.0393,
      "step": 50210
    },
    {
      "epoch": 2.0348048054131804,
      "grad_norm": 0.6445443630218506,
      "learning_rate": 4.693167841697334e-05,
      "loss": 1.0007,
      "step": 50220
    },
    {
      "epoch": 2.035209983590284,
      "grad_norm": 0.6288264989852905,
      "learning_rate": 4.689569562955684e-05,
      "loss": 1.0357,
      "step": 50230
    },
    {
      "epoch": 2.035615161767387,
      "grad_norm": 0.6683019995689392,
      "learning_rate": 4.685972241591956e-05,
      "loss": 1.0319,
      "step": 50240
    },
    {
      "epoch": 2.0360203399444905,
      "grad_norm": 0.626274049282074,
      "learning_rate": 4.6823758782546854e-05,
      "loss": 0.9972,
      "step": 50250
    },
    {
      "epoch": 2.036425518121594,
      "grad_norm": 0.6485503315925598,
      "learning_rate": 4.678780473592238e-05,
      "loss": 1.0596,
      "step": 50260
    },
    {
      "epoch": 2.0368306962986975,
      "grad_norm": 0.6087458729743958,
      "learning_rate": 4.675186028252795e-05,
      "loss": 1.0561,
      "step": 50270
    },
    {
      "epoch": 2.0372358744758006,
      "grad_norm": 0.5699627995491028,
      "learning_rate": 4.6715925428843775e-05,
      "loss": 1.0356,
      "step": 50280
    },
    {
      "epoch": 2.037641052652904,
      "grad_norm": 0.584799587726593,
      "learning_rate": 4.6680000181348196e-05,
      "loss": 1.062,
      "step": 50290
    },
    {
      "epoch": 2.0380462308300076,
      "grad_norm": 0.6917412281036377,
      "learning_rate": 4.6644084546518044e-05,
      "loss": 1.0337,
      "step": 50300
    },
    {
      "epoch": 2.038451409007111,
      "grad_norm": 0.7592881917953491,
      "learning_rate": 4.6608178530828174e-05,
      "loss": 1.0682,
      "step": 50310
    },
    {
      "epoch": 2.038856587184214,
      "grad_norm": 0.6134333610534668,
      "learning_rate": 4.657228214075188e-05,
      "loss": 1.0274,
      "step": 50320
    },
    {
      "epoch": 2.0392617653613176,
      "grad_norm": 0.727453351020813,
      "learning_rate": 4.653639538276058e-05,
      "loss": 1.0466,
      "step": 50330
    },
    {
      "epoch": 2.039666943538421,
      "grad_norm": 0.6605948209762573,
      "learning_rate": 4.650051826332412e-05,
      "loss": 1.0642,
      "step": 50340
    },
    {
      "epoch": 2.040072121715524,
      "grad_norm": 0.6040688157081604,
      "learning_rate": 4.646465078891039e-05,
      "loss": 1.0398,
      "step": 50350
    },
    {
      "epoch": 2.0404772998926277,
      "grad_norm": 0.6581994295120239,
      "learning_rate": 4.642879296598583e-05,
      "loss": 1.0227,
      "step": 50360
    },
    {
      "epoch": 2.040882478069731,
      "grad_norm": 0.6457988619804382,
      "learning_rate": 4.639294480101487e-05,
      "loss": 1.0661,
      "step": 50370
    },
    {
      "epoch": 2.0412876562468347,
      "grad_norm": 0.6158961057662964,
      "learning_rate": 4.6357106300460374e-05,
      "loss": 1.043,
      "step": 50380
    },
    {
      "epoch": 2.041692834423938,
      "grad_norm": 0.6090956926345825,
      "learning_rate": 4.632127747078332e-05,
      "loss": 1.0601,
      "step": 50390
    },
    {
      "epoch": 2.0420980126010413,
      "grad_norm": 0.6499624848365784,
      "learning_rate": 4.6285458318443095e-05,
      "loss": 1.0475,
      "step": 50400
    },
    {
      "epoch": 2.042503190778145,
      "grad_norm": 0.7559104561805725,
      "learning_rate": 4.6249648849897175e-05,
      "loss": 1.0628,
      "step": 50410
    },
    {
      "epoch": 2.042908368955248,
      "grad_norm": 0.5623495578765869,
      "learning_rate": 4.621384907160151e-05,
      "loss": 1.0454,
      "step": 50420
    },
    {
      "epoch": 2.0433135471323514,
      "grad_norm": 0.6444956660270691,
      "learning_rate": 4.617805899001008e-05,
      "loss": 1.0463,
      "step": 50430
    },
    {
      "epoch": 2.043718725309455,
      "grad_norm": 0.6604070067405701,
      "learning_rate": 4.614227861157527e-05,
      "loss": 1.054,
      "step": 50440
    },
    {
      "epoch": 2.0441239034865584,
      "grad_norm": 0.579590916633606,
      "learning_rate": 4.610650794274759e-05,
      "loss": 1.0657,
      "step": 50450
    },
    {
      "epoch": 2.0445290816636614,
      "grad_norm": 0.7100899815559387,
      "learning_rate": 4.607074698997591e-05,
      "loss": 1.0443,
      "step": 50460
    },
    {
      "epoch": 2.044934259840765,
      "grad_norm": 0.7453498244285583,
      "learning_rate": 4.60349957597073e-05,
      "loss": 1.0604,
      "step": 50470
    },
    {
      "epoch": 2.0453394380178684,
      "grad_norm": 0.5705990195274353,
      "learning_rate": 4.599925425838713e-05,
      "loss": 1.0969,
      "step": 50480
    },
    {
      "epoch": 2.045744616194972,
      "grad_norm": 0.6222087740898132,
      "learning_rate": 4.596352249245888e-05,
      "loss": 1.0252,
      "step": 50490
    },
    {
      "epoch": 2.046149794372075,
      "grad_norm": 0.6947829723358154,
      "learning_rate": 4.59278004683644e-05,
      "loss": 1.0453,
      "step": 50500
    },
    {
      "epoch": 2.0465549725491785,
      "grad_norm": 0.6178503632545471,
      "learning_rate": 4.589208819254379e-05,
      "loss": 1.0347,
      "step": 50510
    },
    {
      "epoch": 2.046960150726282,
      "grad_norm": 0.5890249609947205,
      "learning_rate": 4.585638567143529e-05,
      "loss": 1.047,
      "step": 50520
    },
    {
      "epoch": 2.047365328903385,
      "grad_norm": 0.5661311745643616,
      "learning_rate": 4.582069291147546e-05,
      "loss": 1.0706,
      "step": 50530
    },
    {
      "epoch": 2.0477705070804886,
      "grad_norm": 0.7229082584381104,
      "learning_rate": 4.578500991909909e-05,
      "loss": 1.0601,
      "step": 50540
    },
    {
      "epoch": 2.048175685257592,
      "grad_norm": 0.7094923853874207,
      "learning_rate": 4.574933670073922e-05,
      "loss": 1.0712,
      "step": 50550
    },
    {
      "epoch": 2.0485808634346956,
      "grad_norm": 0.6592031121253967,
      "learning_rate": 4.571367326282706e-05,
      "loss": 1.0405,
      "step": 50560
    },
    {
      "epoch": 2.0489860416117986,
      "grad_norm": 0.6858596205711365,
      "learning_rate": 4.5678019611792136e-05,
      "loss": 1.0727,
      "step": 50570
    },
    {
      "epoch": 2.049391219788902,
      "grad_norm": 0.6260494589805603,
      "learning_rate": 4.564237575406217e-05,
      "loss": 1.0191,
      "step": 50580
    },
    {
      "epoch": 2.0497963979660057,
      "grad_norm": 0.5901452302932739,
      "learning_rate": 4.560674169606317e-05,
      "loss": 1.0412,
      "step": 50590
    },
    {
      "epoch": 2.0502015761431087,
      "grad_norm": 0.7516118884086609,
      "learning_rate": 4.557111744421927e-05,
      "loss": 1.0208,
      "step": 50600
    },
    {
      "epoch": 2.050606754320212,
      "grad_norm": 0.6694421768188477,
      "learning_rate": 4.5535503004952954e-05,
      "loss": 1.0368,
      "step": 50610
    },
    {
      "epoch": 2.0510119324973157,
      "grad_norm": 0.6351875066757202,
      "learning_rate": 4.5499898384684827e-05,
      "loss": 1.0595,
      "step": 50620
    },
    {
      "epoch": 2.0514171106744192,
      "grad_norm": 0.6554293036460876,
      "learning_rate": 4.54643035898338e-05,
      "loss": 1.0559,
      "step": 50630
    },
    {
      "epoch": 2.0518222888515223,
      "grad_norm": 0.619621217250824,
      "learning_rate": 4.5428718626817015e-05,
      "loss": 1.042,
      "step": 50640
    },
    {
      "epoch": 2.052227467028626,
      "grad_norm": 0.5921615362167358,
      "learning_rate": 4.5393143502049853e-05,
      "loss": 1.0967,
      "step": 50650
    },
    {
      "epoch": 2.0526326452057293,
      "grad_norm": 0.6304709315299988,
      "learning_rate": 4.5357578221945794e-05,
      "loss": 1.0705,
      "step": 50660
    },
    {
      "epoch": 2.053037823382833,
      "grad_norm": 0.6376930475234985,
      "learning_rate": 4.532202279291673e-05,
      "loss": 1.0285,
      "step": 50670
    },
    {
      "epoch": 2.053443001559936,
      "grad_norm": 0.675715982913971,
      "learning_rate": 4.52864772213726e-05,
      "loss": 1.0516,
      "step": 50680
    },
    {
      "epoch": 2.0538481797370394,
      "grad_norm": 0.6740050315856934,
      "learning_rate": 4.52509415137217e-05,
      "loss": 1.0304,
      "step": 50690
    },
    {
      "epoch": 2.054253357914143,
      "grad_norm": 0.608293890953064,
      "learning_rate": 4.521541567637049e-05,
      "loss": 1.0232,
      "step": 50700
    },
    {
      "epoch": 2.054658536091246,
      "grad_norm": 0.7630971670150757,
      "learning_rate": 4.517989971572369e-05,
      "loss": 1.0493,
      "step": 50710
    },
    {
      "epoch": 2.0550637142683494,
      "grad_norm": 0.6882697343826294,
      "learning_rate": 4.514439363818415e-05,
      "loss": 1.0573,
      "step": 50720
    },
    {
      "epoch": 2.055468892445453,
      "grad_norm": 0.5928632616996765,
      "learning_rate": 4.5108897450153054e-05,
      "loss": 1.0335,
      "step": 50730
    },
    {
      "epoch": 2.0558740706225564,
      "grad_norm": 0.6118112206459045,
      "learning_rate": 4.5073411158029634e-05,
      "loss": 1.0026,
      "step": 50740
    },
    {
      "epoch": 2.0562792487996595,
      "grad_norm": 0.6412435173988342,
      "learning_rate": 4.503793476821161e-05,
      "loss": 1.0233,
      "step": 50750
    },
    {
      "epoch": 2.056684426976763,
      "grad_norm": 0.5953876376152039,
      "learning_rate": 4.5002468287094635e-05,
      "loss": 1.026,
      "step": 50760
    },
    {
      "epoch": 2.0570896051538665,
      "grad_norm": 0.5889370441436768,
      "learning_rate": 4.496701172107277e-05,
      "loss": 1.0702,
      "step": 50770
    },
    {
      "epoch": 2.0574947833309696,
      "grad_norm": 0.6288536190986633,
      "learning_rate": 4.4931565076538154e-05,
      "loss": 1.0433,
      "step": 50780
    },
    {
      "epoch": 2.057899961508073,
      "grad_norm": 0.6103301644325256,
      "learning_rate": 4.4896128359881215e-05,
      "loss": 1.0479,
      "step": 50790
    },
    {
      "epoch": 2.0583051396851766,
      "grad_norm": 0.6946113109588623,
      "learning_rate": 4.4860701577490595e-05,
      "loss": 1.0955,
      "step": 50800
    },
    {
      "epoch": 2.05871031786228,
      "grad_norm": 0.6107717752456665,
      "learning_rate": 4.482528473575315e-05,
      "loss": 1.0641,
      "step": 50810
    },
    {
      "epoch": 2.059115496039383,
      "grad_norm": 0.5936039686203003,
      "learning_rate": 4.478987784105384e-05,
      "loss": 1.0401,
      "step": 50820
    },
    {
      "epoch": 2.0595206742164867,
      "grad_norm": 0.6572657823562622,
      "learning_rate": 4.475448089977595e-05,
      "loss": 1.0255,
      "step": 50830
    },
    {
      "epoch": 2.05992585239359,
      "grad_norm": 0.6349802017211914,
      "learning_rate": 4.471909391830098e-05,
      "loss": 1.0696,
      "step": 50840
    },
    {
      "epoch": 2.0603310305706937,
      "grad_norm": 0.6630573272705078,
      "learning_rate": 4.46837169030085e-05,
      "loss": 1.0451,
      "step": 50850
    },
    {
      "epoch": 2.0607362087477967,
      "grad_norm": 0.6794646382331848,
      "learning_rate": 4.464834986027641e-05,
      "loss": 1.0178,
      "step": 50860
    },
    {
      "epoch": 2.0611413869249002,
      "grad_norm": 0.6787291765213013,
      "learning_rate": 4.461299279648077e-05,
      "loss": 1.0557,
      "step": 50870
    },
    {
      "epoch": 2.0615465651020037,
      "grad_norm": 0.6530352830886841,
      "learning_rate": 4.457764571799586e-05,
      "loss": 1.0521,
      "step": 50880
    },
    {
      "epoch": 2.061951743279107,
      "grad_norm": 0.5974339842796326,
      "learning_rate": 4.454230863119411e-05,
      "loss": 1.0503,
      "step": 50890
    },
    {
      "epoch": 2.0623569214562103,
      "grad_norm": 0.6823073625564575,
      "learning_rate": 4.45069815424462e-05,
      "loss": 1.0352,
      "step": 50900
    },
    {
      "epoch": 2.062762099633314,
      "grad_norm": 0.6281534433364868,
      "learning_rate": 4.447166445812092e-05,
      "loss": 1.0457,
      "step": 50910
    },
    {
      "epoch": 2.0631672778104173,
      "grad_norm": 0.6040210127830505,
      "learning_rate": 4.443635738458546e-05,
      "loss": 1.0441,
      "step": 50920
    },
    {
      "epoch": 2.0635724559875204,
      "grad_norm": 0.6312506794929504,
      "learning_rate": 4.4401060328204946e-05,
      "loss": 1.0372,
      "step": 50930
    },
    {
      "epoch": 2.063977634164624,
      "grad_norm": 0.7646352052688599,
      "learning_rate": 4.436577329534291e-05,
      "loss": 1.0559,
      "step": 50940
    },
    {
      "epoch": 2.0643828123417274,
      "grad_norm": 0.6856712698936462,
      "learning_rate": 4.4330496292360904e-05,
      "loss": 1.0712,
      "step": 50950
    },
    {
      "epoch": 2.0647879905188304,
      "grad_norm": 0.6470664739608765,
      "learning_rate": 4.429522932561884e-05,
      "loss": 1.0159,
      "step": 50960
    },
    {
      "epoch": 2.065193168695934,
      "grad_norm": 0.5777468085289001,
      "learning_rate": 4.425997240147461e-05,
      "loss": 1.0671,
      "step": 50970
    },
    {
      "epoch": 2.0655983468730374,
      "grad_norm": 0.6198233962059021,
      "learning_rate": 4.422472552628458e-05,
      "loss": 1.0504,
      "step": 50980
    },
    {
      "epoch": 2.066003525050141,
      "grad_norm": 0.7129728198051453,
      "learning_rate": 4.4189488706403037e-05,
      "loss": 1.0528,
      "step": 50990
    },
    {
      "epoch": 2.066408703227244,
      "grad_norm": 0.7443998456001282,
      "learning_rate": 4.415426194818264e-05,
      "loss": 1.0521,
      "step": 51000
    },
    {
      "epoch": 2.0668138814043475,
      "grad_norm": 0.5639276504516602,
      "learning_rate": 4.411904525797408e-05,
      "loss": 1.0807,
      "step": 51010
    },
    {
      "epoch": 2.067219059581451,
      "grad_norm": 0.716392457485199,
      "learning_rate": 4.4083838642126384e-05,
      "loss": 1.0313,
      "step": 51020
    },
    {
      "epoch": 2.0676242377585545,
      "grad_norm": 0.7051234841346741,
      "learning_rate": 4.4048642106986596e-05,
      "loss": 1.0089,
      "step": 51030
    },
    {
      "epoch": 2.0680294159356576,
      "grad_norm": 0.6271855235099792,
      "learning_rate": 4.401345565890015e-05,
      "loss": 1.0552,
      "step": 51040
    },
    {
      "epoch": 2.068434594112761,
      "grad_norm": 0.6698451042175293,
      "learning_rate": 4.397827930421047e-05,
      "loss": 1.0742,
      "step": 51050
    },
    {
      "epoch": 2.0688397722898646,
      "grad_norm": 0.659977376461029,
      "learning_rate": 4.39431130492593e-05,
      "loss": 1.0407,
      "step": 51060
    },
    {
      "epoch": 2.0692449504669677,
      "grad_norm": 0.6594642996788025,
      "learning_rate": 4.390795690038643e-05,
      "loss": 1.0541,
      "step": 51070
    },
    {
      "epoch": 2.069650128644071,
      "grad_norm": 0.6392821669578552,
      "learning_rate": 4.387281086392994e-05,
      "loss": 1.0569,
      "step": 51080
    },
    {
      "epoch": 2.0700553068211747,
      "grad_norm": 0.7884118556976318,
      "learning_rate": 4.3837674946226026e-05,
      "loss": 1.0587,
      "step": 51090
    },
    {
      "epoch": 2.070460484998278,
      "grad_norm": 0.6821327805519104,
      "learning_rate": 4.3802549153609154e-05,
      "loss": 1.0288,
      "step": 51100
    },
    {
      "epoch": 2.0708656631753812,
      "grad_norm": 0.5861783623695374,
      "learning_rate": 4.376743349241179e-05,
      "loss": 1.1209,
      "step": 51110
    },
    {
      "epoch": 2.0712708413524847,
      "grad_norm": 0.6561359763145447,
      "learning_rate": 4.3732327968964724e-05,
      "loss": 1.036,
      "step": 51120
    },
    {
      "epoch": 2.0716760195295882,
      "grad_norm": 0.608099102973938,
      "learning_rate": 4.369723258959691e-05,
      "loss": 1.0461,
      "step": 51130
    },
    {
      "epoch": 2.0720811977066917,
      "grad_norm": 0.6305306553840637,
      "learning_rate": 4.366214736063533e-05,
      "loss": 1.0536,
      "step": 51140
    },
    {
      "epoch": 2.072486375883795,
      "grad_norm": 0.6338403224945068,
      "learning_rate": 4.362707228840531e-05,
      "loss": 1.0373,
      "step": 51150
    },
    {
      "epoch": 2.0728915540608983,
      "grad_norm": 0.6415111422538757,
      "learning_rate": 4.359200737923026e-05,
      "loss": 1.0932,
      "step": 51160
    },
    {
      "epoch": 2.073296732238002,
      "grad_norm": 0.6465750336647034,
      "learning_rate": 4.3556952639431804e-05,
      "loss": 1.0352,
      "step": 51170
    },
    {
      "epoch": 2.073701910415105,
      "grad_norm": 0.62632817029953,
      "learning_rate": 4.352190807532963e-05,
      "loss": 1.0634,
      "step": 51180
    },
    {
      "epoch": 2.0741070885922084,
      "grad_norm": 0.6901636719703674,
      "learning_rate": 4.3486873693241706e-05,
      "loss": 1.0326,
      "step": 51190
    },
    {
      "epoch": 2.074512266769312,
      "grad_norm": 0.6546501517295837,
      "learning_rate": 4.34518494994841e-05,
      "loss": 1.0245,
      "step": 51200
    },
    {
      "epoch": 2.0749174449464154,
      "grad_norm": 0.6501778364181519,
      "learning_rate": 4.341683550037111e-05,
      "loss": 1.0597,
      "step": 51210
    },
    {
      "epoch": 2.0753226231235185,
      "grad_norm": 0.5690070986747742,
      "learning_rate": 4.3381831702215084e-05,
      "loss": 1.0563,
      "step": 51220
    },
    {
      "epoch": 2.075727801300622,
      "grad_norm": 0.6688877940177917,
      "learning_rate": 4.334683811132666e-05,
      "loss": 1.0151,
      "step": 51230
    },
    {
      "epoch": 2.0761329794777255,
      "grad_norm": 0.664387583732605,
      "learning_rate": 4.33118547340145e-05,
      "loss": 1.0606,
      "step": 51240
    },
    {
      "epoch": 2.0765381576548285,
      "grad_norm": 0.6513673067092896,
      "learning_rate": 4.327688157658554e-05,
      "loss": 1.0551,
      "step": 51250
    },
    {
      "epoch": 2.076943335831932,
      "grad_norm": 0.7085045576095581,
      "learning_rate": 4.324191864534481e-05,
      "loss": 1.0674,
      "step": 51260
    },
    {
      "epoch": 2.0773485140090355,
      "grad_norm": 0.6950013041496277,
      "learning_rate": 4.320696594659557e-05,
      "loss": 1.0465,
      "step": 51270
    },
    {
      "epoch": 2.077753692186139,
      "grad_norm": 0.697545051574707,
      "learning_rate": 4.31720234866391e-05,
      "loss": 1.064,
      "step": 51280
    },
    {
      "epoch": 2.078158870363242,
      "grad_norm": 0.5620270371437073,
      "learning_rate": 4.3137091271775e-05,
      "loss": 1.0326,
      "step": 51290
    },
    {
      "epoch": 2.0785640485403456,
      "grad_norm": 0.6431417465209961,
      "learning_rate": 4.3102169308300854e-05,
      "loss": 1.0563,
      "step": 51300
    },
    {
      "epoch": 2.078969226717449,
      "grad_norm": 0.6444371938705444,
      "learning_rate": 4.3067257602512536e-05,
      "loss": 1.0608,
      "step": 51310
    },
    {
      "epoch": 2.079374404894552,
      "grad_norm": 0.7244434952735901,
      "learning_rate": 4.303235616070399e-05,
      "loss": 1.035,
      "step": 51320
    },
    {
      "epoch": 2.0797795830716557,
      "grad_norm": 0.6950321793556213,
      "learning_rate": 4.299746498916739e-05,
      "loss": 1.0485,
      "step": 51330
    },
    {
      "epoch": 2.080184761248759,
      "grad_norm": 0.6827163696289062,
      "learning_rate": 4.2962584094192934e-05,
      "loss": 1.0509,
      "step": 51340
    },
    {
      "epoch": 2.0805899394258627,
      "grad_norm": 0.7152417302131653,
      "learning_rate": 4.292771348206911e-05,
      "loss": 1.0292,
      "step": 51350
    },
    {
      "epoch": 2.0809951176029657,
      "grad_norm": 0.703732430934906,
      "learning_rate": 4.289285315908237e-05,
      "loss": 1.0601,
      "step": 51360
    },
    {
      "epoch": 2.0814002957800692,
      "grad_norm": 0.6503881216049194,
      "learning_rate": 4.285800313151756e-05,
      "loss": 1.0528,
      "step": 51370
    },
    {
      "epoch": 2.0818054739571727,
      "grad_norm": 0.6066603660583496,
      "learning_rate": 4.282316340565744e-05,
      "loss": 1.0583,
      "step": 51380
    },
    {
      "epoch": 2.0822106521342763,
      "grad_norm": 0.6550276875495911,
      "learning_rate": 4.278833398778306e-05,
      "loss": 1.0831,
      "step": 51390
    },
    {
      "epoch": 2.0826158303113793,
      "grad_norm": 0.6032612323760986,
      "learning_rate": 4.275351488417349e-05,
      "loss": 1.0392,
      "step": 51400
    },
    {
      "epoch": 2.083021008488483,
      "grad_norm": 0.6965372562408447,
      "learning_rate": 4.271870610110608e-05,
      "loss": 1.0143,
      "step": 51410
    },
    {
      "epoch": 2.0834261866655863,
      "grad_norm": 0.7444655895233154,
      "learning_rate": 4.2683907644856136e-05,
      "loss": 1.0925,
      "step": 51420
    },
    {
      "epoch": 2.0838313648426894,
      "grad_norm": 0.5950588583946228,
      "learning_rate": 4.264911952169735e-05,
      "loss": 1.0473,
      "step": 51430
    },
    {
      "epoch": 2.084236543019793,
      "grad_norm": 0.6177449226379395,
      "learning_rate": 4.261434173790133e-05,
      "loss": 1.0352,
      "step": 51440
    },
    {
      "epoch": 2.0846417211968964,
      "grad_norm": 0.6096029281616211,
      "learning_rate": 4.257957429973795e-05,
      "loss": 1.0676,
      "step": 51450
    },
    {
      "epoch": 2.085046899374,
      "grad_norm": 0.6557916402816772,
      "learning_rate": 4.254481721347511e-05,
      "loss": 1.0779,
      "step": 51460
    },
    {
      "epoch": 2.085452077551103,
      "grad_norm": 0.6681409478187561,
      "learning_rate": 4.251007048537894e-05,
      "loss": 1.0473,
      "step": 51470
    },
    {
      "epoch": 2.0858572557282065,
      "grad_norm": 0.6188157200813293,
      "learning_rate": 4.2475334121713664e-05,
      "loss": 1.0698,
      "step": 51480
    },
    {
      "epoch": 2.08626243390531,
      "grad_norm": 0.6510903835296631,
      "learning_rate": 4.2440608128741686e-05,
      "loss": 1.0403,
      "step": 51490
    },
    {
      "epoch": 2.0866676120824135,
      "grad_norm": 0.6046815514564514,
      "learning_rate": 4.240589251272342e-05,
      "loss": 1.0702,
      "step": 51500
    },
    {
      "epoch": 2.0870727902595165,
      "grad_norm": 0.6202812790870667,
      "learning_rate": 4.237118727991754e-05,
      "loss": 1.0275,
      "step": 51510
    },
    {
      "epoch": 2.08747796843662,
      "grad_norm": 0.7176526188850403,
      "learning_rate": 4.2336492436580786e-05,
      "loss": 1.0637,
      "step": 51520
    },
    {
      "epoch": 2.0878831466137235,
      "grad_norm": 0.6155521869659424,
      "learning_rate": 4.230180798896801e-05,
      "loss": 1.0172,
      "step": 51530
    },
    {
      "epoch": 2.0882883247908266,
      "grad_norm": 0.6973652243614197,
      "learning_rate": 4.2267133943332215e-05,
      "loss": 1.0875,
      "step": 51540
    },
    {
      "epoch": 2.08869350296793,
      "grad_norm": 0.6125881671905518,
      "learning_rate": 4.223247030592454e-05,
      "loss": 1.0101,
      "step": 51550
    },
    {
      "epoch": 2.0890986811450336,
      "grad_norm": 0.6459801197052002,
      "learning_rate": 4.219781708299427e-05,
      "loss": 1.0443,
      "step": 51560
    },
    {
      "epoch": 2.089503859322137,
      "grad_norm": 0.6852020621299744,
      "learning_rate": 4.2163174280788697e-05,
      "loss": 1.031,
      "step": 51570
    },
    {
      "epoch": 2.08990903749924,
      "grad_norm": 0.6370081901550293,
      "learning_rate": 4.21285419055534e-05,
      "loss": 1.0453,
      "step": 51580
    },
    {
      "epoch": 2.0903142156763437,
      "grad_norm": 0.6452752351760864,
      "learning_rate": 4.209391996353187e-05,
      "loss": 1.0778,
      "step": 51590
    },
    {
      "epoch": 2.090719393853447,
      "grad_norm": 0.6531192064285278,
      "learning_rate": 4.205930846096598e-05,
      "loss": 1.0275,
      "step": 51600
    },
    {
      "epoch": 2.0911245720305502,
      "grad_norm": 0.6868224740028381,
      "learning_rate": 4.202470740409549e-05,
      "loss": 1.0486,
      "step": 51610
    },
    {
      "epoch": 2.0915297502076537,
      "grad_norm": 0.628913164138794,
      "learning_rate": 4.199011679915842e-05,
      "loss": 1.1052,
      "step": 51620
    },
    {
      "epoch": 2.0919349283847573,
      "grad_norm": 0.6242732405662537,
      "learning_rate": 4.1955536652390793e-05,
      "loss": 1.0343,
      "step": 51630
    },
    {
      "epoch": 2.0923401065618608,
      "grad_norm": 0.653647780418396,
      "learning_rate": 4.192096697002686e-05,
      "loss": 1.0274,
      "step": 51640
    },
    {
      "epoch": 2.092745284738964,
      "grad_norm": 0.5533059239387512,
      "learning_rate": 4.188640775829885e-05,
      "loss": 1.0933,
      "step": 51650
    },
    {
      "epoch": 2.0931504629160673,
      "grad_norm": 0.668385922908783,
      "learning_rate": 4.185185902343731e-05,
      "loss": 1.0586,
      "step": 51660
    },
    {
      "epoch": 2.093555641093171,
      "grad_norm": 0.6032310128211975,
      "learning_rate": 4.1817320771670676e-05,
      "loss": 1.0421,
      "step": 51670
    },
    {
      "epoch": 2.093960819270274,
      "grad_norm": 0.5834062695503235,
      "learning_rate": 4.178279300922566e-05,
      "loss": 1.0791,
      "step": 51680
    },
    {
      "epoch": 2.0943659974473774,
      "grad_norm": 0.6006143689155579,
      "learning_rate": 4.1748275742326935e-05,
      "loss": 1.0601,
      "step": 51690
    },
    {
      "epoch": 2.094771175624481,
      "grad_norm": 0.6799567341804504,
      "learning_rate": 4.171376897719741e-05,
      "loss": 1.0182,
      "step": 51700
    },
    {
      "epoch": 2.0951763538015844,
      "grad_norm": 0.6284598708152771,
      "learning_rate": 4.167927272005805e-05,
      "loss": 1.0539,
      "step": 51710
    },
    {
      "epoch": 2.0955815319786875,
      "grad_norm": 0.7156328558921814,
      "learning_rate": 4.164478697712797e-05,
      "loss": 1.0476,
      "step": 51720
    },
    {
      "epoch": 2.095986710155791,
      "grad_norm": 0.5961900949478149,
      "learning_rate": 4.1610311754624255e-05,
      "loss": 1.0529,
      "step": 51730
    },
    {
      "epoch": 2.0963918883328945,
      "grad_norm": 0.5663501620292664,
      "learning_rate": 4.157584705876227e-05,
      "loss": 1.055,
      "step": 51740
    },
    {
      "epoch": 2.096797066509998,
      "grad_norm": 0.5877333283424377,
      "learning_rate": 4.154139289575534e-05,
      "loss": 1.0748,
      "step": 51750
    },
    {
      "epoch": 2.097202244687101,
      "grad_norm": 0.6413207650184631,
      "learning_rate": 4.1506949271814954e-05,
      "loss": 1.0495,
      "step": 51760
    },
    {
      "epoch": 2.0976074228642045,
      "grad_norm": 0.6169136166572571,
      "learning_rate": 4.147251619315072e-05,
      "loss": 1.0443,
      "step": 51770
    },
    {
      "epoch": 2.098012601041308,
      "grad_norm": 0.6568485498428345,
      "learning_rate": 4.143809366597037e-05,
      "loss": 1.0529,
      "step": 51780
    },
    {
      "epoch": 2.098417779218411,
      "grad_norm": 0.6289340853691101,
      "learning_rate": 4.140368169647957e-05,
      "loss": 1.0654,
      "step": 51790
    },
    {
      "epoch": 2.0988229573955146,
      "grad_norm": 0.6325077414512634,
      "learning_rate": 4.136928029088227e-05,
      "loss": 1.0642,
      "step": 51800
    },
    {
      "epoch": 2.099228135572618,
      "grad_norm": 0.7343443036079407,
      "learning_rate": 4.133488945538043e-05,
      "loss": 1.058,
      "step": 51810
    },
    {
      "epoch": 2.0996333137497216,
      "grad_norm": 0.7165772914886475,
      "learning_rate": 4.130050919617415e-05,
      "loss": 1.0465,
      "step": 51820
    },
    {
      "epoch": 2.1000384919268247,
      "grad_norm": 0.6516438722610474,
      "learning_rate": 4.126613951946152e-05,
      "loss": 1.0398,
      "step": 51830
    },
    {
      "epoch": 2.100443670103928,
      "grad_norm": 0.623694896697998,
      "learning_rate": 4.123178043143884e-05,
      "loss": 1.0645,
      "step": 51840
    },
    {
      "epoch": 2.1008488482810317,
      "grad_norm": 0.6878968477249146,
      "learning_rate": 4.119743193830048e-05,
      "loss": 1.0748,
      "step": 51850
    },
    {
      "epoch": 2.101254026458135,
      "grad_norm": 0.6184583902359009,
      "learning_rate": 4.1163094046238806e-05,
      "loss": 1.055,
      "step": 51860
    },
    {
      "epoch": 2.1016592046352383,
      "grad_norm": 0.6404325366020203,
      "learning_rate": 4.112876676144436e-05,
      "loss": 1.035,
      "step": 51870
    },
    {
      "epoch": 2.1020643828123418,
      "grad_norm": 0.7039847373962402,
      "learning_rate": 4.1094450090105773e-05,
      "loss": 1.0565,
      "step": 51880
    },
    {
      "epoch": 2.1024695609894453,
      "grad_norm": 0.6710174679756165,
      "learning_rate": 4.106014403840978e-05,
      "loss": 1.0205,
      "step": 51890
    },
    {
      "epoch": 2.1028747391665483,
      "grad_norm": 0.6256112456321716,
      "learning_rate": 4.102584861254107e-05,
      "loss": 1.0436,
      "step": 51900
    },
    {
      "epoch": 2.103279917343652,
      "grad_norm": 0.7446640729904175,
      "learning_rate": 4.099156381868261e-05,
      "loss": 1.0741,
      "step": 51910
    },
    {
      "epoch": 2.1036850955207553,
      "grad_norm": 0.5583058595657349,
      "learning_rate": 4.0957289663015255e-05,
      "loss": 1.0394,
      "step": 51920
    },
    {
      "epoch": 2.104090273697859,
      "grad_norm": 0.5981910228729248,
      "learning_rate": 4.092302615171807e-05,
      "loss": 1.0209,
      "step": 51930
    },
    {
      "epoch": 2.104495451874962,
      "grad_norm": 0.7283002138137817,
      "learning_rate": 4.0888773290968184e-05,
      "loss": 1.0636,
      "step": 51940
    },
    {
      "epoch": 2.1049006300520654,
      "grad_norm": 0.7058972120285034,
      "learning_rate": 4.085453108694083e-05,
      "loss": 1.044,
      "step": 51950
    },
    {
      "epoch": 2.105305808229169,
      "grad_norm": 0.5960850715637207,
      "learning_rate": 4.082029954580919e-05,
      "loss": 1.045,
      "step": 51960
    },
    {
      "epoch": 2.105710986406272,
      "grad_norm": 0.6962130069732666,
      "learning_rate": 4.078607867374469e-05,
      "loss": 1.072,
      "step": 51970
    },
    {
      "epoch": 2.1061161645833755,
      "grad_norm": 0.6285178661346436,
      "learning_rate": 4.0751868476916654e-05,
      "loss": 1.0374,
      "step": 51980
    },
    {
      "epoch": 2.106521342760479,
      "grad_norm": 0.6933303475379944,
      "learning_rate": 4.071766896149273e-05,
      "loss": 1.0501,
      "step": 51990
    },
    {
      "epoch": 2.1069265209375825,
      "grad_norm": 0.7046886682510376,
      "learning_rate": 4.0683480133638377e-05,
      "loss": 1.0698,
      "step": 52000
    },
    {
      "epoch": 2.1073316991146855,
      "grad_norm": 0.6633132100105286,
      "learning_rate": 4.064930199951732e-05,
      "loss": 1.0605,
      "step": 52010
    },
    {
      "epoch": 2.107736877291789,
      "grad_norm": 0.6602674126625061,
      "learning_rate": 4.061513456529121e-05,
      "loss": 1.0378,
      "step": 52020
    },
    {
      "epoch": 2.1081420554688926,
      "grad_norm": 0.6509812474250793,
      "learning_rate": 4.0580977837119904e-05,
      "loss": 1.0821,
      "step": 52030
    },
    {
      "epoch": 2.1085472336459956,
      "grad_norm": 0.6660982966423035,
      "learning_rate": 4.054683182116117e-05,
      "loss": 1.0321,
      "step": 52040
    },
    {
      "epoch": 2.108952411823099,
      "grad_norm": 0.6003744602203369,
      "learning_rate": 4.051269652357106e-05,
      "loss": 1.0586,
      "step": 52050
    },
    {
      "epoch": 2.1093575900002026,
      "grad_norm": 0.6501328945159912,
      "learning_rate": 4.0478571950503486e-05,
      "loss": 1.0467,
      "step": 52060
    },
    {
      "epoch": 2.109762768177306,
      "grad_norm": 0.6175651550292969,
      "learning_rate": 4.044445810811057e-05,
      "loss": 1.0648,
      "step": 52070
    },
    {
      "epoch": 2.110167946354409,
      "grad_norm": 0.6302654147148132,
      "learning_rate": 4.0410355002542385e-05,
      "loss": 1.0448,
      "step": 52080
    },
    {
      "epoch": 2.1105731245315127,
      "grad_norm": 0.6715889573097229,
      "learning_rate": 4.037626263994715e-05,
      "loss": 1.0507,
      "step": 52090
    },
    {
      "epoch": 2.110978302708616,
      "grad_norm": 0.6169251799583435,
      "learning_rate": 4.0342181026471124e-05,
      "loss": 1.0664,
      "step": 52100
    },
    {
      "epoch": 2.1113834808857197,
      "grad_norm": 0.6353077292442322,
      "learning_rate": 4.030811016825867e-05,
      "loss": 1.0546,
      "step": 52110
    },
    {
      "epoch": 2.1117886590628228,
      "grad_norm": 0.6797330975532532,
      "learning_rate": 4.02740500714521e-05,
      "loss": 1.0499,
      "step": 52120
    },
    {
      "epoch": 2.1121938372399263,
      "grad_norm": 0.6830065846443176,
      "learning_rate": 4.024000074219187e-05,
      "loss": 1.0755,
      "step": 52130
    },
    {
      "epoch": 2.1125990154170298,
      "grad_norm": 0.6531261205673218,
      "learning_rate": 4.020596218661653e-05,
      "loss": 1.025,
      "step": 52140
    },
    {
      "epoch": 2.113004193594133,
      "grad_norm": 0.6696150898933411,
      "learning_rate": 4.0171934410862574e-05,
      "loss": 1.0359,
      "step": 52150
    },
    {
      "epoch": 2.1134093717712363,
      "grad_norm": 0.6360255479812622,
      "learning_rate": 4.013791742106464e-05,
      "loss": 1.0287,
      "step": 52160
    },
    {
      "epoch": 2.11381454994834,
      "grad_norm": 0.6547711491584778,
      "learning_rate": 4.01039112233554e-05,
      "loss": 1.0505,
      "step": 52170
    },
    {
      "epoch": 2.1142197281254433,
      "grad_norm": 0.6801090240478516,
      "learning_rate": 4.006991582386561e-05,
      "loss": 1.0457,
      "step": 52180
    },
    {
      "epoch": 2.1146249063025464,
      "grad_norm": 0.6556650400161743,
      "learning_rate": 4.003593122872399e-05,
      "loss": 1.0077,
      "step": 52190
    },
    {
      "epoch": 2.11503008447965,
      "grad_norm": 0.6982724070549011,
      "learning_rate": 4.0001957444057426e-05,
      "loss": 1.0218,
      "step": 52200
    },
    {
      "epoch": 2.1154352626567534,
      "grad_norm": 0.6686224341392517,
      "learning_rate": 3.99679944759907e-05,
      "loss": 1.0215,
      "step": 52210
    },
    {
      "epoch": 2.115840440833857,
      "grad_norm": 0.6492426991462708,
      "learning_rate": 3.993404233064687e-05,
      "loss": 1.093,
      "step": 52220
    },
    {
      "epoch": 2.11624561901096,
      "grad_norm": 0.6036686897277832,
      "learning_rate": 3.990010101414683e-05,
      "loss": 1.0908,
      "step": 52230
    },
    {
      "epoch": 2.1166507971880635,
      "grad_norm": 0.6940036416053772,
      "learning_rate": 3.986617053260967e-05,
      "loss": 1.0369,
      "step": 52240
    },
    {
      "epoch": 2.117055975365167,
      "grad_norm": 0.5795769095420837,
      "learning_rate": 3.983225089215238e-05,
      "loss": 1.0361,
      "step": 52250
    },
    {
      "epoch": 2.11746115354227,
      "grad_norm": 0.6408687233924866,
      "learning_rate": 3.9798342098890165e-05,
      "loss": 1.0196,
      "step": 52260
    },
    {
      "epoch": 2.1178663317193736,
      "grad_norm": 0.771460771560669,
      "learning_rate": 3.976444415893608e-05,
      "loss": 1.0784,
      "step": 52270
    },
    {
      "epoch": 2.118271509896477,
      "grad_norm": 0.5978890657424927,
      "learning_rate": 3.973055707840146e-05,
      "loss": 1.0421,
      "step": 52280
    },
    {
      "epoch": 2.1186766880735806,
      "grad_norm": 0.7877952456474304,
      "learning_rate": 3.969668086339548e-05,
      "loss": 1.0479,
      "step": 52290
    },
    {
      "epoch": 2.1190818662506836,
      "grad_norm": 0.7122960686683655,
      "learning_rate": 3.9662815520025466e-05,
      "loss": 1.0763,
      "step": 52300
    },
    {
      "epoch": 2.119487044427787,
      "grad_norm": 0.6275640726089478,
      "learning_rate": 3.9628961054396696e-05,
      "loss": 1.0883,
      "step": 52310
    },
    {
      "epoch": 2.1198922226048906,
      "grad_norm": 0.6009311079978943,
      "learning_rate": 3.959511747261257e-05,
      "loss": 1.0342,
      "step": 52320
    },
    {
      "epoch": 2.1202974007819937,
      "grad_norm": 0.6398431658744812,
      "learning_rate": 3.95612847807745e-05,
      "loss": 1.0258,
      "step": 52330
    },
    {
      "epoch": 2.120702578959097,
      "grad_norm": 0.6837823390960693,
      "learning_rate": 3.952746298498195e-05,
      "loss": 1.0346,
      "step": 52340
    },
    {
      "epoch": 2.1211077571362007,
      "grad_norm": 0.6628961563110352,
      "learning_rate": 3.949365209133236e-05,
      "loss": 1.0687,
      "step": 52350
    },
    {
      "epoch": 2.121512935313304,
      "grad_norm": 0.6408852338790894,
      "learning_rate": 3.945985210592127e-05,
      "loss": 1.0397,
      "step": 52360
    },
    {
      "epoch": 2.1219181134904073,
      "grad_norm": 0.5686211585998535,
      "learning_rate": 3.9426063034842196e-05,
      "loss": 1.051,
      "step": 52370
    },
    {
      "epoch": 2.1223232916675108,
      "grad_norm": 1.052756428718567,
      "learning_rate": 3.9392284884186735e-05,
      "loss": 1.0545,
      "step": 52380
    },
    {
      "epoch": 2.1227284698446143,
      "grad_norm": 0.6480855941772461,
      "learning_rate": 3.93585176600445e-05,
      "loss": 1.0202,
      "step": 52390
    },
    {
      "epoch": 2.1231336480217173,
      "grad_norm": 0.6022703051567078,
      "learning_rate": 3.9324761368503174e-05,
      "loss": 1.0711,
      "step": 52400
    },
    {
      "epoch": 2.123538826198821,
      "grad_norm": 0.669272780418396,
      "learning_rate": 3.929101601564834e-05,
      "loss": 1.078,
      "step": 52410
    },
    {
      "epoch": 2.1239440043759243,
      "grad_norm": 0.6403139233589172,
      "learning_rate": 3.925728160756379e-05,
      "loss": 1.0639,
      "step": 52420
    },
    {
      "epoch": 2.124349182553028,
      "grad_norm": 0.6188815832138062,
      "learning_rate": 3.9223558150331116e-05,
      "loss": 1.0456,
      "step": 52430
    },
    {
      "epoch": 2.124754360730131,
      "grad_norm": 0.5761057138442993,
      "learning_rate": 3.9189845650030246e-05,
      "loss": 1.0181,
      "step": 52440
    },
    {
      "epoch": 2.1251595389072344,
      "grad_norm": 0.5718631744384766,
      "learning_rate": 3.9156144112738814e-05,
      "loss": 1.0583,
      "step": 52450
    },
    {
      "epoch": 2.125564717084338,
      "grad_norm": 0.654984712600708,
      "learning_rate": 3.912245354453271e-05,
      "loss": 1.0535,
      "step": 52460
    },
    {
      "epoch": 2.1259698952614414,
      "grad_norm": 0.6667295694351196,
      "learning_rate": 3.908877395148568e-05,
      "loss": 1.0717,
      "step": 52470
    },
    {
      "epoch": 2.1263750734385445,
      "grad_norm": 0.5989518761634827,
      "learning_rate": 3.9055105339669595e-05,
      "loss": 1.0674,
      "step": 52480
    },
    {
      "epoch": 2.126780251615648,
      "grad_norm": 0.5777643918991089,
      "learning_rate": 3.902144771515432e-05,
      "loss": 1.0223,
      "step": 52490
    },
    {
      "epoch": 2.1271854297927515,
      "grad_norm": 0.6492853164672852,
      "learning_rate": 3.898780108400779e-05,
      "loss": 1.0724,
      "step": 52500
    },
    {
      "epoch": 2.1275906079698546,
      "grad_norm": 0.7578381299972534,
      "learning_rate": 3.895416545229582e-05,
      "loss": 1.079,
      "step": 52510
    },
    {
      "epoch": 2.127995786146958,
      "grad_norm": 0.6205820441246033,
      "learning_rate": 3.8920540826082364e-05,
      "loss": 1.0735,
      "step": 52520
    },
    {
      "epoch": 2.1284009643240616,
      "grad_norm": 0.6654300689697266,
      "learning_rate": 3.8886927211429394e-05,
      "loss": 1.0539,
      "step": 52530
    },
    {
      "epoch": 2.128806142501165,
      "grad_norm": 0.6225121021270752,
      "learning_rate": 3.8853324614396766e-05,
      "loss": 1.0385,
      "step": 52540
    },
    {
      "epoch": 2.129211320678268,
      "grad_norm": 0.6220117211341858,
      "learning_rate": 3.8819733041042515e-05,
      "loss": 1.0627,
      "step": 52550
    },
    {
      "epoch": 2.1296164988553716,
      "grad_norm": 0.6333680748939514,
      "learning_rate": 3.878615249742259e-05,
      "loss": 1.0709,
      "step": 52560
    },
    {
      "epoch": 2.130021677032475,
      "grad_norm": 0.588590681552887,
      "learning_rate": 3.8752582989591014e-05,
      "loss": 1.02,
      "step": 52570
    },
    {
      "epoch": 2.1304268552095786,
      "grad_norm": 0.6073021292686462,
      "learning_rate": 3.871902452359971e-05,
      "loss": 1.066,
      "step": 52580
    },
    {
      "epoch": 2.1308320333866817,
      "grad_norm": 0.6222146153450012,
      "learning_rate": 3.868547710549878e-05,
      "loss": 1.0455,
      "step": 52590
    },
    {
      "epoch": 2.131237211563785,
      "grad_norm": 0.6666841506958008,
      "learning_rate": 3.86519407413361e-05,
      "loss": 1.0347,
      "step": 52600
    },
    {
      "epoch": 2.1316423897408887,
      "grad_norm": 0.6366895437240601,
      "learning_rate": 3.861841543715785e-05,
      "loss": 1.0228,
      "step": 52610
    },
    {
      "epoch": 2.1320475679179918,
      "grad_norm": 0.6125672459602356,
      "learning_rate": 3.858490119900794e-05,
      "loss": 1.0453,
      "step": 52620
    },
    {
      "epoch": 2.1324527460950953,
      "grad_norm": 0.6229450106620789,
      "learning_rate": 3.8551398032928487e-05,
      "loss": 1.0544,
      "step": 52630
    },
    {
      "epoch": 2.132857924272199,
      "grad_norm": 0.5789170861244202,
      "learning_rate": 3.851790594495944e-05,
      "loss": 1.0556,
      "step": 52640
    },
    {
      "epoch": 2.1332631024493023,
      "grad_norm": 0.6568649411201477,
      "learning_rate": 3.848442494113893e-05,
      "loss": 1.0762,
      "step": 52650
    },
    {
      "epoch": 2.1336682806264053,
      "grad_norm": 0.7004095315933228,
      "learning_rate": 3.845095502750288e-05,
      "loss": 1.0115,
      "step": 52660
    },
    {
      "epoch": 2.134073458803509,
      "grad_norm": 0.6841444969177246,
      "learning_rate": 3.841749621008546e-05,
      "loss": 1.0768,
      "step": 52670
    },
    {
      "epoch": 2.1344786369806124,
      "grad_norm": 0.6051193475723267,
      "learning_rate": 3.838404849491862e-05,
      "loss": 1.0331,
      "step": 52680
    },
    {
      "epoch": 2.1348838151577154,
      "grad_norm": 0.5813019275665283,
      "learning_rate": 3.8350611888032474e-05,
      "loss": 1.0353,
      "step": 52690
    },
    {
      "epoch": 2.135288993334819,
      "grad_norm": 0.6359598636627197,
      "learning_rate": 3.831718639545497e-05,
      "loss": 1.0306,
      "step": 52700
    },
    {
      "epoch": 2.1356941715119224,
      "grad_norm": 0.6386069059371948,
      "learning_rate": 3.828377202321223e-05,
      "loss": 1.0504,
      "step": 52710
    },
    {
      "epoch": 2.136099349689026,
      "grad_norm": 0.5536770820617676,
      "learning_rate": 3.825036877732816e-05,
      "loss": 1.0262,
      "step": 52720
    },
    {
      "epoch": 2.136504527866129,
      "grad_norm": 0.6749328374862671,
      "learning_rate": 3.821697666382494e-05,
      "loss": 1.0438,
      "step": 52730
    },
    {
      "epoch": 2.1369097060432325,
      "grad_norm": 0.6476565599441528,
      "learning_rate": 3.8183595688722474e-05,
      "loss": 1.0308,
      "step": 52740
    },
    {
      "epoch": 2.137314884220336,
      "grad_norm": 0.5688904523849487,
      "learning_rate": 3.815022585803883e-05,
      "loss": 1.0424,
      "step": 52750
    },
    {
      "epoch": 2.137720062397439,
      "grad_norm": 0.6055234670639038,
      "learning_rate": 3.8116867177789936e-05,
      "loss": 1.0503,
      "step": 52760
    },
    {
      "epoch": 2.1381252405745426,
      "grad_norm": 0.6415036916732788,
      "learning_rate": 3.808351965398983e-05,
      "loss": 1.0879,
      "step": 52770
    },
    {
      "epoch": 2.138530418751646,
      "grad_norm": 0.6199552416801453,
      "learning_rate": 3.805018329265048e-05,
      "loss": 1.03,
      "step": 52780
    },
    {
      "epoch": 2.1389355969287496,
      "grad_norm": 0.642274796962738,
      "learning_rate": 3.801685809978188e-05,
      "loss": 1.0767,
      "step": 52790
    },
    {
      "epoch": 2.1393407751058526,
      "grad_norm": 0.6259400844573975,
      "learning_rate": 3.798354408139192e-05,
      "loss": 1.0697,
      "step": 52800
    },
    {
      "epoch": 2.139745953282956,
      "grad_norm": 0.6219567060470581,
      "learning_rate": 3.795024124348657e-05,
      "loss": 1.0222,
      "step": 52810
    },
    {
      "epoch": 2.1401511314600596,
      "grad_norm": 0.6713687181472778,
      "learning_rate": 3.7916949592069786e-05,
      "loss": 1.063,
      "step": 52820
    },
    {
      "epoch": 2.140556309637163,
      "grad_norm": 0.7016239166259766,
      "learning_rate": 3.788366913314339e-05,
      "loss": 1.0278,
      "step": 52830
    },
    {
      "epoch": 2.140961487814266,
      "grad_norm": 0.7128947377204895,
      "learning_rate": 3.785039987270731e-05,
      "loss": 1.0681,
      "step": 52840
    },
    {
      "epoch": 2.1413666659913697,
      "grad_norm": 0.5856156349182129,
      "learning_rate": 3.781714181675941e-05,
      "loss": 1.0638,
      "step": 52850
    },
    {
      "epoch": 2.141771844168473,
      "grad_norm": 0.6049305200576782,
      "learning_rate": 3.778389497129559e-05,
      "loss": 1.0503,
      "step": 52860
    },
    {
      "epoch": 2.1421770223455763,
      "grad_norm": 0.622010350227356,
      "learning_rate": 3.775065934230958e-05,
      "loss": 1.0468,
      "step": 52870
    },
    {
      "epoch": 2.14258220052268,
      "grad_norm": 0.6130561232566833,
      "learning_rate": 3.771743493579323e-05,
      "loss": 1.0266,
      "step": 52880
    },
    {
      "epoch": 2.1429873786997833,
      "grad_norm": 0.6064489483833313,
      "learning_rate": 3.7684221757736327e-05,
      "loss": 1.0415,
      "step": 52890
    },
    {
      "epoch": 2.143392556876887,
      "grad_norm": 0.671116292476654,
      "learning_rate": 3.7651019814126654e-05,
      "loss": 1.0608,
      "step": 52900
    },
    {
      "epoch": 2.14379773505399,
      "grad_norm": 0.56427401304245,
      "learning_rate": 3.7617829110949875e-05,
      "loss": 1.0158,
      "step": 52910
    },
    {
      "epoch": 2.1442029132310934,
      "grad_norm": 0.6576723456382751,
      "learning_rate": 3.758464965418976e-05,
      "loss": 1.0379,
      "step": 52920
    },
    {
      "epoch": 2.144608091408197,
      "grad_norm": 0.639812707901001,
      "learning_rate": 3.755148144982793e-05,
      "loss": 1.0581,
      "step": 52930
    },
    {
      "epoch": 2.1450132695853004,
      "grad_norm": 0.6268312335014343,
      "learning_rate": 3.751832450384405e-05,
      "loss": 1.0304,
      "step": 52940
    },
    {
      "epoch": 2.1454184477624034,
      "grad_norm": 0.7095539569854736,
      "learning_rate": 3.748517882221575e-05,
      "loss": 1.0499,
      "step": 52950
    },
    {
      "epoch": 2.145823625939507,
      "grad_norm": 0.5658072829246521,
      "learning_rate": 3.7452044410918654e-05,
      "loss": 1.0776,
      "step": 52960
    },
    {
      "epoch": 2.1462288041166104,
      "grad_norm": 0.6322121620178223,
      "learning_rate": 3.741892127592625e-05,
      "loss": 1.039,
      "step": 52970
    },
    {
      "epoch": 2.1466339822937135,
      "grad_norm": 0.5905674695968628,
      "learning_rate": 3.738580942321011e-05,
      "loss": 1.0507,
      "step": 52980
    },
    {
      "epoch": 2.147039160470817,
      "grad_norm": 0.5659306645393372,
      "learning_rate": 3.735270885873967e-05,
      "loss": 1.0774,
      "step": 52990
    },
    {
      "epoch": 2.1474443386479205,
      "grad_norm": 0.6821537017822266,
      "learning_rate": 3.731961958848242e-05,
      "loss": 1.071,
      "step": 53000
    },
    {
      "epoch": 2.147849516825024,
      "grad_norm": 0.6402195692062378,
      "learning_rate": 3.728654161840377e-05,
      "loss": 1.0266,
      "step": 53010
    },
    {
      "epoch": 2.148254695002127,
      "grad_norm": 0.5945355296134949,
      "learning_rate": 3.725347495446715e-05,
      "loss": 1.0537,
      "step": 53020
    },
    {
      "epoch": 2.1486598731792306,
      "grad_norm": 0.6305971145629883,
      "learning_rate": 3.7220419602633814e-05,
      "loss": 1.0269,
      "step": 53030
    },
    {
      "epoch": 2.149065051356334,
      "grad_norm": 0.6706334352493286,
      "learning_rate": 3.718737556886316e-05,
      "loss": 1.0273,
      "step": 53040
    },
    {
      "epoch": 2.149470229533437,
      "grad_norm": 0.6000314354896545,
      "learning_rate": 3.7154342859112314e-05,
      "loss": 1.0628,
      "step": 53050
    },
    {
      "epoch": 2.1498754077105406,
      "grad_norm": 0.6407192945480347,
      "learning_rate": 3.712132147933666e-05,
      "loss": 1.0364,
      "step": 53060
    },
    {
      "epoch": 2.150280585887644,
      "grad_norm": 0.6701192855834961,
      "learning_rate": 3.708831143548926e-05,
      "loss": 1.0535,
      "step": 53070
    },
    {
      "epoch": 2.1506857640647477,
      "grad_norm": 0.6099025011062622,
      "learning_rate": 3.705531273352133e-05,
      "loss": 1.0879,
      "step": 53080
    },
    {
      "epoch": 2.1510909422418507,
      "grad_norm": 0.6236820816993713,
      "learning_rate": 3.7022325379381884e-05,
      "loss": 1.0624,
      "step": 53090
    },
    {
      "epoch": 2.151496120418954,
      "grad_norm": 0.6523250937461853,
      "learning_rate": 3.698934937901799e-05,
      "loss": 1.0648,
      "step": 53100
    },
    {
      "epoch": 2.1519012985960577,
      "grad_norm": 0.6398141384124756,
      "learning_rate": 3.695638473837466e-05,
      "loss": 1.0499,
      "step": 53110
    },
    {
      "epoch": 2.152306476773161,
      "grad_norm": 0.6069547533988953,
      "learning_rate": 3.692343146339488e-05,
      "loss": 1.035,
      "step": 53120
    },
    {
      "epoch": 2.1527116549502643,
      "grad_norm": 0.6904507279396057,
      "learning_rate": 3.689048956001947e-05,
      "loss": 1.0355,
      "step": 53130
    },
    {
      "epoch": 2.153116833127368,
      "grad_norm": 0.6383225321769714,
      "learning_rate": 3.685755903418732e-05,
      "loss": 1.0591,
      "step": 53140
    },
    {
      "epoch": 2.1535220113044713,
      "grad_norm": 0.7281789779663086,
      "learning_rate": 3.682463989183526e-05,
      "loss": 1.0395,
      "step": 53150
    },
    {
      "epoch": 2.1539271894815744,
      "grad_norm": 0.6118913888931274,
      "learning_rate": 3.679173213889797e-05,
      "loss": 1.0767,
      "step": 53160
    },
    {
      "epoch": 2.154332367658678,
      "grad_norm": 0.6669185161590576,
      "learning_rate": 3.6758835781308185e-05,
      "loss": 1.0446,
      "step": 53170
    },
    {
      "epoch": 2.1547375458357814,
      "grad_norm": 0.6271657943725586,
      "learning_rate": 3.6725950824996535e-05,
      "loss": 1.0716,
      "step": 53180
    },
    {
      "epoch": 2.155142724012885,
      "grad_norm": 0.6282188296318054,
      "learning_rate": 3.669307727589164e-05,
      "loss": 1.0604,
      "step": 53190
    },
    {
      "epoch": 2.155547902189988,
      "grad_norm": 0.6116662621498108,
      "learning_rate": 3.666021513991996e-05,
      "loss": 1.0789,
      "step": 53200
    },
    {
      "epoch": 2.1559530803670914,
      "grad_norm": 0.6409932971000671,
      "learning_rate": 3.662736442300604e-05,
      "loss": 1.0228,
      "step": 53210
    },
    {
      "epoch": 2.156358258544195,
      "grad_norm": 0.6691927313804626,
      "learning_rate": 3.6594525131072174e-05,
      "loss": 1.0439,
      "step": 53220
    },
    {
      "epoch": 2.156763436721298,
      "grad_norm": 0.6412213444709778,
      "learning_rate": 3.656169727003885e-05,
      "loss": 1.0462,
      "step": 53230
    },
    {
      "epoch": 2.1571686148984015,
      "grad_norm": 0.6344202160835266,
      "learning_rate": 3.652888084582428e-05,
      "loss": 1.0388,
      "step": 53240
    },
    {
      "epoch": 2.157573793075505,
      "grad_norm": 0.6256418824195862,
      "learning_rate": 3.649607586434474e-05,
      "loss": 1.0485,
      "step": 53250
    },
    {
      "epoch": 2.1579789712526085,
      "grad_norm": 0.6584806442260742,
      "learning_rate": 3.6463282331514314e-05,
      "loss": 1.0636,
      "step": 53260
    },
    {
      "epoch": 2.1583841494297116,
      "grad_norm": 0.6093568801879883,
      "learning_rate": 3.64305002532452e-05,
      "loss": 1.0791,
      "step": 53270
    },
    {
      "epoch": 2.158789327606815,
      "grad_norm": 0.669457197189331,
      "learning_rate": 3.639772963544732e-05,
      "loss": 1.0566,
      "step": 53280
    },
    {
      "epoch": 2.1591945057839186,
      "grad_norm": 0.692116379737854,
      "learning_rate": 3.6364970484028785e-05,
      "loss": 1.0297,
      "step": 53290
    },
    {
      "epoch": 2.159599683961022,
      "grad_norm": 0.6617388129234314,
      "learning_rate": 3.633222280489539e-05,
      "loss": 1.0807,
      "step": 53300
    },
    {
      "epoch": 2.160004862138125,
      "grad_norm": 0.6094083786010742,
      "learning_rate": 3.629948660395104e-05,
      "loss": 1.0429,
      "step": 53310
    },
    {
      "epoch": 2.1604100403152287,
      "grad_norm": 0.6369814276695251,
      "learning_rate": 3.626676188709743e-05,
      "loss": 1.0585,
      "step": 53320
    },
    {
      "epoch": 2.160815218492332,
      "grad_norm": 0.6392606496810913,
      "learning_rate": 3.623404866023434e-05,
      "loss": 1.0538,
      "step": 53330
    },
    {
      "epoch": 2.161220396669435,
      "grad_norm": 0.5980780124664307,
      "learning_rate": 3.6201346929259275e-05,
      "loss": 1.0517,
      "step": 53340
    },
    {
      "epoch": 2.1616255748465387,
      "grad_norm": 0.7066354155540466,
      "learning_rate": 3.616865670006792e-05,
      "loss": 1.0494,
      "step": 53350
    },
    {
      "epoch": 2.1620307530236422,
      "grad_norm": 0.5848034620285034,
      "learning_rate": 3.613597797855366e-05,
      "loss": 1.0498,
      "step": 53360
    },
    {
      "epoch": 2.1624359312007457,
      "grad_norm": 0.6847724318504333,
      "learning_rate": 3.6103310770607954e-05,
      "loss": 1.0754,
      "step": 53370
    },
    {
      "epoch": 2.162841109377849,
      "grad_norm": 0.6150724291801453,
      "learning_rate": 3.607065508212008e-05,
      "loss": 1.0216,
      "step": 53380
    },
    {
      "epoch": 2.1632462875549523,
      "grad_norm": 0.638733983039856,
      "learning_rate": 3.603801091897731e-05,
      "loss": 1.0387,
      "step": 53390
    },
    {
      "epoch": 2.163651465732056,
      "grad_norm": 0.6562722325325012,
      "learning_rate": 3.600537828706482e-05,
      "loss": 1.0401,
      "step": 53400
    },
    {
      "epoch": 2.164056643909159,
      "grad_norm": 0.6268374919891357,
      "learning_rate": 3.5972757192265737e-05,
      "loss": 1.0533,
      "step": 53410
    },
    {
      "epoch": 2.1644618220862624,
      "grad_norm": 0.5439851880073547,
      "learning_rate": 3.5940147640461005e-05,
      "loss": 1.0023,
      "step": 53420
    },
    {
      "epoch": 2.164867000263366,
      "grad_norm": 0.7037273645401001,
      "learning_rate": 3.5907549637529636e-05,
      "loss": 1.035,
      "step": 53430
    },
    {
      "epoch": 2.1652721784404694,
      "grad_norm": 0.5923729538917542,
      "learning_rate": 3.587496318934841e-05,
      "loss": 1.0181,
      "step": 53440
    },
    {
      "epoch": 2.1656773566175724,
      "grad_norm": 0.6171359419822693,
      "learning_rate": 3.584238830179211e-05,
      "loss": 1.0454,
      "step": 53450
    },
    {
      "epoch": 2.166082534794676,
      "grad_norm": 0.7638300657272339,
      "learning_rate": 3.580982498073344e-05,
      "loss": 1.0605,
      "step": 53460
    },
    {
      "epoch": 2.1664877129717794,
      "grad_norm": 0.6204900741577148,
      "learning_rate": 3.577727323204303e-05,
      "loss": 1.0723,
      "step": 53470
    },
    {
      "epoch": 2.1668928911488825,
      "grad_norm": 0.6919833421707153,
      "learning_rate": 3.574473306158931e-05,
      "loss": 1.0574,
      "step": 53480
    },
    {
      "epoch": 2.167298069325986,
      "grad_norm": 0.7198923826217651,
      "learning_rate": 3.5712204475238756e-05,
      "loss": 1.0373,
      "step": 53490
    },
    {
      "epoch": 2.1677032475030895,
      "grad_norm": 0.6871110796928406,
      "learning_rate": 3.567968747885569e-05,
      "loss": 1.0469,
      "step": 53500
    },
    {
      "epoch": 2.168108425680193,
      "grad_norm": 0.6396542191505432,
      "learning_rate": 3.564718207830241e-05,
      "loss": 1.0495,
      "step": 53510
    },
    {
      "epoch": 2.168513603857296,
      "grad_norm": 0.6207573413848877,
      "learning_rate": 3.561468827943899e-05,
      "loss": 1.0614,
      "step": 53520
    },
    {
      "epoch": 2.1689187820343996,
      "grad_norm": 0.6900368928909302,
      "learning_rate": 3.5582206088123535e-05,
      "loss": 1.064,
      "step": 53530
    },
    {
      "epoch": 2.169323960211503,
      "grad_norm": 0.6439892053604126,
      "learning_rate": 3.554973551021204e-05,
      "loss": 1.0357,
      "step": 53540
    },
    {
      "epoch": 2.1697291383886066,
      "grad_norm": 0.6591747999191284,
      "learning_rate": 3.5517276551558345e-05,
      "loss": 1.0518,
      "step": 53550
    },
    {
      "epoch": 2.1701343165657097,
      "grad_norm": 0.6374854445457458,
      "learning_rate": 3.548482921801423e-05,
      "loss": 1.0814,
      "step": 53560
    },
    {
      "epoch": 2.170539494742813,
      "grad_norm": 0.683010458946228,
      "learning_rate": 3.545239351542941e-05,
      "loss": 1.0443,
      "step": 53570
    },
    {
      "epoch": 2.1709446729199167,
      "grad_norm": 0.7117885947227478,
      "learning_rate": 3.541996944965149e-05,
      "loss": 1.0305,
      "step": 53580
    },
    {
      "epoch": 2.1713498510970197,
      "grad_norm": 0.6509752869606018,
      "learning_rate": 3.5387557026525916e-05,
      "loss": 1.0388,
      "step": 53590
    },
    {
      "epoch": 2.1717550292741232,
      "grad_norm": 0.657082200050354,
      "learning_rate": 3.5355156251896136e-05,
      "loss": 1.0412,
      "step": 53600
    },
    {
      "epoch": 2.1721602074512267,
      "grad_norm": 0.6366336345672607,
      "learning_rate": 3.532276713160338e-05,
      "loss": 1.0221,
      "step": 53610
    },
    {
      "epoch": 2.1725653856283302,
      "grad_norm": 0.5779707431793213,
      "learning_rate": 3.529038967148689e-05,
      "loss": 1.0291,
      "step": 53620
    },
    {
      "epoch": 2.1729705638054333,
      "grad_norm": 0.5478004813194275,
      "learning_rate": 3.5258023877383727e-05,
      "loss": 1.048,
      "step": 53630
    },
    {
      "epoch": 2.173375741982537,
      "grad_norm": 0.6652435660362244,
      "learning_rate": 3.522566975512893e-05,
      "loss": 1.0481,
      "step": 53640
    },
    {
      "epoch": 2.1737809201596403,
      "grad_norm": 0.5763639211654663,
      "learning_rate": 3.5193327310555314e-05,
      "loss": 1.0451,
      "step": 53650
    },
    {
      "epoch": 2.174186098336744,
      "grad_norm": 0.7552454471588135,
      "learning_rate": 3.516099654949372e-05,
      "loss": 1.034,
      "step": 53660
    },
    {
      "epoch": 2.174591276513847,
      "grad_norm": 0.7000679969787598,
      "learning_rate": 3.5128677477772734e-05,
      "loss": 1.0714,
      "step": 53670
    },
    {
      "epoch": 2.1749964546909504,
      "grad_norm": 0.7062049508094788,
      "learning_rate": 3.509637010121903e-05,
      "loss": 1.0509,
      "step": 53680
    },
    {
      "epoch": 2.175401632868054,
      "grad_norm": 0.6311936378479004,
      "learning_rate": 3.506407442565698e-05,
      "loss": 1.0434,
      "step": 53690
    },
    {
      "epoch": 2.175806811045157,
      "grad_norm": 0.6211927533149719,
      "learning_rate": 3.5031790456909e-05,
      "loss": 1.0445,
      "step": 53700
    },
    {
      "epoch": 2.1762119892222604,
      "grad_norm": 0.60226970911026,
      "learning_rate": 3.499951820079526e-05,
      "loss": 1.0433,
      "step": 53710
    },
    {
      "epoch": 2.176617167399364,
      "grad_norm": 0.6670381426811218,
      "learning_rate": 3.496725766313393e-05,
      "loss": 1.0729,
      "step": 53720
    },
    {
      "epoch": 2.1770223455764675,
      "grad_norm": 0.6315358281135559,
      "learning_rate": 3.4935008849740944e-05,
      "loss": 1.0437,
      "step": 53730
    },
    {
      "epoch": 2.1774275237535705,
      "grad_norm": 0.6171205043792725,
      "learning_rate": 3.490277176643033e-05,
      "loss": 1.0421,
      "step": 53740
    },
    {
      "epoch": 2.177832701930674,
      "grad_norm": 0.616405725479126,
      "learning_rate": 3.487054641901376e-05,
      "loss": 1.0295,
      "step": 53750
    },
    {
      "epoch": 2.1782378801077775,
      "grad_norm": 0.6513734459877014,
      "learning_rate": 3.483833281330098e-05,
      "loss": 1.0366,
      "step": 53760
    },
    {
      "epoch": 2.1786430582848806,
      "grad_norm": 0.6495358347892761,
      "learning_rate": 3.480613095509947e-05,
      "loss": 1.0565,
      "step": 53770
    },
    {
      "epoch": 2.179048236461984,
      "grad_norm": 0.6249848008155823,
      "learning_rate": 3.4773940850214707e-05,
      "loss": 1.016,
      "step": 53780
    },
    {
      "epoch": 2.1794534146390876,
      "grad_norm": 0.6799317598342896,
      "learning_rate": 3.474176250444998e-05,
      "loss": 1.0764,
      "step": 53790
    },
    {
      "epoch": 2.179858592816191,
      "grad_norm": 0.6123341917991638,
      "learning_rate": 3.470959592360654e-05,
      "loss": 1.0311,
      "step": 53800
    },
    {
      "epoch": 2.180263770993294,
      "grad_norm": 0.5917949080467224,
      "learning_rate": 3.467744111348338e-05,
      "loss": 1.0626,
      "step": 53810
    },
    {
      "epoch": 2.1806689491703977,
      "grad_norm": 0.6014786958694458,
      "learning_rate": 3.46452980798775e-05,
      "loss": 1.0538,
      "step": 53820
    },
    {
      "epoch": 2.181074127347501,
      "grad_norm": 0.6259431838989258,
      "learning_rate": 3.461316682858373e-05,
      "loss": 1.057,
      "step": 53830
    },
    {
      "epoch": 2.1814793055246042,
      "grad_norm": 0.6298443675041199,
      "learning_rate": 3.458104736539475e-05,
      "loss": 1.0514,
      "step": 53840
    },
    {
      "epoch": 2.1818844837017077,
      "grad_norm": 0.6561385989189148,
      "learning_rate": 3.454893969610113e-05,
      "loss": 1.0187,
      "step": 53850
    },
    {
      "epoch": 2.1822896618788112,
      "grad_norm": 0.7607010006904602,
      "learning_rate": 3.451684382649136e-05,
      "loss": 1.0265,
      "step": 53860
    },
    {
      "epoch": 2.1826948400559147,
      "grad_norm": 0.6423912644386292,
      "learning_rate": 3.448475976235177e-05,
      "loss": 1.0344,
      "step": 53870
    },
    {
      "epoch": 2.183100018233018,
      "grad_norm": 0.6068238615989685,
      "learning_rate": 3.445268750946651e-05,
      "loss": 1.0279,
      "step": 53880
    },
    {
      "epoch": 2.1835051964101213,
      "grad_norm": 0.6765007376670837,
      "learning_rate": 3.4420627073617694e-05,
      "loss": 1.0901,
      "step": 53890
    },
    {
      "epoch": 2.183910374587225,
      "grad_norm": 0.6313706636428833,
      "learning_rate": 3.4388578460585174e-05,
      "loss": 1.0391,
      "step": 53900
    },
    {
      "epoch": 2.1843155527643283,
      "grad_norm": 0.6333765983581543,
      "learning_rate": 3.435654167614689e-05,
      "loss": 1.0507,
      "step": 53910
    },
    {
      "epoch": 2.1847207309414314,
      "grad_norm": 0.6004877090454102,
      "learning_rate": 3.43245167260784e-05,
      "loss": 1.0365,
      "step": 53920
    },
    {
      "epoch": 2.185125909118535,
      "grad_norm": 0.670911967754364,
      "learning_rate": 3.429250361615333e-05,
      "loss": 1.0433,
      "step": 53930
    },
    {
      "epoch": 2.1855310872956384,
      "grad_norm": 0.6291624903678894,
      "learning_rate": 3.4260502352143e-05,
      "loss": 1.0468,
      "step": 53940
    },
    {
      "epoch": 2.1859362654727414,
      "grad_norm": 0.6051002144813538,
      "learning_rate": 3.422851293981676e-05,
      "loss": 1.0637,
      "step": 53950
    },
    {
      "epoch": 2.186341443649845,
      "grad_norm": 0.6324188709259033,
      "learning_rate": 3.419653538494163e-05,
      "loss": 1.0463,
      "step": 53960
    },
    {
      "epoch": 2.1867466218269485,
      "grad_norm": 0.6150201559066772,
      "learning_rate": 3.4164569693282756e-05,
      "loss": 1.053,
      "step": 53970
    },
    {
      "epoch": 2.187151800004052,
      "grad_norm": 0.6739506125450134,
      "learning_rate": 3.413261587060288e-05,
      "loss": 1.0364,
      "step": 53980
    },
    {
      "epoch": 2.187556978181155,
      "grad_norm": 0.6540142893791199,
      "learning_rate": 3.410067392266279e-05,
      "loss": 1.0702,
      "step": 53990
    },
    {
      "epoch": 2.1879621563582585,
      "grad_norm": 0.6348776817321777,
      "learning_rate": 3.406874385522098e-05,
      "loss": 1.0261,
      "step": 54000
    },
    {
      "epoch": 2.188367334535362,
      "grad_norm": 0.6517127156257629,
      "learning_rate": 3.403682567403394e-05,
      "loss": 1.0455,
      "step": 54010
    },
    {
      "epoch": 2.1887725127124655,
      "grad_norm": 0.7677069902420044,
      "learning_rate": 3.400491938485596e-05,
      "loss": 1.0543,
      "step": 54020
    },
    {
      "epoch": 2.1891776908895686,
      "grad_norm": 0.6457732915878296,
      "learning_rate": 3.3973024993439205e-05,
      "loss": 1.0563,
      "step": 54030
    },
    {
      "epoch": 2.189582869066672,
      "grad_norm": 0.6876969337463379,
      "learning_rate": 3.394114250553363e-05,
      "loss": 1.0405,
      "step": 54040
    },
    {
      "epoch": 2.1899880472437756,
      "grad_norm": 0.7000178694725037,
      "learning_rate": 3.390927192688713e-05,
      "loss": 1.0577,
      "step": 54050
    },
    {
      "epoch": 2.1903932254208787,
      "grad_norm": 0.5574647188186646,
      "learning_rate": 3.387741326324537e-05,
      "loss": 1.0673,
      "step": 54060
    },
    {
      "epoch": 2.190798403597982,
      "grad_norm": 0.687508761882782,
      "learning_rate": 3.384556652035194e-05,
      "loss": 1.0427,
      "step": 54070
    },
    {
      "epoch": 2.1912035817750857,
      "grad_norm": 0.6660807132720947,
      "learning_rate": 3.381373170394825e-05,
      "loss": 1.0207,
      "step": 54080
    },
    {
      "epoch": 2.191608759952189,
      "grad_norm": 0.7511933445930481,
      "learning_rate": 3.378190881977359e-05,
      "loss": 1.0675,
      "step": 54090
    },
    {
      "epoch": 2.1920139381292922,
      "grad_norm": 0.5927016139030457,
      "learning_rate": 3.3750097873565e-05,
      "loss": 1.0465,
      "step": 54100
    },
    {
      "epoch": 2.1924191163063957,
      "grad_norm": 0.6044542193412781,
      "learning_rate": 3.37182988710575e-05,
      "loss": 1.0347,
      "step": 54110
    },
    {
      "epoch": 2.1928242944834992,
      "grad_norm": 0.6381480097770691,
      "learning_rate": 3.368651181798386e-05,
      "loss": 1.0683,
      "step": 54120
    },
    {
      "epoch": 2.1932294726606023,
      "grad_norm": 0.6574164032936096,
      "learning_rate": 3.3654736720074784e-05,
      "loss": 1.045,
      "step": 54130
    },
    {
      "epoch": 2.193634650837706,
      "grad_norm": 0.6094204783439636,
      "learning_rate": 3.3622973583058706e-05,
      "loss": 1.0184,
      "step": 54140
    },
    {
      "epoch": 2.1940398290148093,
      "grad_norm": 0.6397483944892883,
      "learning_rate": 3.359122241266197e-05,
      "loss": 1.0511,
      "step": 54150
    },
    {
      "epoch": 2.194445007191913,
      "grad_norm": 0.6250160932540894,
      "learning_rate": 3.3559483214608824e-05,
      "loss": 1.0122,
      "step": 54160
    },
    {
      "epoch": 2.194850185369016,
      "grad_norm": 0.7058985233306885,
      "learning_rate": 3.35277559946212e-05,
      "loss": 1.0512,
      "step": 54170
    },
    {
      "epoch": 2.1952553635461194,
      "grad_norm": 0.6137279868125916,
      "learning_rate": 3.3496040758419025e-05,
      "loss": 1.0219,
      "step": 54180
    },
    {
      "epoch": 2.195660541723223,
      "grad_norm": 0.7247489094734192,
      "learning_rate": 3.3464337511719965e-05,
      "loss": 1.0699,
      "step": 54190
    },
    {
      "epoch": 2.196065719900326,
      "grad_norm": 0.6042141914367676,
      "learning_rate": 3.343264626023963e-05,
      "loss": 1.041,
      "step": 54200
    },
    {
      "epoch": 2.1964708980774295,
      "grad_norm": 0.6126183271408081,
      "learning_rate": 3.340096700969131e-05,
      "loss": 1.061,
      "step": 54210
    },
    {
      "epoch": 2.196876076254533,
      "grad_norm": 0.6463432312011719,
      "learning_rate": 3.336929976578629e-05,
      "loss": 1.051,
      "step": 54220
    },
    {
      "epoch": 2.1972812544316365,
      "grad_norm": 0.6596893668174744,
      "learning_rate": 3.333764453423357e-05,
      "loss": 1.0158,
      "step": 54230
    },
    {
      "epoch": 2.1976864326087395,
      "grad_norm": 0.7402449250221252,
      "learning_rate": 3.330600132074004e-05,
      "loss": 1.06,
      "step": 54240
    },
    {
      "epoch": 2.198091610785843,
      "grad_norm": 0.7543529272079468,
      "learning_rate": 3.3274370131010456e-05,
      "loss": 1.032,
      "step": 54250
    },
    {
      "epoch": 2.1984967889629465,
      "grad_norm": 0.6659629344940186,
      "learning_rate": 3.3242750970747384e-05,
      "loss": 1.0347,
      "step": 54260
    },
    {
      "epoch": 2.19890196714005,
      "grad_norm": 0.6554586291313171,
      "learning_rate": 3.321114384565114e-05,
      "loss": 1.0551,
      "step": 54270
    },
    {
      "epoch": 2.199307145317153,
      "grad_norm": 0.6498068571090698,
      "learning_rate": 3.317954876142e-05,
      "loss": 1.0529,
      "step": 54280
    },
    {
      "epoch": 2.1997123234942566,
      "grad_norm": 0.6864092350006104,
      "learning_rate": 3.314796572374991e-05,
      "loss": 1.0414,
      "step": 54290
    },
    {
      "epoch": 2.20011750167136,
      "grad_norm": 0.6805278062820435,
      "learning_rate": 3.3116394738334866e-05,
      "loss": 1.0643,
      "step": 54300
    },
    {
      "epoch": 2.200522679848463,
      "grad_norm": 0.6776809692382812,
      "learning_rate": 3.3084835810866485e-05,
      "loss": 1.0442,
      "step": 54310
    },
    {
      "epoch": 2.2009278580255667,
      "grad_norm": 0.6038316488265991,
      "learning_rate": 3.3053288947034346e-05,
      "loss": 1.0495,
      "step": 54320
    },
    {
      "epoch": 2.20133303620267,
      "grad_norm": 0.6668871641159058,
      "learning_rate": 3.302175415252573e-05,
      "loss": 1.0218,
      "step": 54330
    },
    {
      "epoch": 2.2017382143797737,
      "grad_norm": 0.6040464639663696,
      "learning_rate": 3.2990231433025875e-05,
      "loss": 1.0759,
      "step": 54340
    },
    {
      "epoch": 2.2021433925568767,
      "grad_norm": 0.6131914258003235,
      "learning_rate": 3.295872079421769e-05,
      "loss": 1.0397,
      "step": 54350
    },
    {
      "epoch": 2.2025485707339802,
      "grad_norm": 0.7004358172416687,
      "learning_rate": 3.2927222241782105e-05,
      "loss": 1.0554,
      "step": 54360
    },
    {
      "epoch": 2.2029537489110838,
      "grad_norm": 0.6163952946662903,
      "learning_rate": 3.2895735781397685e-05,
      "loss": 1.0166,
      "step": 54370
    },
    {
      "epoch": 2.2033589270881873,
      "grad_norm": 0.6435933709144592,
      "learning_rate": 3.2864261418740925e-05,
      "loss": 1.061,
      "step": 54380
    },
    {
      "epoch": 2.2037641052652903,
      "grad_norm": 0.6508039236068726,
      "learning_rate": 3.283279915948607e-05,
      "loss": 1.059,
      "step": 54390
    },
    {
      "epoch": 2.204169283442394,
      "grad_norm": 0.6059533357620239,
      "learning_rate": 3.280134900930523e-05,
      "loss": 1.0464,
      "step": 54400
    },
    {
      "epoch": 2.2045744616194973,
      "grad_norm": 0.5878006219863892,
      "learning_rate": 3.276991097386831e-05,
      "loss": 1.0122,
      "step": 54410
    },
    {
      "epoch": 2.2049796397966004,
      "grad_norm": 0.5942531824111938,
      "learning_rate": 3.273848505884309e-05,
      "loss": 1.0682,
      "step": 54420
    },
    {
      "epoch": 2.205384817973704,
      "grad_norm": 0.7153899669647217,
      "learning_rate": 3.270707126989505e-05,
      "loss": 1.034,
      "step": 54430
    },
    {
      "epoch": 2.2057899961508074,
      "grad_norm": 0.6617687344551086,
      "learning_rate": 3.2675669612687565e-05,
      "loss": 1.0505,
      "step": 54440
    },
    {
      "epoch": 2.206195174327911,
      "grad_norm": 0.656867265701294,
      "learning_rate": 3.264428009288185e-05,
      "loss": 1.049,
      "step": 54450
    },
    {
      "epoch": 2.206600352505014,
      "grad_norm": 0.6705126166343689,
      "learning_rate": 3.261290271613682e-05,
      "loss": 1.0485,
      "step": 54460
    },
    {
      "epoch": 2.2070055306821175,
      "grad_norm": 0.629774808883667,
      "learning_rate": 3.258153748810929e-05,
      "loss": 1.0801,
      "step": 54470
    },
    {
      "epoch": 2.207410708859221,
      "grad_norm": 0.6279151439666748,
      "learning_rate": 3.255018441445389e-05,
      "loss": 1.0401,
      "step": 54480
    },
    {
      "epoch": 2.207815887036324,
      "grad_norm": 0.6297650933265686,
      "learning_rate": 3.2518843500823025e-05,
      "loss": 1.034,
      "step": 54490
    },
    {
      "epoch": 2.2082210652134275,
      "grad_norm": 0.7182501554489136,
      "learning_rate": 3.248751475286689e-05,
      "loss": 1.0395,
      "step": 54500
    },
    {
      "epoch": 2.208626243390531,
      "grad_norm": 0.6281754970550537,
      "learning_rate": 3.2456198176233543e-05,
      "loss": 1.0747,
      "step": 54510
    },
    {
      "epoch": 2.2090314215676345,
      "grad_norm": 0.6347669959068298,
      "learning_rate": 3.2424893776568774e-05,
      "loss": 1.0461,
      "step": 54520
    },
    {
      "epoch": 2.2094365997447376,
      "grad_norm": 0.6008911728858948,
      "learning_rate": 3.239360155951624e-05,
      "loss": 1.0037,
      "step": 54530
    },
    {
      "epoch": 2.209841777921841,
      "grad_norm": 0.7399722337722778,
      "learning_rate": 3.23623215307174e-05,
      "loss": 1.0737,
      "step": 54540
    },
    {
      "epoch": 2.2102469560989446,
      "grad_norm": 0.6889854669570923,
      "learning_rate": 3.2331053695811495e-05,
      "loss": 1.0751,
      "step": 54550
    },
    {
      "epoch": 2.2106521342760477,
      "grad_norm": 0.6463508605957031,
      "learning_rate": 3.229979806043555e-05,
      "loss": 1.0633,
      "step": 54560
    },
    {
      "epoch": 2.211057312453151,
      "grad_norm": 0.6828311085700989,
      "learning_rate": 3.2268554630224435e-05,
      "loss": 1.0411,
      "step": 54570
    },
    {
      "epoch": 2.2114624906302547,
      "grad_norm": 0.6504213213920593,
      "learning_rate": 3.2237323410810715e-05,
      "loss": 1.0255,
      "step": 54580
    },
    {
      "epoch": 2.211867668807358,
      "grad_norm": 0.6396294236183167,
      "learning_rate": 3.2206104407824965e-05,
      "loss": 1.0605,
      "step": 54590
    },
    {
      "epoch": 2.2122728469844612,
      "grad_norm": 0.6608656644821167,
      "learning_rate": 3.2174897626895326e-05,
      "loss": 1.0411,
      "step": 54600
    },
    {
      "epoch": 2.2126780251615648,
      "grad_norm": 0.6414101719856262,
      "learning_rate": 3.21437030736479e-05,
      "loss": 1.0725,
      "step": 54610
    },
    {
      "epoch": 2.2130832033386683,
      "grad_norm": 0.6555875539779663,
      "learning_rate": 3.211252075370647e-05,
      "loss": 1.046,
      "step": 54620
    },
    {
      "epoch": 2.2134883815157718,
      "grad_norm": 0.7074677348136902,
      "learning_rate": 3.2081350672692666e-05,
      "loss": 1.0544,
      "step": 54630
    },
    {
      "epoch": 2.213893559692875,
      "grad_norm": 0.6461824774742126,
      "learning_rate": 3.205019283622594e-05,
      "loss": 1.0192,
      "step": 54640
    },
    {
      "epoch": 2.2142987378699783,
      "grad_norm": 0.6437401175498962,
      "learning_rate": 3.201904724992352e-05,
      "loss": 1.1026,
      "step": 54650
    },
    {
      "epoch": 2.214703916047082,
      "grad_norm": 0.6563968062400818,
      "learning_rate": 3.198791391940036e-05,
      "loss": 1.0604,
      "step": 54660
    },
    {
      "epoch": 2.215109094224185,
      "grad_norm": 0.6316072940826416,
      "learning_rate": 3.19567928502693e-05,
      "loss": 1.0333,
      "step": 54670
    },
    {
      "epoch": 2.2155142724012884,
      "grad_norm": 0.5997112393379211,
      "learning_rate": 3.192568404814088e-05,
      "loss": 1.0676,
      "step": 54680
    },
    {
      "epoch": 2.215919450578392,
      "grad_norm": 0.6385384798049927,
      "learning_rate": 3.189458751862351e-05,
      "loss": 1.031,
      "step": 54690
    },
    {
      "epoch": 2.2163246287554954,
      "grad_norm": 0.5894071459770203,
      "learning_rate": 3.186350326732333e-05,
      "loss": 1.0417,
      "step": 54700
    },
    {
      "epoch": 2.2167298069325985,
      "grad_norm": 0.6690185070037842,
      "learning_rate": 3.183243129984433e-05,
      "loss": 1.0627,
      "step": 54710
    },
    {
      "epoch": 2.217134985109702,
      "grad_norm": 0.66822749376297,
      "learning_rate": 3.18013716217882e-05,
      "loss": 1.0307,
      "step": 54720
    },
    {
      "epoch": 2.2175401632868055,
      "grad_norm": 0.6551576256752014,
      "learning_rate": 3.177032423875449e-05,
      "loss": 1.0639,
      "step": 54730
    },
    {
      "epoch": 2.217945341463909,
      "grad_norm": 0.6492186188697815,
      "learning_rate": 3.173928915634042e-05,
      "loss": 1.0313,
      "step": 54740
    },
    {
      "epoch": 2.218350519641012,
      "grad_norm": 0.567099392414093,
      "learning_rate": 3.170826638014121e-05,
      "loss": 1.0409,
      "step": 54750
    },
    {
      "epoch": 2.2187556978181155,
      "grad_norm": 0.6018804907798767,
      "learning_rate": 3.1677255915749606e-05,
      "loss": 1.0452,
      "step": 54760
    },
    {
      "epoch": 2.219160875995219,
      "grad_norm": 0.6685236096382141,
      "learning_rate": 3.164625776875634e-05,
      "loss": 1.062,
      "step": 54770
    },
    {
      "epoch": 2.219566054172322,
      "grad_norm": 0.6416273713111877,
      "learning_rate": 3.161527194474976e-05,
      "loss": 1.1078,
      "step": 54780
    },
    {
      "epoch": 2.2199712323494256,
      "grad_norm": 0.6279966235160828,
      "learning_rate": 3.158429844931611e-05,
      "loss": 1.056,
      "step": 54790
    },
    {
      "epoch": 2.220376410526529,
      "grad_norm": 0.6663196682929993,
      "learning_rate": 3.1553337288039364e-05,
      "loss": 1.0581,
      "step": 54800
    },
    {
      "epoch": 2.2207815887036326,
      "grad_norm": 0.7373790740966797,
      "learning_rate": 3.1522388466501304e-05,
      "loss": 1.0594,
      "step": 54810
    },
    {
      "epoch": 2.2211867668807357,
      "grad_norm": 0.6620219945907593,
      "learning_rate": 3.149145199028142e-05,
      "loss": 1.055,
      "step": 54820
    },
    {
      "epoch": 2.221591945057839,
      "grad_norm": 0.6288800239562988,
      "learning_rate": 3.1460527864957034e-05,
      "loss": 1.0561,
      "step": 54830
    },
    {
      "epoch": 2.2219971232349427,
      "grad_norm": 0.6511500477790833,
      "learning_rate": 3.1429616096103266e-05,
      "loss": 1.0015,
      "step": 54840
    },
    {
      "epoch": 2.2224023014120458,
      "grad_norm": 0.6722560524940491,
      "learning_rate": 3.1398716689292893e-05,
      "loss": 1.0315,
      "step": 54850
    },
    {
      "epoch": 2.2228074795891493,
      "grad_norm": 0.6668419241905212,
      "learning_rate": 3.136782965009658e-05,
      "loss": 1.0255,
      "step": 54860
    },
    {
      "epoch": 2.2232126577662528,
      "grad_norm": 0.773957371711731,
      "learning_rate": 3.1336954984082725e-05,
      "loss": 1.0684,
      "step": 54870
    },
    {
      "epoch": 2.2236178359433563,
      "grad_norm": 0.6181584596633911,
      "learning_rate": 3.1306092696817515e-05,
      "loss": 1.0259,
      "step": 54880
    },
    {
      "epoch": 2.2240230141204593,
      "grad_norm": 0.6155064105987549,
      "learning_rate": 3.1275242793864814e-05,
      "loss": 1.0194,
      "step": 54890
    },
    {
      "epoch": 2.224428192297563,
      "grad_norm": 0.6177005767822266,
      "learning_rate": 3.1244405280786396e-05,
      "loss": 1.0692,
      "step": 54900
    },
    {
      "epoch": 2.2248333704746663,
      "grad_norm": 0.6818276047706604,
      "learning_rate": 3.121358016314162e-05,
      "loss": 1.0689,
      "step": 54910
    },
    {
      "epoch": 2.2252385486517694,
      "grad_norm": 0.5897628664970398,
      "learning_rate": 3.118276744648786e-05,
      "loss": 1.0672,
      "step": 54920
    },
    {
      "epoch": 2.225643726828873,
      "grad_norm": 0.6461270451545715,
      "learning_rate": 3.115196713638e-05,
      "loss": 1.0143,
      "step": 54930
    },
    {
      "epoch": 2.2260489050059764,
      "grad_norm": 0.7057691216468811,
      "learning_rate": 3.112117923837088e-05,
      "loss": 1.1014,
      "step": 54940
    },
    {
      "epoch": 2.22645408318308,
      "grad_norm": 0.5709009766578674,
      "learning_rate": 3.1090403758010934e-05,
      "loss": 1.0501,
      "step": 54950
    },
    {
      "epoch": 2.226859261360183,
      "grad_norm": 0.6262062788009644,
      "learning_rate": 3.1059640700848524e-05,
      "loss": 1.0819,
      "step": 54960
    },
    {
      "epoch": 2.2272644395372865,
      "grad_norm": 0.6120871305465698,
      "learning_rate": 3.102889007242959e-05,
      "loss": 1.0436,
      "step": 54970
    },
    {
      "epoch": 2.22766961771439,
      "grad_norm": 0.6254662275314331,
      "learning_rate": 3.0998151878298055e-05,
      "loss": 1.0478,
      "step": 54980
    },
    {
      "epoch": 2.2280747958914935,
      "grad_norm": 0.6068799495697021,
      "learning_rate": 3.0967426123995393e-05,
      "loss": 1.0205,
      "step": 54990
    },
    {
      "epoch": 2.2284799740685965,
      "grad_norm": 0.7061730027198792,
      "learning_rate": 3.093671281506099e-05,
      "loss": 1.0568,
      "step": 55000
    },
    {
      "epoch": 2.2288851522457,
      "grad_norm": 0.6869016289710999,
      "learning_rate": 3.0906011957031846e-05,
      "loss": 1.0329,
      "step": 55010
    },
    {
      "epoch": 2.2292903304228036,
      "grad_norm": 0.6463878154754639,
      "learning_rate": 3.0875323555442846e-05,
      "loss": 1.086,
      "step": 55020
    },
    {
      "epoch": 2.2296955085999066,
      "grad_norm": 0.5963410139083862,
      "learning_rate": 3.0844647615826497e-05,
      "loss": 1.0382,
      "step": 55030
    },
    {
      "epoch": 2.23010068677701,
      "grad_norm": 0.612596333026886,
      "learning_rate": 3.081398414371325e-05,
      "loss": 1.0391,
      "step": 55040
    },
    {
      "epoch": 2.2305058649541136,
      "grad_norm": 0.6495521068572998,
      "learning_rate": 3.078333314463111e-05,
      "loss": 1.0346,
      "step": 55050
    },
    {
      "epoch": 2.230911043131217,
      "grad_norm": 0.6306731700897217,
      "learning_rate": 3.075269462410596e-05,
      "loss": 1.0706,
      "step": 55060
    },
    {
      "epoch": 2.23131622130832,
      "grad_norm": 0.6343292593955994,
      "learning_rate": 3.072206858766134e-05,
      "loss": 1.0795,
      "step": 55070
    },
    {
      "epoch": 2.2317213994854237,
      "grad_norm": 0.6452826261520386,
      "learning_rate": 3.069145504081864e-05,
      "loss": 1.034,
      "step": 55080
    },
    {
      "epoch": 2.232126577662527,
      "grad_norm": 0.703775942325592,
      "learning_rate": 3.0660853989096915e-05,
      "loss": 1.0473,
      "step": 55090
    },
    {
      "epoch": 2.2325317558396307,
      "grad_norm": 0.6965835094451904,
      "learning_rate": 3.063026543801305e-05,
      "loss": 1.0419,
      "step": 55100
    },
    {
      "epoch": 2.2329369340167338,
      "grad_norm": 0.645672619342804,
      "learning_rate": 3.059968939308155e-05,
      "loss": 1.0162,
      "step": 55110
    },
    {
      "epoch": 2.2333421121938373,
      "grad_norm": 0.6982325911521912,
      "learning_rate": 3.0569125859814797e-05,
      "loss": 1.0198,
      "step": 55120
    },
    {
      "epoch": 2.2337472903709408,
      "grad_norm": 0.6326655745506287,
      "learning_rate": 3.053857484372288e-05,
      "loss": 1.0534,
      "step": 55130
    },
    {
      "epoch": 2.234152468548044,
      "grad_norm": 0.6022027134895325,
      "learning_rate": 3.050803635031355e-05,
      "loss": 1.0639,
      "step": 55140
    },
    {
      "epoch": 2.2345576467251473,
      "grad_norm": 0.6611401438713074,
      "learning_rate": 3.047751038509239e-05,
      "loss": 1.0458,
      "step": 55150
    },
    {
      "epoch": 2.234962824902251,
      "grad_norm": 0.6656332612037659,
      "learning_rate": 3.0446996953562723e-05,
      "loss": 1.0584,
      "step": 55160
    },
    {
      "epoch": 2.2353680030793543,
      "grad_norm": 0.6671144366264343,
      "learning_rate": 3.0416496061225586e-05,
      "loss": 1.0527,
      "step": 55170
    },
    {
      "epoch": 2.2357731812564574,
      "grad_norm": 0.6702262163162231,
      "learning_rate": 3.0386007713579724e-05,
      "loss": 1.0169,
      "step": 55180
    },
    {
      "epoch": 2.236178359433561,
      "grad_norm": 0.6877419352531433,
      "learning_rate": 3.0355531916121704e-05,
      "loss": 1.0483,
      "step": 55190
    },
    {
      "epoch": 2.2365835376106644,
      "grad_norm": 0.6109175086021423,
      "learning_rate": 3.0325068674345682e-05,
      "loss": 1.0769,
      "step": 55200
    },
    {
      "epoch": 2.2369887157877675,
      "grad_norm": 0.6967020034790039,
      "learning_rate": 3.029461799374378e-05,
      "loss": 1.0934,
      "step": 55210
    },
    {
      "epoch": 2.237393893964871,
      "grad_norm": 0.6225603222846985,
      "learning_rate": 3.026417987980563e-05,
      "loss": 1.0252,
      "step": 55220
    },
    {
      "epoch": 2.2377990721419745,
      "grad_norm": 0.6825239062309265,
      "learning_rate": 3.023375433801876e-05,
      "loss": 1.0649,
      "step": 55230
    },
    {
      "epoch": 2.238204250319078,
      "grad_norm": 0.6592296361923218,
      "learning_rate": 3.020334137386829e-05,
      "loss": 1.0305,
      "step": 55240
    },
    {
      "epoch": 2.238609428496181,
      "grad_norm": 0.6264320611953735,
      "learning_rate": 3.0172940992837194e-05,
      "loss": 1.052,
      "step": 55250
    },
    {
      "epoch": 2.2390146066732846,
      "grad_norm": 0.6958689093589783,
      "learning_rate": 3.0142553200406108e-05,
      "loss": 1.0953,
      "step": 55260
    },
    {
      "epoch": 2.239419784850388,
      "grad_norm": 0.6903350949287415,
      "learning_rate": 3.011217800205347e-05,
      "loss": 1.0413,
      "step": 55270
    },
    {
      "epoch": 2.239824963027491,
      "grad_norm": 0.5723147988319397,
      "learning_rate": 3.0081815403255332e-05,
      "loss": 1.0139,
      "step": 55280
    },
    {
      "epoch": 2.2402301412045946,
      "grad_norm": 0.6260440349578857,
      "learning_rate": 3.0051465409485603e-05,
      "loss": 1.0809,
      "step": 55290
    },
    {
      "epoch": 2.240635319381698,
      "grad_norm": 0.6741366386413574,
      "learning_rate": 3.0021128026215784e-05,
      "loss": 1.0724,
      "step": 55300
    },
    {
      "epoch": 2.2410404975588016,
      "grad_norm": 0.70045405626297,
      "learning_rate": 2.9990803258915223e-05,
      "loss": 1.085,
      "step": 55310
    },
    {
      "epoch": 2.2414456757359047,
      "grad_norm": 0.6448278427124023,
      "learning_rate": 2.9960491113050937e-05,
      "loss": 1.0823,
      "step": 55320
    },
    {
      "epoch": 2.241850853913008,
      "grad_norm": 0.6323899626731873,
      "learning_rate": 2.993019159408771e-05,
      "loss": 1.0478,
      "step": 55330
    },
    {
      "epoch": 2.2422560320901117,
      "grad_norm": 0.6433828473091125,
      "learning_rate": 2.9899904707487948e-05,
      "loss": 1.059,
      "step": 55340
    },
    {
      "epoch": 2.242661210267215,
      "grad_norm": 0.6222715377807617,
      "learning_rate": 2.9869630458711927e-05,
      "loss": 1.0411,
      "step": 55350
    },
    {
      "epoch": 2.2430663884443183,
      "grad_norm": 0.6195199489593506,
      "learning_rate": 2.9839368853217453e-05,
      "loss": 1.0518,
      "step": 55360
    },
    {
      "epoch": 2.2434715666214218,
      "grad_norm": 0.67914879322052,
      "learning_rate": 2.9809119896460302e-05,
      "loss": 1.0585,
      "step": 55370
    },
    {
      "epoch": 2.2438767447985253,
      "grad_norm": 0.7056317329406738,
      "learning_rate": 2.9778883593893736e-05,
      "loss": 1.0423,
      "step": 55380
    },
    {
      "epoch": 2.2442819229756283,
      "grad_norm": 0.6255557537078857,
      "learning_rate": 2.9748659950968894e-05,
      "loss": 1.0458,
      "step": 55390
    },
    {
      "epoch": 2.244687101152732,
      "grad_norm": 0.5045526027679443,
      "learning_rate": 2.9718448973134518e-05,
      "loss": 1.0556,
      "step": 55400
    },
    {
      "epoch": 2.2450922793298353,
      "grad_norm": 0.6189839243888855,
      "learning_rate": 2.9688250665837135e-05,
      "loss": 1.0364,
      "step": 55410
    },
    {
      "epoch": 2.245497457506939,
      "grad_norm": 0.7039285898208618,
      "learning_rate": 2.9658065034520978e-05,
      "loss": 1.0634,
      "step": 55420
    },
    {
      "epoch": 2.245902635684042,
      "grad_norm": 0.655666708946228,
      "learning_rate": 2.9627892084628027e-05,
      "loss": 1.0031,
      "step": 55430
    },
    {
      "epoch": 2.2463078138611454,
      "grad_norm": 0.6916324496269226,
      "learning_rate": 2.9597731821597874e-05,
      "loss": 1.0292,
      "step": 55440
    },
    {
      "epoch": 2.246712992038249,
      "grad_norm": 0.784453272819519,
      "learning_rate": 2.956758425086792e-05,
      "loss": 1.0816,
      "step": 55450
    },
    {
      "epoch": 2.2471181702153524,
      "grad_norm": 0.6983439922332764,
      "learning_rate": 2.9537449377873273e-05,
      "loss": 1.0697,
      "step": 55460
    },
    {
      "epoch": 2.2475233483924555,
      "grad_norm": 0.6076111197471619,
      "learning_rate": 2.9507327208046666e-05,
      "loss": 1.0631,
      "step": 55470
    },
    {
      "epoch": 2.247928526569559,
      "grad_norm": 0.6898669600486755,
      "learning_rate": 2.947721774681862e-05,
      "loss": 1.0517,
      "step": 55480
    },
    {
      "epoch": 2.2483337047466625,
      "grad_norm": 0.633270800113678,
      "learning_rate": 2.944712099961736e-05,
      "loss": 1.0478,
      "step": 55490
    },
    {
      "epoch": 2.2487388829237656,
      "grad_norm": 0.6434231996536255,
      "learning_rate": 2.9417036971868837e-05,
      "loss": 1.0534,
      "step": 55500
    },
    {
      "epoch": 2.249144061100869,
      "grad_norm": 0.8000975847244263,
      "learning_rate": 2.9386965668996602e-05,
      "loss": 1.0348,
      "step": 55510
    },
    {
      "epoch": 2.2495492392779726,
      "grad_norm": 0.6342495679855347,
      "learning_rate": 2.935690709642205e-05,
      "loss": 1.006,
      "step": 55520
    },
    {
      "epoch": 2.249954417455076,
      "grad_norm": 0.6507908701896667,
      "learning_rate": 2.932686125956413e-05,
      "loss": 1.0438,
      "step": 55530
    },
    {
      "epoch": 2.250359595632179,
      "grad_norm": 0.6130563020706177,
      "learning_rate": 2.9296828163839706e-05,
      "loss": 1.0378,
      "step": 55540
    },
    {
      "epoch": 2.2507647738092826,
      "grad_norm": 0.6823384165763855,
      "learning_rate": 2.926680781466312e-05,
      "loss": 1.0829,
      "step": 55550
    },
    {
      "epoch": 2.251169951986386,
      "grad_norm": 0.6905949115753174,
      "learning_rate": 2.9236800217446593e-05,
      "loss": 1.0453,
      "step": 55560
    },
    {
      "epoch": 2.251575130163489,
      "grad_norm": 0.6699017286300659,
      "learning_rate": 2.92068053775999e-05,
      "loss": 1.0382,
      "step": 55570
    },
    {
      "epoch": 2.2519803083405927,
      "grad_norm": 0.5688890218734741,
      "learning_rate": 2.9176823300530642e-05,
      "loss": 1.0278,
      "step": 55580
    },
    {
      "epoch": 2.252385486517696,
      "grad_norm": 0.6496602892875671,
      "learning_rate": 2.9146853991643984e-05,
      "loss": 1.0792,
      "step": 55590
    },
    {
      "epoch": 2.2527906646947997,
      "grad_norm": 0.6362513899803162,
      "learning_rate": 2.9116897456342986e-05,
      "loss": 1.0533,
      "step": 55600
    },
    {
      "epoch": 2.2531958428719028,
      "grad_norm": 0.6281322240829468,
      "learning_rate": 2.9086953700028197e-05,
      "loss": 1.0733,
      "step": 55610
    },
    {
      "epoch": 2.2536010210490063,
      "grad_norm": 0.5817422866821289,
      "learning_rate": 2.9057022728098015e-05,
      "loss": 1.0247,
      "step": 55620
    },
    {
      "epoch": 2.25400619922611,
      "grad_norm": 0.6715766787528992,
      "learning_rate": 2.9027104545948414e-05,
      "loss": 1.0405,
      "step": 55630
    },
    {
      "epoch": 2.254411377403213,
      "grad_norm": 0.6007046103477478,
      "learning_rate": 2.8997199158973178e-05,
      "loss": 1.0161,
      "step": 55640
    },
    {
      "epoch": 2.2548165555803164,
      "grad_norm": 0.7473832964897156,
      "learning_rate": 2.8967306572563636e-05,
      "loss": 1.0711,
      "step": 55650
    },
    {
      "epoch": 2.25522173375742,
      "grad_norm": 0.61555016040802,
      "learning_rate": 2.8937426792109023e-05,
      "loss": 1.0469,
      "step": 55660
    },
    {
      "epoch": 2.2556269119345234,
      "grad_norm": 0.6324512958526611,
      "learning_rate": 2.8907559822996045e-05,
      "loss": 1.0101,
      "step": 55670
    },
    {
      "epoch": 2.2560320901116264,
      "grad_norm": 0.6125521659851074,
      "learning_rate": 2.8877705670609266e-05,
      "loss": 1.033,
      "step": 55680
    },
    {
      "epoch": 2.25643726828873,
      "grad_norm": 0.6607523560523987,
      "learning_rate": 2.8847864340330788e-05,
      "loss": 1.0411,
      "step": 55690
    },
    {
      "epoch": 2.2568424464658334,
      "grad_norm": 0.6692434549331665,
      "learning_rate": 2.8818035837540537e-05,
      "loss": 1.0498,
      "step": 55700
    },
    {
      "epoch": 2.257247624642937,
      "grad_norm": 0.6802632212638855,
      "learning_rate": 2.8788220167616065e-05,
      "loss": 1.0638,
      "step": 55710
    },
    {
      "epoch": 2.25765280282004,
      "grad_norm": 0.6383114457130432,
      "learning_rate": 2.875841733593263e-05,
      "loss": 1.0204,
      "step": 55720
    },
    {
      "epoch": 2.2580579809971435,
      "grad_norm": 0.6414245367050171,
      "learning_rate": 2.8728627347863125e-05,
      "loss": 1.0632,
      "step": 55730
    },
    {
      "epoch": 2.258463159174247,
      "grad_norm": 0.6280530691146851,
      "learning_rate": 2.8698850208778206e-05,
      "loss": 1.0521,
      "step": 55740
    },
    {
      "epoch": 2.25886833735135,
      "grad_norm": 0.6922106146812439,
      "learning_rate": 2.8669085924046123e-05,
      "loss": 1.0502,
      "step": 55750
    },
    {
      "epoch": 2.2592735155284536,
      "grad_norm": 0.6661469340324402,
      "learning_rate": 2.8639334499032878e-05,
      "loss": 1.0718,
      "step": 55760
    },
    {
      "epoch": 2.259678693705557,
      "grad_norm": 0.690352737903595,
      "learning_rate": 2.8609595939102153e-05,
      "loss": 1.0273,
      "step": 55770
    },
    {
      "epoch": 2.2600838718826606,
      "grad_norm": 0.6832296848297119,
      "learning_rate": 2.8579870249615302e-05,
      "loss": 1.0409,
      "step": 55780
    },
    {
      "epoch": 2.2604890500597636,
      "grad_norm": 0.6955117583274841,
      "learning_rate": 2.8550157435931292e-05,
      "loss": 1.0327,
      "step": 55790
    },
    {
      "epoch": 2.260894228236867,
      "grad_norm": 0.7331977486610413,
      "learning_rate": 2.852045750340685e-05,
      "loss": 1.0131,
      "step": 55800
    },
    {
      "epoch": 2.2612994064139706,
      "grad_norm": 0.6682060360908508,
      "learning_rate": 2.849077045739641e-05,
      "loss": 1.1056,
      "step": 55810
    },
    {
      "epoch": 2.261704584591074,
      "grad_norm": 0.6116932034492493,
      "learning_rate": 2.8461096303251934e-05,
      "loss": 1.0399,
      "step": 55820
    },
    {
      "epoch": 2.262109762768177,
      "grad_norm": 0.588554859161377,
      "learning_rate": 2.8431435046323208e-05,
      "loss": 1.0706,
      "step": 55830
    },
    {
      "epoch": 2.2625149409452807,
      "grad_norm": 0.8350497484207153,
      "learning_rate": 2.840178669195763e-05,
      "loss": 1.0723,
      "step": 55840
    },
    {
      "epoch": 2.262920119122384,
      "grad_norm": 0.632033109664917,
      "learning_rate": 2.837215124550031e-05,
      "loss": 1.0671,
      "step": 55850
    },
    {
      "epoch": 2.2633252972994873,
      "grad_norm": 0.6697351336479187,
      "learning_rate": 2.8342528712293948e-05,
      "loss": 1.0292,
      "step": 55860
    },
    {
      "epoch": 2.263730475476591,
      "grad_norm": 0.6317589282989502,
      "learning_rate": 2.8312919097678992e-05,
      "loss": 1.0414,
      "step": 55870
    },
    {
      "epoch": 2.2641356536536943,
      "grad_norm": 0.7065878510475159,
      "learning_rate": 2.8283322406993552e-05,
      "loss": 1.0379,
      "step": 55880
    },
    {
      "epoch": 2.2645408318307974,
      "grad_norm": 0.6658704876899719,
      "learning_rate": 2.8253738645573402e-05,
      "loss": 1.0434,
      "step": 55890
    },
    {
      "epoch": 2.264946010007901,
      "grad_norm": 0.6473925709724426,
      "learning_rate": 2.8224167818751947e-05,
      "loss": 1.0344,
      "step": 55900
    },
    {
      "epoch": 2.2653511881850044,
      "grad_norm": 0.6336280107498169,
      "learning_rate": 2.8194609931860316e-05,
      "loss": 1.0246,
      "step": 55910
    },
    {
      "epoch": 2.265756366362108,
      "grad_norm": 0.6263164281845093,
      "learning_rate": 2.8165064990227252e-05,
      "loss": 1.0684,
      "step": 55920
    },
    {
      "epoch": 2.266161544539211,
      "grad_norm": 0.650327742099762,
      "learning_rate": 2.8135532999179204e-05,
      "loss": 1.0437,
      "step": 55930
    },
    {
      "epoch": 2.2665667227163144,
      "grad_norm": 0.7132824063301086,
      "learning_rate": 2.810601396404029e-05,
      "loss": 1.0228,
      "step": 55940
    },
    {
      "epoch": 2.266971900893418,
      "grad_norm": 0.5761677622795105,
      "learning_rate": 2.8076507890132288e-05,
      "loss": 1.0514,
      "step": 55950
    },
    {
      "epoch": 2.2673770790705214,
      "grad_norm": 0.7160036563873291,
      "learning_rate": 2.8047014782774584e-05,
      "loss": 1.0447,
      "step": 55960
    },
    {
      "epoch": 2.2677822572476245,
      "grad_norm": 0.6661969423294067,
      "learning_rate": 2.8017534647284317e-05,
      "loss": 1.0501,
      "step": 55970
    },
    {
      "epoch": 2.268187435424728,
      "grad_norm": 0.5795893669128418,
      "learning_rate": 2.7988067488976156e-05,
      "loss": 1.069,
      "step": 55980
    },
    {
      "epoch": 2.2685926136018315,
      "grad_norm": 0.703864574432373,
      "learning_rate": 2.7958613313162642e-05,
      "loss": 1.0434,
      "step": 55990
    },
    {
      "epoch": 2.2689977917789346,
      "grad_norm": 0.5719252824783325,
      "learning_rate": 2.7929172125153746e-05,
      "loss": 1.0619,
      "step": 56000
    },
    {
      "epoch": 2.269402969956038,
      "grad_norm": 0.6821990609169006,
      "learning_rate": 2.7899743930257272e-05,
      "loss": 1.0387,
      "step": 56010
    },
    {
      "epoch": 2.2698081481331416,
      "grad_norm": 0.6921728849411011,
      "learning_rate": 2.787032873377855e-05,
      "loss": 1.0697,
      "step": 56020
    },
    {
      "epoch": 2.270213326310245,
      "grad_norm": 0.5791845917701721,
      "learning_rate": 2.7840926541020683e-05,
      "loss": 0.9957,
      "step": 56030
    },
    {
      "epoch": 2.270618504487348,
      "grad_norm": 0.7392063140869141,
      "learning_rate": 2.7811537357284278e-05,
      "loss": 1.0696,
      "step": 56040
    },
    {
      "epoch": 2.2710236826644516,
      "grad_norm": 0.6404615044593811,
      "learning_rate": 2.7782161187867818e-05,
      "loss": 1.0641,
      "step": 56050
    },
    {
      "epoch": 2.271428860841555,
      "grad_norm": 0.7310931086540222,
      "learning_rate": 2.7752798038067217e-05,
      "loss": 1.0343,
      "step": 56060
    },
    {
      "epoch": 2.2718340390186587,
      "grad_norm": 0.6737620830535889,
      "learning_rate": 2.7723447913176213e-05,
      "loss": 1.0096,
      "step": 56070
    },
    {
      "epoch": 2.2722392171957617,
      "grad_norm": 0.6160619258880615,
      "learning_rate": 2.7694110818486054e-05,
      "loss": 1.0612,
      "step": 56080
    },
    {
      "epoch": 2.2726443953728652,
      "grad_norm": 0.6518389582633972,
      "learning_rate": 2.7664786759285723e-05,
      "loss": 1.0348,
      "step": 56090
    },
    {
      "epoch": 2.2730495735499687,
      "grad_norm": 0.6721343994140625,
      "learning_rate": 2.7635475740861848e-05,
      "loss": 1.0582,
      "step": 56100
    },
    {
      "epoch": 2.273454751727072,
      "grad_norm": 0.6647982597351074,
      "learning_rate": 2.760617776849873e-05,
      "loss": 1.0641,
      "step": 56110
    },
    {
      "epoch": 2.2738599299041753,
      "grad_norm": 0.6459821462631226,
      "learning_rate": 2.7576892847478207e-05,
      "loss": 1.0639,
      "step": 56120
    },
    {
      "epoch": 2.274265108081279,
      "grad_norm": 0.6334119439125061,
      "learning_rate": 2.754762098307988e-05,
      "loss": 1.0303,
      "step": 56130
    },
    {
      "epoch": 2.2746702862583823,
      "grad_norm": 0.5799537301063538,
      "learning_rate": 2.7518362180580994e-05,
      "loss": 1.0583,
      "step": 56140
    },
    {
      "epoch": 2.2750754644354854,
      "grad_norm": 0.6618200540542603,
      "learning_rate": 2.7489116445256314e-05,
      "loss": 1.0784,
      "step": 56150
    },
    {
      "epoch": 2.275480642612589,
      "grad_norm": 0.8368687033653259,
      "learning_rate": 2.7459883782378392e-05,
      "loss": 1.0021,
      "step": 56160
    },
    {
      "epoch": 2.2758858207896924,
      "grad_norm": 0.6060855984687805,
      "learning_rate": 2.7430664197217347e-05,
      "loss": 1.0389,
      "step": 56170
    },
    {
      "epoch": 2.276290998966796,
      "grad_norm": 0.6663442850112915,
      "learning_rate": 2.7401457695041e-05,
      "loss": 1.012,
      "step": 56180
    },
    {
      "epoch": 2.276696177143899,
      "grad_norm": 0.6662267446517944,
      "learning_rate": 2.737226428111471e-05,
      "loss": 1.0326,
      "step": 56190
    },
    {
      "epoch": 2.2771013553210024,
      "grad_norm": 0.6489757299423218,
      "learning_rate": 2.734308396070161e-05,
      "loss": 1.0471,
      "step": 56200
    },
    {
      "epoch": 2.277506533498106,
      "grad_norm": 0.6254786849021912,
      "learning_rate": 2.7313916739062308e-05,
      "loss": 1.08,
      "step": 56210
    },
    {
      "epoch": 2.277911711675209,
      "grad_norm": 0.6205195188522339,
      "learning_rate": 2.7284762621455252e-05,
      "loss": 1.0722,
      "step": 56220
    },
    {
      "epoch": 2.2783168898523125,
      "grad_norm": 0.6444239616394043,
      "learning_rate": 2.725562161313634e-05,
      "loss": 1.0345,
      "step": 56230
    },
    {
      "epoch": 2.278722068029416,
      "grad_norm": 0.6185008883476257,
      "learning_rate": 2.722649371935926e-05,
      "loss": 1.0464,
      "step": 56240
    },
    {
      "epoch": 2.279127246206519,
      "grad_norm": 0.6429424285888672,
      "learning_rate": 2.7197378945375175e-05,
      "loss": 1.0674,
      "step": 56250
    },
    {
      "epoch": 2.2795324243836226,
      "grad_norm": 0.6310341954231262,
      "learning_rate": 2.7168277296433053e-05,
      "loss": 1.0584,
      "step": 56260
    },
    {
      "epoch": 2.279937602560726,
      "grad_norm": 0.6796350479125977,
      "learning_rate": 2.7139188777779313e-05,
      "loss": 1.0798,
      "step": 56270
    },
    {
      "epoch": 2.2803427807378296,
      "grad_norm": 0.6229960322380066,
      "learning_rate": 2.7110113394658243e-05,
      "loss": 1.029,
      "step": 56280
    },
    {
      "epoch": 2.2807479589149326,
      "grad_norm": 0.644614577293396,
      "learning_rate": 2.708105115231151e-05,
      "loss": 0.9964,
      "step": 56290
    },
    {
      "epoch": 2.281153137092036,
      "grad_norm": 0.6383907794952393,
      "learning_rate": 2.705200205597861e-05,
      "loss": 1.0089,
      "step": 56300
    },
    {
      "epoch": 2.2815583152691397,
      "grad_norm": 0.7242418527603149,
      "learning_rate": 2.702296611089652e-05,
      "loss": 1.0733,
      "step": 56310
    },
    {
      "epoch": 2.281963493446243,
      "grad_norm": 0.6196380257606506,
      "learning_rate": 2.6993943322299953e-05,
      "loss": 1.0311,
      "step": 56320
    },
    {
      "epoch": 2.2823686716233462,
      "grad_norm": 0.6015592217445374,
      "learning_rate": 2.6964933695421192e-05,
      "loss": 1.0584,
      "step": 56330
    },
    {
      "epoch": 2.2827738498004497,
      "grad_norm": 0.7340041399002075,
      "learning_rate": 2.6935937235490215e-05,
      "loss": 1.0632,
      "step": 56340
    },
    {
      "epoch": 2.2831790279775532,
      "grad_norm": 0.5908708572387695,
      "learning_rate": 2.6906953947734516e-05,
      "loss": 1.0327,
      "step": 56350
    },
    {
      "epoch": 2.2835842061546563,
      "grad_norm": 0.7221953868865967,
      "learning_rate": 2.687798383737933e-05,
      "loss": 1.0454,
      "step": 56360
    },
    {
      "epoch": 2.28398938433176,
      "grad_norm": 0.5763204097747803,
      "learning_rate": 2.6849026909647403e-05,
      "loss": 1.0499,
      "step": 56370
    },
    {
      "epoch": 2.2843945625088633,
      "grad_norm": 0.7920113801956177,
      "learning_rate": 2.682008316975919e-05,
      "loss": 1.0374,
      "step": 56380
    },
    {
      "epoch": 2.284799740685967,
      "grad_norm": 0.6387757062911987,
      "learning_rate": 2.6791152622932746e-05,
      "loss": 1.0684,
      "step": 56390
    },
    {
      "epoch": 2.28520491886307,
      "grad_norm": 0.6420643329620361,
      "learning_rate": 2.6762235274383772e-05,
      "loss": 1.064,
      "step": 56400
    },
    {
      "epoch": 2.2856100970401734,
      "grad_norm": 0.7290070056915283,
      "learning_rate": 2.673333112932551e-05,
      "loss": 1.0213,
      "step": 56410
    },
    {
      "epoch": 2.286015275217277,
      "grad_norm": 0.6447329521179199,
      "learning_rate": 2.6704440192968895e-05,
      "loss": 1.0588,
      "step": 56420
    },
    {
      "epoch": 2.2864204533943804,
      "grad_norm": 0.6553655862808228,
      "learning_rate": 2.6675562470522488e-05,
      "loss": 1.048,
      "step": 56430
    },
    {
      "epoch": 2.2868256315714834,
      "grad_norm": 0.6424555778503418,
      "learning_rate": 2.664669796719238e-05,
      "loss": 1.048,
      "step": 56440
    },
    {
      "epoch": 2.287230809748587,
      "grad_norm": 0.7197215557098389,
      "learning_rate": 2.6617846688182367e-05,
      "loss": 1.04,
      "step": 56450
    },
    {
      "epoch": 2.2876359879256905,
      "grad_norm": 0.6145633459091187,
      "learning_rate": 2.6589008638693834e-05,
      "loss": 1.0226,
      "step": 56460
    },
    {
      "epoch": 2.2880411661027935,
      "grad_norm": 0.6794785857200623,
      "learning_rate": 2.65601838239258e-05,
      "loss": 1.052,
      "step": 56470
    },
    {
      "epoch": 2.288446344279897,
      "grad_norm": 0.6646171808242798,
      "learning_rate": 2.6531372249074827e-05,
      "loss": 1.048,
      "step": 56480
    },
    {
      "epoch": 2.2888515224570005,
      "grad_norm": 0.729332447052002,
      "learning_rate": 2.650257391933516e-05,
      "loss": 1.068,
      "step": 56490
    },
    {
      "epoch": 2.289256700634104,
      "grad_norm": 0.6335723996162415,
      "learning_rate": 2.647378883989864e-05,
      "loss": 1.0116,
      "step": 56500
    },
    {
      "epoch": 2.289661878811207,
      "grad_norm": 0.5968024730682373,
      "learning_rate": 2.644501701595474e-05,
      "loss": 1.0548,
      "step": 56510
    },
    {
      "epoch": 2.2900670569883106,
      "grad_norm": 0.6600881814956665,
      "learning_rate": 2.641625845269047e-05,
      "loss": 1.0599,
      "step": 56520
    },
    {
      "epoch": 2.290472235165414,
      "grad_norm": 0.7034698724746704,
      "learning_rate": 2.6387513155290535e-05,
      "loss": 1.0315,
      "step": 56530
    },
    {
      "epoch": 2.2908774133425176,
      "grad_norm": 0.6492295265197754,
      "learning_rate": 2.6358781128937172e-05,
      "loss": 1.063,
      "step": 56540
    },
    {
      "epoch": 2.2912825915196207,
      "grad_norm": 0.664213240146637,
      "learning_rate": 2.633006237881028e-05,
      "loss": 1.0388,
      "step": 56550
    },
    {
      "epoch": 2.291687769696724,
      "grad_norm": 0.689761757850647,
      "learning_rate": 2.630135691008736e-05,
      "loss": 1.0539,
      "step": 56560
    },
    {
      "epoch": 2.2920929478738277,
      "grad_norm": 0.649197518825531,
      "learning_rate": 2.6272664727943518e-05,
      "loss": 1.0375,
      "step": 56570
    },
    {
      "epoch": 2.2924981260509307,
      "grad_norm": 0.7230448126792908,
      "learning_rate": 2.624398583755142e-05,
      "loss": 1.0663,
      "step": 56580
    },
    {
      "epoch": 2.2929033042280342,
      "grad_norm": 0.682887077331543,
      "learning_rate": 2.6215320244081408e-05,
      "loss": 1.0477,
      "step": 56590
    },
    {
      "epoch": 2.2933084824051377,
      "grad_norm": 0.6743301749229431,
      "learning_rate": 2.6186667952701304e-05,
      "loss": 1.0867,
      "step": 56600
    },
    {
      "epoch": 2.293713660582241,
      "grad_norm": 0.6887637376785278,
      "learning_rate": 2.6158028968576743e-05,
      "loss": 1.0737,
      "step": 56610
    },
    {
      "epoch": 2.2941188387593443,
      "grad_norm": 0.5826771855354309,
      "learning_rate": 2.6129403296870736e-05,
      "loss": 1.0698,
      "step": 56620
    },
    {
      "epoch": 2.294524016936448,
      "grad_norm": 0.6228000521659851,
      "learning_rate": 2.6100790942744057e-05,
      "loss": 1.0317,
      "step": 56630
    },
    {
      "epoch": 2.2949291951135513,
      "grad_norm": 0.614663302898407,
      "learning_rate": 2.607219191135495e-05,
      "loss": 1.0484,
      "step": 56640
    },
    {
      "epoch": 2.2953343732906544,
      "grad_norm": 0.6163212656974792,
      "learning_rate": 2.6043606207859384e-05,
      "loss": 1.0417,
      "step": 56650
    },
    {
      "epoch": 2.295739551467758,
      "grad_norm": 0.6638268828392029,
      "learning_rate": 2.6015033837410773e-05,
      "loss": 1.0587,
      "step": 56660
    },
    {
      "epoch": 2.2961447296448614,
      "grad_norm": 0.6013580560684204,
      "learning_rate": 2.5986474805160332e-05,
      "loss": 1.0244,
      "step": 56670
    },
    {
      "epoch": 2.296549907821965,
      "grad_norm": 0.6609846949577332,
      "learning_rate": 2.5957929116256675e-05,
      "loss": 1.0544,
      "step": 56680
    },
    {
      "epoch": 2.296955085999068,
      "grad_norm": 0.6574509143829346,
      "learning_rate": 2.592939677584614e-05,
      "loss": 1.035,
      "step": 56690
    },
    {
      "epoch": 2.2973602641761715,
      "grad_norm": 0.6693636178970337,
      "learning_rate": 2.5900877789072555e-05,
      "loss": 1.0134,
      "step": 56700
    },
    {
      "epoch": 2.297765442353275,
      "grad_norm": 0.7143217325210571,
      "learning_rate": 2.587237216107745e-05,
      "loss": 1.0456,
      "step": 56710
    },
    {
      "epoch": 2.298170620530378,
      "grad_norm": 0.6699699759483337,
      "learning_rate": 2.584387989699982e-05,
      "loss": 1.0276,
      "step": 56720
    },
    {
      "epoch": 2.2985757987074815,
      "grad_norm": 0.5952284336090088,
      "learning_rate": 2.5815401001976436e-05,
      "loss": 1.0185,
      "step": 56730
    },
    {
      "epoch": 2.298980976884585,
      "grad_norm": 0.698862612247467,
      "learning_rate": 2.5786935481141437e-05,
      "loss": 1.0383,
      "step": 56740
    },
    {
      "epoch": 2.2993861550616885,
      "grad_norm": 0.6765432953834534,
      "learning_rate": 2.5758483339626738e-05,
      "loss": 1.0198,
      "step": 56750
    },
    {
      "epoch": 2.2997913332387916,
      "grad_norm": 0.6382190585136414,
      "learning_rate": 2.573004458256171e-05,
      "loss": 1.0157,
      "step": 56760
    },
    {
      "epoch": 2.300196511415895,
      "grad_norm": 0.6408072710037231,
      "learning_rate": 2.5701619215073368e-05,
      "loss": 1.023,
      "step": 56770
    },
    {
      "epoch": 2.3006016895929986,
      "grad_norm": 0.6725713610649109,
      "learning_rate": 2.5673207242286333e-05,
      "loss": 1.0737,
      "step": 56780
    },
    {
      "epoch": 2.301006867770102,
      "grad_norm": 0.6962560415267944,
      "learning_rate": 2.5644808669322806e-05,
      "loss": 1.0414,
      "step": 56790
    },
    {
      "epoch": 2.301412045947205,
      "grad_norm": 0.6331755518913269,
      "learning_rate": 2.56164235013025e-05,
      "loss": 1.037,
      "step": 56800
    },
    {
      "epoch": 2.3018172241243087,
      "grad_norm": 0.5822718143463135,
      "learning_rate": 2.55880517433428e-05,
      "loss": 1.0638,
      "step": 56810
    },
    {
      "epoch": 2.302222402301412,
      "grad_norm": 0.7130267024040222,
      "learning_rate": 2.5559693400558658e-05,
      "loss": 1.0445,
      "step": 56820
    },
    {
      "epoch": 2.3026275804785152,
      "grad_norm": 0.6402716040611267,
      "learning_rate": 2.5531348478062534e-05,
      "loss": 0.9919,
      "step": 56830
    },
    {
      "epoch": 2.3030327586556187,
      "grad_norm": 0.651801586151123,
      "learning_rate": 2.5503016980964543e-05,
      "loss": 1.0498,
      "step": 56840
    },
    {
      "epoch": 2.3034379368327222,
      "grad_norm": 0.6985906958580017,
      "learning_rate": 2.5474698914372375e-05,
      "loss": 1.0215,
      "step": 56850
    },
    {
      "epoch": 2.3038431150098257,
      "grad_norm": 0.6844033598899841,
      "learning_rate": 2.5446394283391306e-05,
      "loss": 1.0718,
      "step": 56860
    },
    {
      "epoch": 2.304248293186929,
      "grad_norm": 0.6353937983512878,
      "learning_rate": 2.541810309312411e-05,
      "loss": 1.05,
      "step": 56870
    },
    {
      "epoch": 2.3046534713640323,
      "grad_norm": 0.5916405320167542,
      "learning_rate": 2.5389825348671247e-05,
      "loss": 1.0481,
      "step": 56880
    },
    {
      "epoch": 2.305058649541136,
      "grad_norm": 0.6486285328865051,
      "learning_rate": 2.536156105513062e-05,
      "loss": 1.0128,
      "step": 56890
    },
    {
      "epoch": 2.3054638277182393,
      "grad_norm": 0.6263782382011414,
      "learning_rate": 2.5333310217597906e-05,
      "loss": 1.0601,
      "step": 56900
    },
    {
      "epoch": 2.3058690058953424,
      "grad_norm": 0.6845192909240723,
      "learning_rate": 2.5305072841166132e-05,
      "loss": 1.0265,
      "step": 56910
    },
    {
      "epoch": 2.306274184072446,
      "grad_norm": 0.6700462698936462,
      "learning_rate": 2.527684893092609e-05,
      "loss": 1.0506,
      "step": 56920
    },
    {
      "epoch": 2.3066793622495494,
      "grad_norm": 0.9081945419311523,
      "learning_rate": 2.524863849196598e-05,
      "loss": 1.0889,
      "step": 56930
    },
    {
      "epoch": 2.3070845404266525,
      "grad_norm": 0.6793072819709778,
      "learning_rate": 2.5220441529371698e-05,
      "loss": 1.0331,
      "step": 56940
    },
    {
      "epoch": 2.307489718603756,
      "grad_norm": 0.6034645438194275,
      "learning_rate": 2.5192258048226645e-05,
      "loss": 1.0614,
      "step": 56950
    },
    {
      "epoch": 2.3078948967808595,
      "grad_norm": 0.6056315898895264,
      "learning_rate": 2.5164088053611845e-05,
      "loss": 1.0479,
      "step": 56960
    },
    {
      "epoch": 2.3083000749579625,
      "grad_norm": 0.6649630665779114,
      "learning_rate": 2.5135931550605808e-05,
      "loss": 1.0372,
      "step": 56970
    },
    {
      "epoch": 2.308705253135066,
      "grad_norm": 0.7183367609977722,
      "learning_rate": 2.5107788544284714e-05,
      "loss": 1.0291,
      "step": 56980
    },
    {
      "epoch": 2.3091104313121695,
      "grad_norm": 0.6439093351364136,
      "learning_rate": 2.507965903972219e-05,
      "loss": 1.0319,
      "step": 56990
    },
    {
      "epoch": 2.309515609489273,
      "grad_norm": 0.6764087080955505,
      "learning_rate": 2.5051543041989534e-05,
      "loss": 1.0677,
      "step": 57000
    },
    {
      "epoch": 2.309920787666376,
      "grad_norm": 0.5912669897079468,
      "learning_rate": 2.5023440556155563e-05,
      "loss": 1.0722,
      "step": 57010
    },
    {
      "epoch": 2.3103259658434796,
      "grad_norm": 0.6029124855995178,
      "learning_rate": 2.4995351587286696e-05,
      "loss": 1.0837,
      "step": 57020
    },
    {
      "epoch": 2.310731144020583,
      "grad_norm": 0.6045593619346619,
      "learning_rate": 2.4967276140446826e-05,
      "loss": 1.0433,
      "step": 57030
    },
    {
      "epoch": 2.3111363221976866,
      "grad_norm": 0.6627344489097595,
      "learning_rate": 2.4939214220697526e-05,
      "loss": 1.0388,
      "step": 57040
    },
    {
      "epoch": 2.3115415003747897,
      "grad_norm": 0.6179713606834412,
      "learning_rate": 2.4911165833097782e-05,
      "loss": 1.0302,
      "step": 57050
    },
    {
      "epoch": 2.311946678551893,
      "grad_norm": 0.6580325365066528,
      "learning_rate": 2.488313098270435e-05,
      "loss": 1.0298,
      "step": 57060
    },
    {
      "epoch": 2.3123518567289967,
      "grad_norm": 0.7134519219398499,
      "learning_rate": 2.485510967457132e-05,
      "loss": 1.0594,
      "step": 57070
    },
    {
      "epoch": 2.3127570349060997,
      "grad_norm": 0.6690366268157959,
      "learning_rate": 2.4827101913750518e-05,
      "loss": 1.072,
      "step": 57080
    },
    {
      "epoch": 2.3131622130832032,
      "grad_norm": 0.6124110221862793,
      "learning_rate": 2.4799107705291193e-05,
      "loss": 1.0472,
      "step": 57090
    },
    {
      "epoch": 2.3135673912603067,
      "grad_norm": 0.6635573506355286,
      "learning_rate": 2.477112705424024e-05,
      "loss": 1.0507,
      "step": 57100
    },
    {
      "epoch": 2.3139725694374103,
      "grad_norm": 0.6212814450263977,
      "learning_rate": 2.474315996564209e-05,
      "loss": 1.0372,
      "step": 57110
    },
    {
      "epoch": 2.3143777476145133,
      "grad_norm": 0.5741321444511414,
      "learning_rate": 2.471520644453873e-05,
      "loss": 1.0392,
      "step": 57120
    },
    {
      "epoch": 2.314782925791617,
      "grad_norm": 0.7147338390350342,
      "learning_rate": 2.4687266495969663e-05,
      "loss": 1.0582,
      "step": 57130
    },
    {
      "epoch": 2.3151881039687203,
      "grad_norm": 0.6533762216567993,
      "learning_rate": 2.4659340124971975e-05,
      "loss": 1.0577,
      "step": 57140
    },
    {
      "epoch": 2.315593282145824,
      "grad_norm": 0.623112142086029,
      "learning_rate": 2.463142733658035e-05,
      "loss": 1.0409,
      "step": 57150
    },
    {
      "epoch": 2.315998460322927,
      "grad_norm": 0.6676403880119324,
      "learning_rate": 2.4603528135826902e-05,
      "loss": 1.0733,
      "step": 57160
    },
    {
      "epoch": 2.3164036385000304,
      "grad_norm": 0.6756773591041565,
      "learning_rate": 2.4575642527741415e-05,
      "loss": 1.0742,
      "step": 57170
    },
    {
      "epoch": 2.316808816677134,
      "grad_norm": 0.6008915901184082,
      "learning_rate": 2.4547770517351177e-05,
      "loss": 1.0574,
      "step": 57180
    },
    {
      "epoch": 2.317213994854237,
      "grad_norm": 0.6508607268333435,
      "learning_rate": 2.451991210968104e-05,
      "loss": 1.0393,
      "step": 57190
    },
    {
      "epoch": 2.3176191730313405,
      "grad_norm": 0.6683459877967834,
      "learning_rate": 2.4492067309753354e-05,
      "loss": 1.0418,
      "step": 57200
    },
    {
      "epoch": 2.318024351208444,
      "grad_norm": 0.6071228384971619,
      "learning_rate": 2.446423612258808e-05,
      "loss": 0.9913,
      "step": 57210
    },
    {
      "epoch": 2.3184295293855475,
      "grad_norm": 0.670391321182251,
      "learning_rate": 2.4436418553202633e-05,
      "loss": 1.048,
      "step": 57220
    },
    {
      "epoch": 2.3188347075626505,
      "grad_norm": 0.6485451459884644,
      "learning_rate": 2.4408614606612136e-05,
      "loss": 1.0535,
      "step": 57230
    },
    {
      "epoch": 2.319239885739754,
      "grad_norm": 0.5832093358039856,
      "learning_rate": 2.4380824287829074e-05,
      "loss": 1.0399,
      "step": 57240
    },
    {
      "epoch": 2.3196450639168575,
      "grad_norm": 0.6465844511985779,
      "learning_rate": 2.435304760186361e-05,
      "loss": 1.0694,
      "step": 57250
    },
    {
      "epoch": 2.320050242093961,
      "grad_norm": 0.6308861374855042,
      "learning_rate": 2.432528455372335e-05,
      "loss": 1.0981,
      "step": 57260
    },
    {
      "epoch": 2.320455420271064,
      "grad_norm": 0.6862841248512268,
      "learning_rate": 2.429753514841352e-05,
      "loss": 1.0282,
      "step": 57270
    },
    {
      "epoch": 2.3208605984481676,
      "grad_norm": 0.6674177050590515,
      "learning_rate": 2.4269799390936798e-05,
      "loss": 1.0534,
      "step": 57280
    },
    {
      "epoch": 2.321265776625271,
      "grad_norm": 0.633453369140625,
      "learning_rate": 2.424207728629355e-05,
      "loss": 1.0656,
      "step": 57290
    },
    {
      "epoch": 2.321670954802374,
      "grad_norm": 0.6845259070396423,
      "learning_rate": 2.4214368839481515e-05,
      "loss": 1.006,
      "step": 57300
    },
    {
      "epoch": 2.3220761329794777,
      "grad_norm": 0.6499998569488525,
      "learning_rate": 2.4186674055496083e-05,
      "loss": 1.0329,
      "step": 57310
    },
    {
      "epoch": 2.322481311156581,
      "grad_norm": 0.6871117353439331,
      "learning_rate": 2.4158992939330104e-05,
      "loss": 1.0581,
      "step": 57320
    },
    {
      "epoch": 2.3228864893336842,
      "grad_norm": 0.6471672058105469,
      "learning_rate": 2.4131325495974046e-05,
      "loss": 1.0713,
      "step": 57330
    },
    {
      "epoch": 2.3232916675107878,
      "grad_norm": 0.6510379314422607,
      "learning_rate": 2.4103671730415767e-05,
      "loss": 1.0668,
      "step": 57340
    },
    {
      "epoch": 2.3236968456878913,
      "grad_norm": 0.604845404624939,
      "learning_rate": 2.4076031647640896e-05,
      "loss": 1.0428,
      "step": 57350
    },
    {
      "epoch": 2.3241020238649948,
      "grad_norm": 0.6148521900177002,
      "learning_rate": 2.4048405252632366e-05,
      "loss": 1.0631,
      "step": 57360
    },
    {
      "epoch": 2.324507202042098,
      "grad_norm": 0.7093408107757568,
      "learning_rate": 2.4020792550370784e-05,
      "loss": 1.0617,
      "step": 57370
    },
    {
      "epoch": 2.3249123802192013,
      "grad_norm": 0.5920068621635437,
      "learning_rate": 2.399319354583418e-05,
      "loss": 1.062,
      "step": 57380
    },
    {
      "epoch": 2.325317558396305,
      "grad_norm": 0.6779560446739197,
      "learning_rate": 2.3965608243998217e-05,
      "loss": 1.067,
      "step": 57390
    },
    {
      "epoch": 2.3257227365734083,
      "grad_norm": 0.677445650100708,
      "learning_rate": 2.3938036649836025e-05,
      "loss": 1.0673,
      "step": 57400
    },
    {
      "epoch": 2.3261279147505114,
      "grad_norm": 0.6860247850418091,
      "learning_rate": 2.391047876831831e-05,
      "loss": 1.0394,
      "step": 57410
    },
    {
      "epoch": 2.326533092927615,
      "grad_norm": 0.6656750440597534,
      "learning_rate": 2.3882934604413242e-05,
      "loss": 1.0074,
      "step": 57420
    },
    {
      "epoch": 2.3269382711047184,
      "grad_norm": 0.6555866599082947,
      "learning_rate": 2.3855404163086558e-05,
      "loss": 1.0839,
      "step": 57430
    },
    {
      "epoch": 2.3273434492818215,
      "grad_norm": 0.7267152070999146,
      "learning_rate": 2.3827887449301556e-05,
      "loss": 1.0746,
      "step": 57440
    },
    {
      "epoch": 2.327748627458925,
      "grad_norm": 0.6745039820671082,
      "learning_rate": 2.3800384468018954e-05,
      "loss": 1.0519,
      "step": 57450
    },
    {
      "epoch": 2.3281538056360285,
      "grad_norm": 0.9352084398269653,
      "learning_rate": 2.37728952241971e-05,
      "loss": 1.0585,
      "step": 57460
    },
    {
      "epoch": 2.328558983813132,
      "grad_norm": 0.6149017214775085,
      "learning_rate": 2.3745419722791805e-05,
      "loss": 1.0636,
      "step": 57470
    },
    {
      "epoch": 2.328964161990235,
      "grad_norm": 0.6716863512992859,
      "learning_rate": 2.3717957968756476e-05,
      "loss": 1.1029,
      "step": 57480
    },
    {
      "epoch": 2.3293693401673385,
      "grad_norm": 0.6731002330780029,
      "learning_rate": 2.3690509967041918e-05,
      "loss": 0.9758,
      "step": 57490
    },
    {
      "epoch": 2.329774518344442,
      "grad_norm": 0.6418077945709229,
      "learning_rate": 2.3663075722596583e-05,
      "loss": 1.0308,
      "step": 57500
    },
    {
      "epoch": 2.3301796965215456,
      "grad_norm": 0.671885073184967,
      "learning_rate": 2.3635655240366305e-05,
      "loss": 1.0647,
      "step": 57510
    },
    {
      "epoch": 2.3305848746986486,
      "grad_norm": 0.7639000415802002,
      "learning_rate": 2.3608248525294628e-05,
      "loss": 1.0464,
      "step": 57520
    },
    {
      "epoch": 2.330990052875752,
      "grad_norm": 0.6468989253044128,
      "learning_rate": 2.3580855582322425e-05,
      "loss": 1.0538,
      "step": 57530
    },
    {
      "epoch": 2.3313952310528556,
      "grad_norm": 0.604217529296875,
      "learning_rate": 2.355347641638822e-05,
      "loss": 1.0743,
      "step": 57540
    },
    {
      "epoch": 2.3318004092299587,
      "grad_norm": 0.7018433213233948,
      "learning_rate": 2.3526111032427933e-05,
      "loss": 1.0394,
      "step": 57550
    },
    {
      "epoch": 2.332205587407062,
      "grad_norm": 0.672577440738678,
      "learning_rate": 2.3498759435375105e-05,
      "loss": 1.0376,
      "step": 57560
    },
    {
      "epoch": 2.3326107655841657,
      "grad_norm": 0.6036897897720337,
      "learning_rate": 2.347142163016074e-05,
      "loss": 1.0693,
      "step": 57570
    },
    {
      "epoch": 2.333015943761269,
      "grad_norm": 0.6345599293708801,
      "learning_rate": 2.344409762171341e-05,
      "loss": 1.0541,
      "step": 57580
    },
    {
      "epoch": 2.3334211219383723,
      "grad_norm": 0.6532865762710571,
      "learning_rate": 2.3416787414959097e-05,
      "loss": 1.0223,
      "step": 57590
    },
    {
      "epoch": 2.3338263001154758,
      "grad_norm": 0.6102273464202881,
      "learning_rate": 2.3389491014821395e-05,
      "loss": 1.0466,
      "step": 57600
    },
    {
      "epoch": 2.3342314782925793,
      "grad_norm": 0.6772716641426086,
      "learning_rate": 2.336220842622133e-05,
      "loss": 1.0385,
      "step": 57610
    },
    {
      "epoch": 2.3346366564696828,
      "grad_norm": 0.7215996384620667,
      "learning_rate": 2.3334939654077503e-05,
      "loss": 1.066,
      "step": 57620
    },
    {
      "epoch": 2.335041834646786,
      "grad_norm": 0.7528132796287537,
      "learning_rate": 2.3307684703305998e-05,
      "loss": 1.0604,
      "step": 57630
    },
    {
      "epoch": 2.3354470128238893,
      "grad_norm": 0.6593201160430908,
      "learning_rate": 2.328044357882042e-05,
      "loss": 1.012,
      "step": 57640
    },
    {
      "epoch": 2.335852191000993,
      "grad_norm": 0.628547728061676,
      "learning_rate": 2.3253216285531832e-05,
      "loss": 1.0849,
      "step": 57650
    },
    {
      "epoch": 2.336257369178096,
      "grad_norm": 0.6694075465202332,
      "learning_rate": 2.322600282834888e-05,
      "loss": 1.0631,
      "step": 57660
    },
    {
      "epoch": 2.3366625473551994,
      "grad_norm": 0.6752238273620605,
      "learning_rate": 2.31988032121776e-05,
      "loss": 1.0631,
      "step": 57670
    },
    {
      "epoch": 2.337067725532303,
      "grad_norm": 0.5653058290481567,
      "learning_rate": 2.317161744192171e-05,
      "loss": 1.013,
      "step": 57680
    },
    {
      "epoch": 2.337472903709406,
      "grad_norm": 0.6135436296463013,
      "learning_rate": 2.3144445522482273e-05,
      "loss": 1.052,
      "step": 57690
    },
    {
      "epoch": 2.3378780818865095,
      "grad_norm": 0.6465479135513306,
      "learning_rate": 2.311728745875793e-05,
      "loss": 1.0156,
      "step": 57700
    },
    {
      "epoch": 2.338283260063613,
      "grad_norm": 0.6298975944519043,
      "learning_rate": 2.309014325564478e-05,
      "loss": 1.0832,
      "step": 57710
    },
    {
      "epoch": 2.3386884382407165,
      "grad_norm": 0.6936813592910767,
      "learning_rate": 2.3063012918036486e-05,
      "loss": 1.0335,
      "step": 57720
    },
    {
      "epoch": 2.3390936164178195,
      "grad_norm": 0.5714254379272461,
      "learning_rate": 2.303589645082411e-05,
      "loss": 1.0508,
      "step": 57730
    },
    {
      "epoch": 2.339498794594923,
      "grad_norm": 0.6619501113891602,
      "learning_rate": 2.3008793858896372e-05,
      "loss": 1.0391,
      "step": 57740
    },
    {
      "epoch": 2.3399039727720266,
      "grad_norm": 0.6950544714927673,
      "learning_rate": 2.298170514713931e-05,
      "loss": 1.0444,
      "step": 57750
    },
    {
      "epoch": 2.34030915094913,
      "grad_norm": 0.7381371259689331,
      "learning_rate": 2.295463032043662e-05,
      "loss": 1.0527,
      "step": 57760
    },
    {
      "epoch": 2.340714329126233,
      "grad_norm": 0.7286491990089417,
      "learning_rate": 2.2927569383669356e-05,
      "loss": 1.0721,
      "step": 57770
    },
    {
      "epoch": 2.3411195073033366,
      "grad_norm": 0.689113974571228,
      "learning_rate": 2.2900522341716156e-05,
      "loss": 1.0517,
      "step": 57780
    },
    {
      "epoch": 2.34152468548044,
      "grad_norm": 0.7046433687210083,
      "learning_rate": 2.2873489199453135e-05,
      "loss": 1.0409,
      "step": 57790
    },
    {
      "epoch": 2.341929863657543,
      "grad_norm": 0.6613414287567139,
      "learning_rate": 2.2846469961753915e-05,
      "loss": 1.0296,
      "step": 57800
    },
    {
      "epoch": 2.3423350418346467,
      "grad_norm": 0.6228213310241699,
      "learning_rate": 2.281946463348955e-05,
      "loss": 1.0587,
      "step": 57810
    },
    {
      "epoch": 2.34274022001175,
      "grad_norm": 0.7852486968040466,
      "learning_rate": 2.2792473219528655e-05,
      "loss": 1.038,
      "step": 57820
    },
    {
      "epoch": 2.3431453981888537,
      "grad_norm": 0.6409097909927368,
      "learning_rate": 2.276549572473734e-05,
      "loss": 1.0538,
      "step": 57830
    },
    {
      "epoch": 2.3435505763659568,
      "grad_norm": 0.626924991607666,
      "learning_rate": 2.2738532153979108e-05,
      "loss": 1.0381,
      "step": 57840
    },
    {
      "epoch": 2.3439557545430603,
      "grad_norm": 0.6788647770881653,
      "learning_rate": 2.2711582512115058e-05,
      "loss": 1.0743,
      "step": 57850
    },
    {
      "epoch": 2.3443609327201638,
      "grad_norm": 0.6562239527702332,
      "learning_rate": 2.268464680400374e-05,
      "loss": 1.044,
      "step": 57860
    },
    {
      "epoch": 2.3447661108972673,
      "grad_norm": 0.6330710649490356,
      "learning_rate": 2.265772503450122e-05,
      "loss": 1.0709,
      "step": 57870
    },
    {
      "epoch": 2.3451712890743703,
      "grad_norm": 0.7328686714172363,
      "learning_rate": 2.2630817208460965e-05,
      "loss": 1.0502,
      "step": 57880
    },
    {
      "epoch": 2.345576467251474,
      "grad_norm": 0.6977989077568054,
      "learning_rate": 2.2603923330734045e-05,
      "loss": 1.0332,
      "step": 57890
    },
    {
      "epoch": 2.3459816454285773,
      "grad_norm": 0.6445995569229126,
      "learning_rate": 2.2577043406168875e-05,
      "loss": 1.0175,
      "step": 57900
    },
    {
      "epoch": 2.3463868236056804,
      "grad_norm": 0.5807134509086609,
      "learning_rate": 2.255017743961153e-05,
      "loss": 1.0329,
      "step": 57910
    },
    {
      "epoch": 2.346792001782784,
      "grad_norm": 0.6417651176452637,
      "learning_rate": 2.2523325435905418e-05,
      "loss": 1.0686,
      "step": 57920
    },
    {
      "epoch": 2.3471971799598874,
      "grad_norm": 0.6441731452941895,
      "learning_rate": 2.249648739989153e-05,
      "loss": 1.0698,
      "step": 57930
    },
    {
      "epoch": 2.347602358136991,
      "grad_norm": 0.6075198650360107,
      "learning_rate": 2.246966333640823e-05,
      "loss": 1.0445,
      "step": 57940
    },
    {
      "epoch": 2.348007536314094,
      "grad_norm": 0.6124346852302551,
      "learning_rate": 2.2442853250291507e-05,
      "loss": 1.0256,
      "step": 57950
    },
    {
      "epoch": 2.3484127144911975,
      "grad_norm": 0.6125638484954834,
      "learning_rate": 2.241605714637465e-05,
      "loss": 1.0365,
      "step": 57960
    },
    {
      "epoch": 2.348817892668301,
      "grad_norm": 0.7719059586524963,
      "learning_rate": 2.2389275029488633e-05,
      "loss": 1.0497,
      "step": 57970
    },
    {
      "epoch": 2.3492230708454045,
      "grad_norm": 0.6280518174171448,
      "learning_rate": 2.2362506904461745e-05,
      "loss": 1.0542,
      "step": 57980
    },
    {
      "epoch": 2.3496282490225076,
      "grad_norm": 0.6659229397773743,
      "learning_rate": 2.2335752776119856e-05,
      "loss": 1.0483,
      "step": 57990
    },
    {
      "epoch": 2.350033427199611,
      "grad_norm": 0.6924601793289185,
      "learning_rate": 2.230901264928621e-05,
      "loss": 1.0418,
      "step": 58000
    },
    {
      "epoch": 2.3504386053767146,
      "grad_norm": 0.6887273192405701,
      "learning_rate": 2.2282286528781605e-05,
      "loss": 1.0619,
      "step": 58010
    },
    {
      "epoch": 2.3508437835538176,
      "grad_norm": 0.6617075800895691,
      "learning_rate": 2.2255574419424318e-05,
      "loss": 1.0567,
      "step": 58020
    },
    {
      "epoch": 2.351248961730921,
      "grad_norm": 0.7297280430793762,
      "learning_rate": 2.222887632603008e-05,
      "loss": 1.0238,
      "step": 58030
    },
    {
      "epoch": 2.3516541399080246,
      "grad_norm": 0.6333580613136292,
      "learning_rate": 2.2202192253412045e-05,
      "loss": 1.0589,
      "step": 58040
    },
    {
      "epoch": 2.3520593180851277,
      "grad_norm": 0.6550553441047668,
      "learning_rate": 2.217552220638095e-05,
      "loss": 1.0484,
      "step": 58050
    },
    {
      "epoch": 2.352464496262231,
      "grad_norm": 0.6718667149543762,
      "learning_rate": 2.2148866189744878e-05,
      "loss": 1.0746,
      "step": 58060
    },
    {
      "epoch": 2.3528696744393347,
      "grad_norm": 0.631168782711029,
      "learning_rate": 2.212222420830946e-05,
      "loss": 1.0475,
      "step": 58070
    },
    {
      "epoch": 2.353274852616438,
      "grad_norm": 0.6771163940429688,
      "learning_rate": 2.2095596266877782e-05,
      "loss": 1.0526,
      "step": 58080
    },
    {
      "epoch": 2.3536800307935413,
      "grad_norm": 0.65813148021698,
      "learning_rate": 2.2068982370250447e-05,
      "loss": 1.0915,
      "step": 58090
    },
    {
      "epoch": 2.3540852089706448,
      "grad_norm": 0.7099075317382812,
      "learning_rate": 2.2042382523225392e-05,
      "loss": 1.0134,
      "step": 58100
    },
    {
      "epoch": 2.3544903871477483,
      "grad_norm": 0.6993504166603088,
      "learning_rate": 2.2015796730598158e-05,
      "loss": 1.0045,
      "step": 58110
    },
    {
      "epoch": 2.354895565324852,
      "grad_norm": 0.7104092240333557,
      "learning_rate": 2.1989224997161706e-05,
      "loss": 1.0494,
      "step": 58120
    },
    {
      "epoch": 2.355300743501955,
      "grad_norm": 0.6249347925186157,
      "learning_rate": 2.196266732770641e-05,
      "loss": 1.0734,
      "step": 58130
    },
    {
      "epoch": 2.3557059216790583,
      "grad_norm": 0.6225873827934265,
      "learning_rate": 2.1936123727020187e-05,
      "loss": 1.0859,
      "step": 58140
    },
    {
      "epoch": 2.356111099856162,
      "grad_norm": 0.7345860600471497,
      "learning_rate": 2.1909594199888372e-05,
      "loss": 1.0186,
      "step": 58150
    },
    {
      "epoch": 2.356516278033265,
      "grad_norm": 0.7037703990936279,
      "learning_rate": 2.1883078751093822e-05,
      "loss": 1.0557,
      "step": 58160
    },
    {
      "epoch": 2.3569214562103684,
      "grad_norm": 0.6878785490989685,
      "learning_rate": 2.1856577385416732e-05,
      "loss": 1.0413,
      "step": 58170
    },
    {
      "epoch": 2.357326634387472,
      "grad_norm": 0.6111530661582947,
      "learning_rate": 2.183009010763488e-05,
      "loss": 1.0474,
      "step": 58180
    },
    {
      "epoch": 2.3577318125645754,
      "grad_norm": 0.6618905663490295,
      "learning_rate": 2.180361692252345e-05,
      "loss": 1.0636,
      "step": 58190
    },
    {
      "epoch": 2.3581369907416785,
      "grad_norm": 0.6867774724960327,
      "learning_rate": 2.1777157834855132e-05,
      "loss": 1.0236,
      "step": 58200
    },
    {
      "epoch": 2.358542168918782,
      "grad_norm": 0.6144596338272095,
      "learning_rate": 2.175071284939998e-05,
      "loss": 1.0762,
      "step": 58210
    },
    {
      "epoch": 2.3589473470958855,
      "grad_norm": 0.6692119836807251,
      "learning_rate": 2.172428197092561e-05,
      "loss": 1.0286,
      "step": 58220
    },
    {
      "epoch": 2.359352525272989,
      "grad_norm": 0.6927149295806885,
      "learning_rate": 2.1697865204197e-05,
      "loss": 1.0819,
      "step": 58230
    },
    {
      "epoch": 2.359757703450092,
      "grad_norm": 0.6123718619346619,
      "learning_rate": 2.167146255397666e-05,
      "loss": 1.0691,
      "step": 58240
    },
    {
      "epoch": 2.3601628816271956,
      "grad_norm": 0.6617963314056396,
      "learning_rate": 2.1645074025024524e-05,
      "loss": 1.0336,
      "step": 58250
    },
    {
      "epoch": 2.360568059804299,
      "grad_norm": 0.655885636806488,
      "learning_rate": 2.1618699622098016e-05,
      "loss": 1.0494,
      "step": 58260
    },
    {
      "epoch": 2.360973237981402,
      "grad_norm": 0.724457323551178,
      "learning_rate": 2.1592339349951916e-05,
      "loss": 1.0419,
      "step": 58270
    },
    {
      "epoch": 2.3613784161585056,
      "grad_norm": 0.6037267446517944,
      "learning_rate": 2.1565993213338588e-05,
      "loss": 1.1094,
      "step": 58280
    },
    {
      "epoch": 2.361783594335609,
      "grad_norm": 0.568701446056366,
      "learning_rate": 2.153966121700769e-05,
      "loss": 1.0463,
      "step": 58290
    },
    {
      "epoch": 2.3621887725127126,
      "grad_norm": 0.6398252248764038,
      "learning_rate": 2.151334336570653e-05,
      "loss": 1.0776,
      "step": 58300
    },
    {
      "epoch": 2.3625939506898157,
      "grad_norm": 0.5729824900627136,
      "learning_rate": 2.1487039664179685e-05,
      "loss": 1.0465,
      "step": 58310
    },
    {
      "epoch": 2.362999128866919,
      "grad_norm": 0.7074784636497498,
      "learning_rate": 2.146075011716929e-05,
      "loss": 1.0388,
      "step": 58320
    },
    {
      "epoch": 2.3634043070440227,
      "grad_norm": 0.6676985621452332,
      "learning_rate": 2.1434474729414855e-05,
      "loss": 1.0478,
      "step": 58330
    },
    {
      "epoch": 2.363809485221126,
      "grad_norm": 0.5631843209266663,
      "learning_rate": 2.140821350565343e-05,
      "loss": 1.0726,
      "step": 58340
    },
    {
      "epoch": 2.3642146633982293,
      "grad_norm": 0.678950309753418,
      "learning_rate": 2.138196645061935e-05,
      "loss": 1.0697,
      "step": 58350
    },
    {
      "epoch": 2.364619841575333,
      "grad_norm": 0.6082729697227478,
      "learning_rate": 2.1355733569044635e-05,
      "loss": 1.0758,
      "step": 58360
    },
    {
      "epoch": 2.3650250197524363,
      "grad_norm": 0.7339044809341431,
      "learning_rate": 2.132951486565852e-05,
      "loss": 1.0217,
      "step": 58370
    },
    {
      "epoch": 2.3654301979295393,
      "grad_norm": 0.7266520261764526,
      "learning_rate": 2.1303310345187844e-05,
      "loss": 1.0249,
      "step": 58380
    },
    {
      "epoch": 2.365835376106643,
      "grad_norm": 0.7655950784683228,
      "learning_rate": 2.1277120012356755e-05,
      "loss": 1.0622,
      "step": 58390
    },
    {
      "epoch": 2.3662405542837464,
      "grad_norm": 0.6145873069763184,
      "learning_rate": 2.125094387188695e-05,
      "loss": 1.0741,
      "step": 58400
    },
    {
      "epoch": 2.3666457324608494,
      "grad_norm": 0.7033446431159973,
      "learning_rate": 2.122478192849753e-05,
      "loss": 1.0938,
      "step": 58410
    },
    {
      "epoch": 2.367050910637953,
      "grad_norm": 0.6478738784790039,
      "learning_rate": 2.1198634186905064e-05,
      "loss": 1.0239,
      "step": 58420
    },
    {
      "epoch": 2.3674560888150564,
      "grad_norm": 0.6836292147636414,
      "learning_rate": 2.117250065182349e-05,
      "loss": 1.0292,
      "step": 58430
    },
    {
      "epoch": 2.36786126699216,
      "grad_norm": 0.7316177487373352,
      "learning_rate": 2.1146381327964228e-05,
      "loss": 1.047,
      "step": 58440
    },
    {
      "epoch": 2.368266445169263,
      "grad_norm": 0.6522375345230103,
      "learning_rate": 2.1120276220036184e-05,
      "loss": 1.0313,
      "step": 58450
    },
    {
      "epoch": 2.3686716233463665,
      "grad_norm": 0.5849481821060181,
      "learning_rate": 2.10941853327456e-05,
      "loss": 1.0468,
      "step": 58460
    },
    {
      "epoch": 2.36907680152347,
      "grad_norm": 0.7266647219657898,
      "learning_rate": 2.1068108670796227e-05,
      "loss": 1.0552,
      "step": 58470
    },
    {
      "epoch": 2.3694819797005735,
      "grad_norm": 0.6515419483184814,
      "learning_rate": 2.1042046238889236e-05,
      "loss": 1.0482,
      "step": 58480
    },
    {
      "epoch": 2.3698871578776766,
      "grad_norm": 0.6722886562347412,
      "learning_rate": 2.1015998041723252e-05,
      "loss": 1.0248,
      "step": 58490
    },
    {
      "epoch": 2.37029233605478,
      "grad_norm": 0.623701274394989,
      "learning_rate": 2.0989964083994252e-05,
      "loss": 1.0436,
      "step": 58500
    },
    {
      "epoch": 2.3706975142318836,
      "grad_norm": 0.6534514427185059,
      "learning_rate": 2.0963944370395773e-05,
      "loss": 1.0441,
      "step": 58510
    },
    {
      "epoch": 2.3711026924089866,
      "grad_norm": 0.6601418852806091,
      "learning_rate": 2.0937938905618627e-05,
      "loss": 1.0546,
      "step": 58520
    },
    {
      "epoch": 2.37150787058609,
      "grad_norm": 0.6837037801742554,
      "learning_rate": 2.091194769435124e-05,
      "loss": 1.0478,
      "step": 58530
    },
    {
      "epoch": 2.3719130487631936,
      "grad_norm": 0.6196555495262146,
      "learning_rate": 2.0885970741279316e-05,
      "loss": 1.0777,
      "step": 58540
    },
    {
      "epoch": 2.372318226940297,
      "grad_norm": 0.577675461769104,
      "learning_rate": 2.0860008051086077e-05,
      "loss": 0.9919,
      "step": 58550
    },
    {
      "epoch": 2.3727234051174,
      "grad_norm": 0.6575630307197571,
      "learning_rate": 2.083405962845212e-05,
      "loss": 1.037,
      "step": 58560
    },
    {
      "epoch": 2.3731285832945037,
      "grad_norm": 0.634506106376648,
      "learning_rate": 2.0808125478055505e-05,
      "loss": 1.0687,
      "step": 58570
    },
    {
      "epoch": 2.373533761471607,
      "grad_norm": 0.6461396217346191,
      "learning_rate": 2.078220560457167e-05,
      "loss": 1.0287,
      "step": 58580
    },
    {
      "epoch": 2.3739389396487107,
      "grad_norm": 0.6757138967514038,
      "learning_rate": 2.075630001267359e-05,
      "loss": 1.0142,
      "step": 58590
    },
    {
      "epoch": 2.374344117825814,
      "grad_norm": 0.6719546914100647,
      "learning_rate": 2.0730408707031522e-05,
      "loss": 1.0821,
      "step": 58600
    },
    {
      "epoch": 2.3747492960029173,
      "grad_norm": 0.6798965334892273,
      "learning_rate": 2.0704531692313268e-05,
      "loss": 1.0695,
      "step": 58610
    },
    {
      "epoch": 2.375154474180021,
      "grad_norm": 0.8047559857368469,
      "learning_rate": 2.0678668973183958e-05,
      "loss": 1.0491,
      "step": 58620
    },
    {
      "epoch": 2.375559652357124,
      "grad_norm": 0.6807461977005005,
      "learning_rate": 2.0652820554306205e-05,
      "loss": 1.0558,
      "step": 58630
    },
    {
      "epoch": 2.3759648305342274,
      "grad_norm": 0.6123344302177429,
      "learning_rate": 2.0626986440340035e-05,
      "loss": 1.0314,
      "step": 58640
    },
    {
      "epoch": 2.376370008711331,
      "grad_norm": 0.6586763858795166,
      "learning_rate": 2.0601166635942913e-05,
      "loss": 1.0459,
      "step": 58650
    },
    {
      "epoch": 2.3767751868884344,
      "grad_norm": 0.6388865113258362,
      "learning_rate": 2.0575361145769633e-05,
      "loss": 1.0548,
      "step": 58660
    },
    {
      "epoch": 2.3771803650655374,
      "grad_norm": 0.6495730876922607,
      "learning_rate": 2.054956997447255e-05,
      "loss": 1.0256,
      "step": 58670
    },
    {
      "epoch": 2.377585543242641,
      "grad_norm": 0.6040590405464172,
      "learning_rate": 2.0523793126701295e-05,
      "loss": 1.0782,
      "step": 58680
    },
    {
      "epoch": 2.3779907214197444,
      "grad_norm": 0.6262520551681519,
      "learning_rate": 2.0498030607103002e-05,
      "loss": 1.0573,
      "step": 58690
    },
    {
      "epoch": 2.378395899596848,
      "grad_norm": 0.623823344707489,
      "learning_rate": 2.0472282420322232e-05,
      "loss": 1.0466,
      "step": 58700
    },
    {
      "epoch": 2.378801077773951,
      "grad_norm": 0.7127469778060913,
      "learning_rate": 2.0446548571000935e-05,
      "loss": 1.0578,
      "step": 58710
    },
    {
      "epoch": 2.3792062559510545,
      "grad_norm": 0.6709697246551514,
      "learning_rate": 2.0420829063778425e-05,
      "loss": 1.0479,
      "step": 58720
    },
    {
      "epoch": 2.379611434128158,
      "grad_norm": 0.6607086658477783,
      "learning_rate": 2.0395123903291534e-05,
      "loss": 1.0516,
      "step": 58730
    },
    {
      "epoch": 2.380016612305261,
      "grad_norm": 0.6657174229621887,
      "learning_rate": 2.0369433094174416e-05,
      "loss": 1.0189,
      "step": 58740
    },
    {
      "epoch": 2.3804217904823646,
      "grad_norm": 0.6443727612495422,
      "learning_rate": 2.0343756641058688e-05,
      "loss": 1.052,
      "step": 58750
    },
    {
      "epoch": 2.380826968659468,
      "grad_norm": 0.6599297523498535,
      "learning_rate": 2.0318094548573365e-05,
      "loss": 1.04,
      "step": 58760
    },
    {
      "epoch": 2.381232146836571,
      "grad_norm": 0.6443716287612915,
      "learning_rate": 2.0292446821344903e-05,
      "loss": 1.0124,
      "step": 58770
    },
    {
      "epoch": 2.3816373250136746,
      "grad_norm": 0.7008005976676941,
      "learning_rate": 2.0266813463997092e-05,
      "loss": 1.0662,
      "step": 58780
    },
    {
      "epoch": 2.382042503190778,
      "grad_norm": 0.6821445822715759,
      "learning_rate": 2.0241194481151203e-05,
      "loss": 1.0675,
      "step": 58790
    },
    {
      "epoch": 2.3824476813678817,
      "grad_norm": 0.6204022169113159,
      "learning_rate": 2.021558987742589e-05,
      "loss": 1.0471,
      "step": 58800
    },
    {
      "epoch": 2.3828528595449847,
      "grad_norm": 0.6264798045158386,
      "learning_rate": 2.018999965743724e-05,
      "loss": 1.0491,
      "step": 58810
    },
    {
      "epoch": 2.383258037722088,
      "grad_norm": 0.6351835131645203,
      "learning_rate": 2.0164423825798674e-05,
      "loss": 1.0328,
      "step": 58820
    },
    {
      "epoch": 2.3836632158991917,
      "grad_norm": 0.6371484398841858,
      "learning_rate": 2.0138862387121095e-05,
      "loss": 1.0483,
      "step": 58830
    },
    {
      "epoch": 2.3840683940762952,
      "grad_norm": 0.7512366771697998,
      "learning_rate": 2.0113315346012807e-05,
      "loss": 1.047,
      "step": 58840
    },
    {
      "epoch": 2.3844735722533983,
      "grad_norm": 0.729417085647583,
      "learning_rate": 2.008778270707944e-05,
      "loss": 1.0429,
      "step": 58850
    },
    {
      "epoch": 2.384878750430502,
      "grad_norm": 0.629753053188324,
      "learning_rate": 2.006226447492412e-05,
      "loss": 1.0889,
      "step": 58860
    },
    {
      "epoch": 2.3852839286076053,
      "grad_norm": 0.7308348417282104,
      "learning_rate": 2.0036760654147335e-05,
      "loss": 1.0501,
      "step": 58870
    },
    {
      "epoch": 2.3856891067847084,
      "grad_norm": 0.7079418301582336,
      "learning_rate": 2.0011271249347008e-05,
      "loss": 1.0885,
      "step": 58880
    },
    {
      "epoch": 2.386094284961812,
      "grad_norm": 0.723892092704773,
      "learning_rate": 1.998579626511836e-05,
      "loss": 1.0394,
      "step": 58890
    },
    {
      "epoch": 2.3864994631389154,
      "grad_norm": 0.7976542115211487,
      "learning_rate": 1.9960335706054168e-05,
      "loss": 1.0551,
      "step": 58900
    },
    {
      "epoch": 2.386904641316019,
      "grad_norm": 0.6008467674255371,
      "learning_rate": 1.993488957674443e-05,
      "loss": 1.0485,
      "step": 58910
    },
    {
      "epoch": 2.387309819493122,
      "grad_norm": 0.632396936416626,
      "learning_rate": 1.990945788177676e-05,
      "loss": 1.0517,
      "step": 58920
    },
    {
      "epoch": 2.3877149976702254,
      "grad_norm": 0.6405857801437378,
      "learning_rate": 1.9884040625735945e-05,
      "loss": 1.0192,
      "step": 58930
    },
    {
      "epoch": 2.388120175847329,
      "grad_norm": 0.6411457061767578,
      "learning_rate": 1.985863781320435e-05,
      "loss": 1.0684,
      "step": 58940
    },
    {
      "epoch": 2.3885253540244324,
      "grad_norm": 0.6405032873153687,
      "learning_rate": 1.9833249448761592e-05,
      "loss": 1.0639,
      "step": 58950
    },
    {
      "epoch": 2.3889305322015355,
      "grad_norm": 0.6511332988739014,
      "learning_rate": 1.98078755369848e-05,
      "loss": 1.0609,
      "step": 58960
    },
    {
      "epoch": 2.389335710378639,
      "grad_norm": 0.675067663192749,
      "learning_rate": 1.9782516082448377e-05,
      "loss": 1.0613,
      "step": 58970
    },
    {
      "epoch": 2.3897408885557425,
      "grad_norm": 0.6389851570129395,
      "learning_rate": 1.975717108972429e-05,
      "loss": 1.0396,
      "step": 58980
    },
    {
      "epoch": 2.3901460667328456,
      "grad_norm": 0.6730941534042358,
      "learning_rate": 1.973184056338173e-05,
      "loss": 1.0716,
      "step": 58990
    },
    {
      "epoch": 2.390551244909949,
      "grad_norm": 0.6425848007202148,
      "learning_rate": 1.970652450798739e-05,
      "loss": 1.0443,
      "step": 59000
    },
    {
      "epoch": 2.3909564230870526,
      "grad_norm": 0.7030326128005981,
      "learning_rate": 1.968122292810527e-05,
      "loss": 1.0625,
      "step": 59010
    },
    {
      "epoch": 2.391361601264156,
      "grad_norm": 0.6122674345970154,
      "learning_rate": 1.9655935828296844e-05,
      "loss": 1.009,
      "step": 59020
    },
    {
      "epoch": 2.391766779441259,
      "grad_norm": 0.6584659814834595,
      "learning_rate": 1.963066321312086e-05,
      "loss": 1.0553,
      "step": 59030
    },
    {
      "epoch": 2.3921719576183627,
      "grad_norm": 0.6167944073677063,
      "learning_rate": 1.9605405087133643e-05,
      "loss": 1.0472,
      "step": 59040
    },
    {
      "epoch": 2.392577135795466,
      "grad_norm": 0.6163232326507568,
      "learning_rate": 1.95801614548887e-05,
      "loss": 1.0131,
      "step": 59050
    },
    {
      "epoch": 2.3929823139725697,
      "grad_norm": 0.6044601202011108,
      "learning_rate": 1.955493232093708e-05,
      "loss": 1.0235,
      "step": 59060
    },
    {
      "epoch": 2.3933874921496727,
      "grad_norm": 0.69999760389328,
      "learning_rate": 1.95297176898271e-05,
      "loss": 1.1001,
      "step": 59070
    },
    {
      "epoch": 2.3937926703267762,
      "grad_norm": 0.5877159237861633,
      "learning_rate": 1.9504517566104542e-05,
      "loss": 1.0386,
      "step": 59080
    },
    {
      "epoch": 2.3941978485038797,
      "grad_norm": 0.5981888175010681,
      "learning_rate": 1.9479331954312553e-05,
      "loss": 1.0379,
      "step": 59090
    },
    {
      "epoch": 2.394603026680983,
      "grad_norm": 0.7054407000541687,
      "learning_rate": 1.945416085899169e-05,
      "loss": 1.023,
      "step": 59100
    },
    {
      "epoch": 2.3950082048580863,
      "grad_norm": 0.7585617899894714,
      "learning_rate": 1.94290042846798e-05,
      "loss": 1.0316,
      "step": 59110
    },
    {
      "epoch": 2.39541338303519,
      "grad_norm": 0.684179425239563,
      "learning_rate": 1.9403862235912196e-05,
      "loss": 1.0547,
      "step": 59120
    },
    {
      "epoch": 2.395818561212293,
      "grad_norm": 0.6574723124504089,
      "learning_rate": 1.937873471722158e-05,
      "loss": 1.0269,
      "step": 59130
    },
    {
      "epoch": 2.3962237393893964,
      "grad_norm": 0.6155020594596863,
      "learning_rate": 1.935362173313796e-05,
      "loss": 0.9843,
      "step": 59140
    },
    {
      "epoch": 2.3966289175665,
      "grad_norm": 0.579308807849884,
      "learning_rate": 1.9328523288188804e-05,
      "loss": 0.989,
      "step": 59150
    },
    {
      "epoch": 2.3970340957436034,
      "grad_norm": 0.6410502195358276,
      "learning_rate": 1.9303439386898903e-05,
      "loss": 1.0263,
      "step": 59160
    },
    {
      "epoch": 2.3974392739207064,
      "grad_norm": 0.6446061134338379,
      "learning_rate": 1.927837003379047e-05,
      "loss": 1.0309,
      "step": 59170
    },
    {
      "epoch": 2.39784445209781,
      "grad_norm": 0.5676600337028503,
      "learning_rate": 1.925331523338304e-05,
      "loss": 1.0382,
      "step": 59180
    },
    {
      "epoch": 2.3982496302749134,
      "grad_norm": 0.6141343116760254,
      "learning_rate": 1.9228274990193607e-05,
      "loss": 1.0549,
      "step": 59190
    },
    {
      "epoch": 2.398654808452017,
      "grad_norm": 0.7301005125045776,
      "learning_rate": 1.9203249308736394e-05,
      "loss": 1.0744,
      "step": 59200
    },
    {
      "epoch": 2.39905998662912,
      "grad_norm": 0.7880656123161316,
      "learning_rate": 1.9178238193523213e-05,
      "loss": 1.0604,
      "step": 59210
    },
    {
      "epoch": 2.3994651648062235,
      "grad_norm": 0.6863526105880737,
      "learning_rate": 1.9153241649063035e-05,
      "loss": 1.0597,
      "step": 59220
    },
    {
      "epoch": 2.399870342983327,
      "grad_norm": 0.6691502332687378,
      "learning_rate": 1.912825967986238e-05,
      "loss": 1.0405,
      "step": 59230
    },
    {
      "epoch": 2.40027552116043,
      "grad_norm": 0.6616203784942627,
      "learning_rate": 1.9103292290425e-05,
      "loss": 1.0903,
      "step": 59240
    },
    {
      "epoch": 2.4006806993375336,
      "grad_norm": 0.7077987194061279,
      "learning_rate": 1.9078339485252084e-05,
      "loss": 1.0924,
      "step": 59250
    },
    {
      "epoch": 2.401085877514637,
      "grad_norm": 0.6631951928138733,
      "learning_rate": 1.905340126884222e-05,
      "loss": 1.0443,
      "step": 59260
    },
    {
      "epoch": 2.4014910556917406,
      "grad_norm": 0.7607508897781372,
      "learning_rate": 1.9028477645691335e-05,
      "loss": 1.0581,
      "step": 59270
    },
    {
      "epoch": 2.4018962338688437,
      "grad_norm": 0.7119076251983643,
      "learning_rate": 1.9003568620292678e-05,
      "loss": 1.0447,
      "step": 59280
    },
    {
      "epoch": 2.402301412045947,
      "grad_norm": 0.7777712941169739,
      "learning_rate": 1.8978674197136968e-05,
      "loss": 1.029,
      "step": 59290
    },
    {
      "epoch": 2.4027065902230507,
      "grad_norm": 0.6110389232635498,
      "learning_rate": 1.8953794380712187e-05,
      "loss": 1.0578,
      "step": 59300
    },
    {
      "epoch": 2.403111768400154,
      "grad_norm": 0.6334032416343689,
      "learning_rate": 1.892892917550374e-05,
      "loss": 1.0386,
      "step": 59310
    },
    {
      "epoch": 2.4035169465772572,
      "grad_norm": 0.6639935374259949,
      "learning_rate": 1.8904078585994412e-05,
      "loss": 1.0517,
      "step": 59320
    },
    {
      "epoch": 2.4039221247543607,
      "grad_norm": 0.6472568511962891,
      "learning_rate": 1.8879242616664338e-05,
      "loss": 1.0516,
      "step": 59330
    },
    {
      "epoch": 2.4043273029314642,
      "grad_norm": 0.600551426410675,
      "learning_rate": 1.8854421271990964e-05,
      "loss": 1.0174,
      "step": 59340
    },
    {
      "epoch": 2.4047324811085673,
      "grad_norm": 0.6313623189926147,
      "learning_rate": 1.8829614556449203e-05,
      "loss": 1.0558,
      "step": 59350
    },
    {
      "epoch": 2.405137659285671,
      "grad_norm": 0.6604694724082947,
      "learning_rate": 1.8804822474511208e-05,
      "loss": 1.0273,
      "step": 59360
    },
    {
      "epoch": 2.4055428374627743,
      "grad_norm": 0.7274693250656128,
      "learning_rate": 1.8780045030646597e-05,
      "loss": 1.0193,
      "step": 59370
    },
    {
      "epoch": 2.405948015639878,
      "grad_norm": 0.5722888112068176,
      "learning_rate": 1.8755282229322302e-05,
      "loss": 1.0727,
      "step": 59380
    },
    {
      "epoch": 2.406353193816981,
      "grad_norm": 0.625357449054718,
      "learning_rate": 1.873053407500266e-05,
      "loss": 1.0525,
      "step": 59390
    },
    {
      "epoch": 2.4067583719940844,
      "grad_norm": 0.6584646701812744,
      "learning_rate": 1.8705800572149268e-05,
      "loss": 1.006,
      "step": 59400
    },
    {
      "epoch": 2.407163550171188,
      "grad_norm": 0.6823252439498901,
      "learning_rate": 1.8681081725221184e-05,
      "loss": 1.0684,
      "step": 59410
    },
    {
      "epoch": 2.4075687283482914,
      "grad_norm": 0.7265653610229492,
      "learning_rate": 1.8656377538674774e-05,
      "loss": 1.0356,
      "step": 59420
    },
    {
      "epoch": 2.4079739065253944,
      "grad_norm": 0.737095296382904,
      "learning_rate": 1.86316880169638e-05,
      "loss": 1.0672,
      "step": 59430
    },
    {
      "epoch": 2.408379084702498,
      "grad_norm": 0.7122402787208557,
      "learning_rate": 1.86070131645393e-05,
      "loss": 1.0376,
      "step": 59440
    },
    {
      "epoch": 2.4087842628796015,
      "grad_norm": 0.6079379916191101,
      "learning_rate": 1.8582352985849737e-05,
      "loss": 1.0489,
      "step": 59450
    },
    {
      "epoch": 2.4091894410567045,
      "grad_norm": 0.6669487357139587,
      "learning_rate": 1.8557707485340957e-05,
      "loss": 1.0427,
      "step": 59460
    },
    {
      "epoch": 2.409594619233808,
      "grad_norm": 0.7192121148109436,
      "learning_rate": 1.853307666745605e-05,
      "loss": 1.0126,
      "step": 59470
    },
    {
      "epoch": 2.4099997974109115,
      "grad_norm": 0.6357536315917969,
      "learning_rate": 1.850846053663554e-05,
      "loss": 1.0295,
      "step": 59480
    },
    {
      "epoch": 2.4104049755880146,
      "grad_norm": 0.6996358036994934,
      "learning_rate": 1.84838590973173e-05,
      "loss": 1.0411,
      "step": 59490
    },
    {
      "epoch": 2.410810153765118,
      "grad_norm": 0.6639929413795471,
      "learning_rate": 1.8459272353936564e-05,
      "loss": 1.0793,
      "step": 59500
    },
    {
      "epoch": 2.4112153319422216,
      "grad_norm": 0.6417388916015625,
      "learning_rate": 1.8434700310925834e-05,
      "loss": 1.0348,
      "step": 59510
    },
    {
      "epoch": 2.411620510119325,
      "grad_norm": 0.6734386086463928,
      "learning_rate": 1.8410142972715073e-05,
      "loss": 1.0422,
      "step": 59520
    },
    {
      "epoch": 2.412025688296428,
      "grad_norm": 0.644486665725708,
      "learning_rate": 1.8385600343731468e-05,
      "loss": 1.0454,
      "step": 59530
    },
    {
      "epoch": 2.4124308664735317,
      "grad_norm": 0.6633933186531067,
      "learning_rate": 1.836107242839973e-05,
      "loss": 1.095,
      "step": 59540
    },
    {
      "epoch": 2.412836044650635,
      "grad_norm": 0.6782891154289246,
      "learning_rate": 1.8336559231141726e-05,
      "loss": 1.0716,
      "step": 59550
    },
    {
      "epoch": 2.4132412228277387,
      "grad_norm": 0.6809507012367249,
      "learning_rate": 1.8312060756376814e-05,
      "loss": 1.0467,
      "step": 59560
    },
    {
      "epoch": 2.4136464010048417,
      "grad_norm": 0.6637238264083862,
      "learning_rate": 1.8287577008521607e-05,
      "loss": 1.0772,
      "step": 59570
    },
    {
      "epoch": 2.4140515791819452,
      "grad_norm": 0.6843333840370178,
      "learning_rate": 1.8263107991990124e-05,
      "loss": 1.0359,
      "step": 59580
    },
    {
      "epoch": 2.4144567573590487,
      "grad_norm": 0.6409594416618347,
      "learning_rate": 1.8238653711193643e-05,
      "loss": 1.0497,
      "step": 59590
    },
    {
      "epoch": 2.414861935536152,
      "grad_norm": 0.700142502784729,
      "learning_rate": 1.8214214170540944e-05,
      "loss": 1.0442,
      "step": 59600
    },
    {
      "epoch": 2.4152671137132553,
      "grad_norm": 0.6531118750572205,
      "learning_rate": 1.818978937443796e-05,
      "loss": 1.0499,
      "step": 59610
    },
    {
      "epoch": 2.415672291890359,
      "grad_norm": 0.611826479434967,
      "learning_rate": 1.8165379327288113e-05,
      "loss": 1.0146,
      "step": 59620
    },
    {
      "epoch": 2.4160774700674623,
      "grad_norm": 0.6359519958496094,
      "learning_rate": 1.814098403349206e-05,
      "loss": 1.0331,
      "step": 59630
    },
    {
      "epoch": 2.4164826482445654,
      "grad_norm": 0.6894932985305786,
      "learning_rate": 1.8116603497447905e-05,
      "loss": 1.0818,
      "step": 59640
    },
    {
      "epoch": 2.416887826421669,
      "grad_norm": 0.6701063513755798,
      "learning_rate": 1.8092237723550943e-05,
      "loss": 1.0724,
      "step": 59650
    },
    {
      "epoch": 2.4172930045987724,
      "grad_norm": 0.6669261455535889,
      "learning_rate": 1.8067886716193993e-05,
      "loss": 1.0537,
      "step": 59660
    },
    {
      "epoch": 2.417698182775876,
      "grad_norm": 0.7227350473403931,
      "learning_rate": 1.804355047976707e-05,
      "loss": 1.0137,
      "step": 59670
    },
    {
      "epoch": 2.418103360952979,
      "grad_norm": 0.6961343884468079,
      "learning_rate": 1.8019229018657603e-05,
      "loss": 1.0433,
      "step": 59680
    },
    {
      "epoch": 2.4185085391300825,
      "grad_norm": 0.6688957810401917,
      "learning_rate": 1.7994922337250274e-05,
      "loss": 1.0189,
      "step": 59690
    },
    {
      "epoch": 2.418913717307186,
      "grad_norm": 0.6371031999588013,
      "learning_rate": 1.7970630439927184e-05,
      "loss": 1.0612,
      "step": 59700
    },
    {
      "epoch": 2.419318895484289,
      "grad_norm": 0.6631408333778381,
      "learning_rate": 1.7946353331067733e-05,
      "loss": 1.0838,
      "step": 59710
    },
    {
      "epoch": 2.4197240736613925,
      "grad_norm": 0.6626921892166138,
      "learning_rate": 1.7922091015048704e-05,
      "loss": 1.0612,
      "step": 59720
    },
    {
      "epoch": 2.420129251838496,
      "grad_norm": 0.594695508480072,
      "learning_rate": 1.7897843496244092e-05,
      "loss": 1.0716,
      "step": 59730
    },
    {
      "epoch": 2.4205344300155995,
      "grad_norm": 0.6124787926673889,
      "learning_rate": 1.787361077902534e-05,
      "loss": 1.0459,
      "step": 59740
    },
    {
      "epoch": 2.4209396081927026,
      "grad_norm": 0.6646514534950256,
      "learning_rate": 1.7849392867761216e-05,
      "loss": 1.0424,
      "step": 59750
    },
    {
      "epoch": 2.421344786369806,
      "grad_norm": 0.6428131461143494,
      "learning_rate": 1.7825189766817728e-05,
      "loss": 1.0448,
      "step": 59760
    },
    {
      "epoch": 2.4217499645469096,
      "grad_norm": 0.7067396640777588,
      "learning_rate": 1.78010014805583e-05,
      "loss": 1.0441,
      "step": 59770
    },
    {
      "epoch": 2.422155142724013,
      "grad_norm": 0.6952794194221497,
      "learning_rate": 1.777682801334367e-05,
      "loss": 1.0419,
      "step": 59780
    },
    {
      "epoch": 2.422560320901116,
      "grad_norm": 0.6875140070915222,
      "learning_rate": 1.775266936953186e-05,
      "loss": 1.0266,
      "step": 59790
    },
    {
      "epoch": 2.4229654990782197,
      "grad_norm": 0.6793460845947266,
      "learning_rate": 1.772852555347828e-05,
      "loss": 1.0415,
      "step": 59800
    },
    {
      "epoch": 2.423370677255323,
      "grad_norm": 0.6133608222007751,
      "learning_rate": 1.770439656953564e-05,
      "loss": 1.0727,
      "step": 59810
    },
    {
      "epoch": 2.4237758554324262,
      "grad_norm": 0.7567294239997864,
      "learning_rate": 1.768028242205395e-05,
      "loss": 1.0896,
      "step": 59820
    },
    {
      "epoch": 2.4241810336095297,
      "grad_norm": 0.7113649249076843,
      "learning_rate": 1.7656183115380577e-05,
      "loss": 1.0409,
      "step": 59830
    },
    {
      "epoch": 2.4245862117866332,
      "grad_norm": 0.6846340298652649,
      "learning_rate": 1.7632098653860217e-05,
      "loss": 1.0654,
      "step": 59840
    },
    {
      "epoch": 2.4249913899637363,
      "grad_norm": 0.6103577017784119,
      "learning_rate": 1.7608029041834894e-05,
      "loss": 1.0429,
      "step": 59850
    },
    {
      "epoch": 2.42539656814084,
      "grad_norm": 0.6800231337547302,
      "learning_rate": 1.75839742836439e-05,
      "loss": 1.0403,
      "step": 59860
    },
    {
      "epoch": 2.4258017463179433,
      "grad_norm": 0.6304474472999573,
      "learning_rate": 1.75599343836239e-05,
      "loss": 1.0422,
      "step": 59870
    },
    {
      "epoch": 2.426206924495047,
      "grad_norm": 0.615571916103363,
      "learning_rate": 1.753590934610888e-05,
      "loss": 1.0373,
      "step": 59880
    },
    {
      "epoch": 2.42661210267215,
      "grad_norm": 0.6116135716438293,
      "learning_rate": 1.7511899175430157e-05,
      "loss": 1.0267,
      "step": 59890
    },
    {
      "epoch": 2.4270172808492534,
      "grad_norm": 0.5915014743804932,
      "learning_rate": 1.748790387591629e-05,
      "loss": 1.0191,
      "step": 59900
    },
    {
      "epoch": 2.427422459026357,
      "grad_norm": 0.6794490814208984,
      "learning_rate": 1.746392345189326e-05,
      "loss": 1.0444,
      "step": 59910
    },
    {
      "epoch": 2.4278276372034604,
      "grad_norm": 0.6526592969894409,
      "learning_rate": 1.743995790768428e-05,
      "loss": 1.0791,
      "step": 59920
    },
    {
      "epoch": 2.4282328153805635,
      "grad_norm": 0.6025851964950562,
      "learning_rate": 1.7416007247609943e-05,
      "loss": 1.0242,
      "step": 59930
    },
    {
      "epoch": 2.428637993557667,
      "grad_norm": 0.6561797261238098,
      "learning_rate": 1.7392071475988125e-05,
      "loss": 1.0589,
      "step": 59940
    },
    {
      "epoch": 2.4290431717347705,
      "grad_norm": 0.6101284027099609,
      "learning_rate": 1.7368150597134058e-05,
      "loss": 1.0425,
      "step": 59950
    },
    {
      "epoch": 2.4294483499118735,
      "grad_norm": 0.5876065492630005,
      "learning_rate": 1.7344244615360216e-05,
      "loss": 1.0404,
      "step": 59960
    },
    {
      "epoch": 2.429853528088977,
      "grad_norm": 0.6530161499977112,
      "learning_rate": 1.7320353534976474e-05,
      "loss": 1.0546,
      "step": 59970
    },
    {
      "epoch": 2.4302587062660805,
      "grad_norm": 0.6374059319496155,
      "learning_rate": 1.729647736028993e-05,
      "loss": 1.0527,
      "step": 59980
    },
    {
      "epoch": 2.430663884443184,
      "grad_norm": 0.6424882411956787,
      "learning_rate": 1.7272616095605054e-05,
      "loss": 1.0429,
      "step": 59990
    },
    {
      "epoch": 2.431069062620287,
      "grad_norm": 0.6629862189292908,
      "learning_rate": 1.7248769745223638e-05,
      "loss": 1.0223,
      "step": 60000
    },
    {
      "epoch": 2.4314742407973906,
      "grad_norm": 0.567160964012146,
      "learning_rate": 1.722493831344476e-05,
      "loss": 1.0872,
      "step": 60010
    },
    {
      "epoch": 2.431879418974494,
      "grad_norm": 0.6303004622459412,
      "learning_rate": 1.720112180456479e-05,
      "loss": 1.05,
      "step": 60020
    },
    {
      "epoch": 2.4322845971515976,
      "grad_norm": 0.6050437688827515,
      "learning_rate": 1.717732022287746e-05,
      "loss": 1.0691,
      "step": 60030
    },
    {
      "epoch": 2.4326897753287007,
      "grad_norm": 0.6745299100875854,
      "learning_rate": 1.715353357267371e-05,
      "loss": 1.0599,
      "step": 60040
    },
    {
      "epoch": 2.433094953505804,
      "grad_norm": 0.6353136301040649,
      "learning_rate": 1.7129761858241964e-05,
      "loss": 1.0379,
      "step": 60050
    },
    {
      "epoch": 2.4335001316829077,
      "grad_norm": 0.7003951668739319,
      "learning_rate": 1.710600508386776e-05,
      "loss": 1.0191,
      "step": 60060
    },
    {
      "epoch": 2.4339053098600107,
      "grad_norm": 0.5845526456832886,
      "learning_rate": 1.7082263253834086e-05,
      "loss": 1.0389,
      "step": 60070
    },
    {
      "epoch": 2.4343104880371143,
      "grad_norm": 0.6554606556892395,
      "learning_rate": 1.705853637242113e-05,
      "loss": 1.0329,
      "step": 60080
    },
    {
      "epoch": 2.4347156662142178,
      "grad_norm": 0.6143736839294434,
      "learning_rate": 1.703482444390646e-05,
      "loss": 1.0343,
      "step": 60090
    },
    {
      "epoch": 2.4351208443913213,
      "grad_norm": 0.6299846768379211,
      "learning_rate": 1.7011127472564924e-05,
      "loss": 1.0444,
      "step": 60100
    },
    {
      "epoch": 2.4355260225684243,
      "grad_norm": 0.7418878674507141,
      "learning_rate": 1.6987445462668694e-05,
      "loss": 1.0236,
      "step": 60110
    },
    {
      "epoch": 2.435931200745528,
      "grad_norm": 0.6296384334564209,
      "learning_rate": 1.6963778418487163e-05,
      "loss": 1.0604,
      "step": 60120
    },
    {
      "epoch": 2.4363363789226313,
      "grad_norm": 0.6433230638504028,
      "learning_rate": 1.694012634428713e-05,
      "loss": 1.0353,
      "step": 60130
    },
    {
      "epoch": 2.436741557099735,
      "grad_norm": 0.6397788524627686,
      "learning_rate": 1.6916489244332677e-05,
      "loss": 1.0452,
      "step": 60140
    },
    {
      "epoch": 2.437146735276838,
      "grad_norm": 0.6080282330513,
      "learning_rate": 1.6892867122885102e-05,
      "loss": 1.0541,
      "step": 60150
    },
    {
      "epoch": 2.4375519134539414,
      "grad_norm": 0.6382706761360168,
      "learning_rate": 1.6869259984203078e-05,
      "loss": 1.0478,
      "step": 60160
    },
    {
      "epoch": 2.437957091631045,
      "grad_norm": 0.6241779923439026,
      "learning_rate": 1.6845667832542565e-05,
      "loss": 1.0311,
      "step": 60170
    },
    {
      "epoch": 2.438362269808148,
      "grad_norm": 0.642695963382721,
      "learning_rate": 1.6822090672156855e-05,
      "loss": 1.056,
      "step": 60180
    },
    {
      "epoch": 2.4387674479852515,
      "grad_norm": 0.5967938303947449,
      "learning_rate": 1.679852850729644e-05,
      "loss": 1.0183,
      "step": 60190
    },
    {
      "epoch": 2.439172626162355,
      "grad_norm": 0.7157670855522156,
      "learning_rate": 1.6774981342209208e-05,
      "loss": 1.0553,
      "step": 60200
    },
    {
      "epoch": 2.439577804339458,
      "grad_norm": 0.705159604549408,
      "learning_rate": 1.6751449181140233e-05,
      "loss": 1.0544,
      "step": 60210
    },
    {
      "epoch": 2.4399829825165615,
      "grad_norm": 0.6588870286941528,
      "learning_rate": 1.672793202833206e-05,
      "loss": 1.0355,
      "step": 60220
    },
    {
      "epoch": 2.440388160693665,
      "grad_norm": 0.6632559895515442,
      "learning_rate": 1.6704429888024332e-05,
      "loss": 1.0603,
      "step": 60230
    },
    {
      "epoch": 2.4407933388707685,
      "grad_norm": 0.6302182078361511,
      "learning_rate": 1.6680942764454133e-05,
      "loss": 1.0524,
      "step": 60240
    },
    {
      "epoch": 2.4411985170478716,
      "grad_norm": 0.6035864949226379,
      "learning_rate": 1.6657470661855745e-05,
      "loss": 1.0297,
      "step": 60250
    },
    {
      "epoch": 2.441603695224975,
      "grad_norm": 0.7166916728019714,
      "learning_rate": 1.6634013584460807e-05,
      "loss": 1.0753,
      "step": 60260
    },
    {
      "epoch": 2.4420088734020786,
      "grad_norm": 0.6407885551452637,
      "learning_rate": 1.6610571536498143e-05,
      "loss": 1.0111,
      "step": 60270
    },
    {
      "epoch": 2.442414051579182,
      "grad_norm": 0.6442380547523499,
      "learning_rate": 1.6587144522194063e-05,
      "loss": 1.0692,
      "step": 60280
    },
    {
      "epoch": 2.442819229756285,
      "grad_norm": 0.5766764879226685,
      "learning_rate": 1.656373254577197e-05,
      "loss": 1.0647,
      "step": 60290
    },
    {
      "epoch": 2.4432244079333887,
      "grad_norm": 0.6645486354827881,
      "learning_rate": 1.6540335611452685e-05,
      "loss": 1.0426,
      "step": 60300
    },
    {
      "epoch": 2.443629586110492,
      "grad_norm": 0.6585373878479004,
      "learning_rate": 1.651695372345421e-05,
      "loss": 1.0884,
      "step": 60310
    },
    {
      "epoch": 2.4440347642875953,
      "grad_norm": 0.6145046353340149,
      "learning_rate": 1.649358688599191e-05,
      "loss": 1.0492,
      "step": 60320
    },
    {
      "epoch": 2.4444399424646988,
      "grad_norm": 0.5485001802444458,
      "learning_rate": 1.6470235103278442e-05,
      "loss": 1.0497,
      "step": 60330
    },
    {
      "epoch": 2.4448451206418023,
      "grad_norm": 0.642357349395752,
      "learning_rate": 1.644689837952372e-05,
      "loss": 1.041,
      "step": 60340
    },
    {
      "epoch": 2.4452502988189058,
      "grad_norm": 0.7342958450317383,
      "learning_rate": 1.642357671893492e-05,
      "loss": 1.0764,
      "step": 60350
    },
    {
      "epoch": 2.445655476996009,
      "grad_norm": 0.6648983955383301,
      "learning_rate": 1.6400270125716577e-05,
      "loss": 1.0363,
      "step": 60360
    },
    {
      "epoch": 2.4460606551731123,
      "grad_norm": 0.6601636409759521,
      "learning_rate": 1.6376978604070402e-05,
      "loss": 1.0717,
      "step": 60370
    },
    {
      "epoch": 2.446465833350216,
      "grad_norm": 0.6380355358123779,
      "learning_rate": 1.6353702158195493e-05,
      "loss": 1.03,
      "step": 60380
    },
    {
      "epoch": 2.4468710115273193,
      "grad_norm": 0.6350869536399841,
      "learning_rate": 1.633044079228817e-05,
      "loss": 1.0692,
      "step": 60390
    },
    {
      "epoch": 2.4472761897044224,
      "grad_norm": 0.6687099933624268,
      "learning_rate": 1.6307194510542077e-05,
      "loss": 1.0576,
      "step": 60400
    },
    {
      "epoch": 2.447681367881526,
      "grad_norm": 0.6747086644172668,
      "learning_rate": 1.6283963317148076e-05,
      "loss": 1.0216,
      "step": 60410
    },
    {
      "epoch": 2.4480865460586294,
      "grad_norm": 0.6998465657234192,
      "learning_rate": 1.626074721629436e-05,
      "loss": 1.0234,
      "step": 60420
    },
    {
      "epoch": 2.4484917242357325,
      "grad_norm": 0.7094212174415588,
      "learning_rate": 1.623754621216641e-05,
      "loss": 1.0356,
      "step": 60430
    },
    {
      "epoch": 2.448896902412836,
      "grad_norm": 0.7915600538253784,
      "learning_rate": 1.621436030894693e-05,
      "loss": 1.0373,
      "step": 60440
    },
    {
      "epoch": 2.4493020805899395,
      "grad_norm": 0.7231775522232056,
      "learning_rate": 1.619118951081594e-05,
      "loss": 1.031,
      "step": 60450
    },
    {
      "epoch": 2.449707258767043,
      "grad_norm": 0.651868999004364,
      "learning_rate": 1.6168033821950735e-05,
      "loss": 1.074,
      "step": 60460
    },
    {
      "epoch": 2.450112436944146,
      "grad_norm": 0.6734997630119324,
      "learning_rate": 1.6144893246525906e-05,
      "loss": 1.0448,
      "step": 60470
    },
    {
      "epoch": 2.4505176151212495,
      "grad_norm": 0.6514572501182556,
      "learning_rate": 1.612176778871325e-05,
      "loss": 1.0632,
      "step": 60480
    },
    {
      "epoch": 2.450922793298353,
      "grad_norm": 0.5671477317810059,
      "learning_rate": 1.6098657452681896e-05,
      "loss": 1.0468,
      "step": 60490
    },
    {
      "epoch": 2.4513279714754566,
      "grad_norm": 0.6319550275802612,
      "learning_rate": 1.607556224259825e-05,
      "loss": 1.0254,
      "step": 60500
    },
    {
      "epoch": 2.4517331496525596,
      "grad_norm": 0.644673228263855,
      "learning_rate": 1.6052482162625992e-05,
      "loss": 1.0736,
      "step": 60510
    },
    {
      "epoch": 2.452138327829663,
      "grad_norm": 0.6371615529060364,
      "learning_rate": 1.602941721692601e-05,
      "loss": 1.0714,
      "step": 60520
    },
    {
      "epoch": 2.4525435060067666,
      "grad_norm": 0.693604052066803,
      "learning_rate": 1.6006367409656564e-05,
      "loss": 1.0405,
      "step": 60530
    },
    {
      "epoch": 2.4529486841838697,
      "grad_norm": 0.6225358843803406,
      "learning_rate": 1.5983332744973078e-05,
      "loss": 1.0558,
      "step": 60540
    },
    {
      "epoch": 2.453353862360973,
      "grad_norm": 0.6936606764793396,
      "learning_rate": 1.5960313227028322e-05,
      "loss": 1.0829,
      "step": 60550
    },
    {
      "epoch": 2.4537590405380767,
      "grad_norm": 0.6586780548095703,
      "learning_rate": 1.5937308859972323e-05,
      "loss": 1.0463,
      "step": 60560
    },
    {
      "epoch": 2.4541642187151798,
      "grad_norm": 0.722313642501831,
      "learning_rate": 1.591431964795238e-05,
      "loss": 1.0752,
      "step": 60570
    },
    {
      "epoch": 2.4545693968922833,
      "grad_norm": 0.6529853940010071,
      "learning_rate": 1.5891345595112993e-05,
      "loss": 1.0557,
      "step": 60580
    },
    {
      "epoch": 2.4549745750693868,
      "grad_norm": 0.6676023006439209,
      "learning_rate": 1.586838670559606e-05,
      "loss": 1.0466,
      "step": 60590
    },
    {
      "epoch": 2.4553797532464903,
      "grad_norm": 0.6427491903305054,
      "learning_rate": 1.584544298354059e-05,
      "loss": 1.0634,
      "step": 60600
    },
    {
      "epoch": 2.4557849314235933,
      "grad_norm": 0.6969166398048401,
      "learning_rate": 1.582251443308297e-05,
      "loss": 1.0384,
      "step": 60610
    },
    {
      "epoch": 2.456190109600697,
      "grad_norm": 0.6994515061378479,
      "learning_rate": 1.5799601058356816e-05,
      "loss": 1.0707,
      "step": 60620
    },
    {
      "epoch": 2.4565952877778003,
      "grad_norm": 0.621282696723938,
      "learning_rate": 1.5776702863493043e-05,
      "loss": 1.0524,
      "step": 60630
    },
    {
      "epoch": 2.457000465954904,
      "grad_norm": 0.6929033994674683,
      "learning_rate": 1.5753819852619733e-05,
      "loss": 1.0863,
      "step": 60640
    },
    {
      "epoch": 2.457405644132007,
      "grad_norm": 0.5506204962730408,
      "learning_rate": 1.5730952029862354e-05,
      "loss": 1.0704,
      "step": 60650
    },
    {
      "epoch": 2.4578108223091104,
      "grad_norm": 0.6588799357414246,
      "learning_rate": 1.5708099399343502e-05,
      "loss": 1.0917,
      "step": 60660
    },
    {
      "epoch": 2.458216000486214,
      "grad_norm": 0.6246358156204224,
      "learning_rate": 1.5685261965183195e-05,
      "loss": 1.0008,
      "step": 60670
    },
    {
      "epoch": 2.458621178663317,
      "grad_norm": 0.8208995461463928,
      "learning_rate": 1.5662439731498556e-05,
      "loss": 1.0489,
      "step": 60680
    },
    {
      "epoch": 2.4590263568404205,
      "grad_norm": 0.7315219640731812,
      "learning_rate": 1.5639632702404083e-05,
      "loss": 1.0066,
      "step": 60690
    },
    {
      "epoch": 2.459431535017524,
      "grad_norm": 0.693507730960846,
      "learning_rate": 1.5616840882011443e-05,
      "loss": 1.0886,
      "step": 60700
    },
    {
      "epoch": 2.4598367131946275,
      "grad_norm": 0.7399535775184631,
      "learning_rate": 1.559406427442962e-05,
      "loss": 1.0697,
      "step": 60710
    },
    {
      "epoch": 2.4602418913717305,
      "grad_norm": 0.6469250917434692,
      "learning_rate": 1.557130288376484e-05,
      "loss": 1.0347,
      "step": 60720
    },
    {
      "epoch": 2.460647069548834,
      "grad_norm": 0.6677451133728027,
      "learning_rate": 1.554855671412061e-05,
      "loss": 1.0772,
      "step": 60730
    },
    {
      "epoch": 2.4610522477259376,
      "grad_norm": 0.6674231886863708,
      "learning_rate": 1.5525825769597624e-05,
      "loss": 1.0484,
      "step": 60740
    },
    {
      "epoch": 2.461457425903041,
      "grad_norm": 0.6711300611495972,
      "learning_rate": 1.5503110054293892e-05,
      "loss": 1.048,
      "step": 60750
    },
    {
      "epoch": 2.461862604080144,
      "grad_norm": 0.6466948986053467,
      "learning_rate": 1.5480409572304678e-05,
      "loss": 1.0566,
      "step": 60760
    },
    {
      "epoch": 2.4622677822572476,
      "grad_norm": 0.7562391757965088,
      "learning_rate": 1.5457724327722457e-05,
      "loss": 1.0229,
      "step": 60770
    },
    {
      "epoch": 2.462672960434351,
      "grad_norm": 0.6648483276367188,
      "learning_rate": 1.5435054324636976e-05,
      "loss": 1.0244,
      "step": 60780
    },
    {
      "epoch": 2.463078138611454,
      "grad_norm": 0.6634892821311951,
      "learning_rate": 1.5412399567135273e-05,
      "loss": 1.0289,
      "step": 60790
    },
    {
      "epoch": 2.4634833167885577,
      "grad_norm": 0.6147224307060242,
      "learning_rate": 1.5389760059301617e-05,
      "loss": 1.0545,
      "step": 60800
    },
    {
      "epoch": 2.463888494965661,
      "grad_norm": 0.8517619967460632,
      "learning_rate": 1.5367135805217458e-05,
      "loss": 1.0472,
      "step": 60810
    },
    {
      "epoch": 2.4642936731427647,
      "grad_norm": 0.6185300350189209,
      "learning_rate": 1.534452680896161e-05,
      "loss": 1.028,
      "step": 60820
    },
    {
      "epoch": 2.4646988513198678,
      "grad_norm": 0.6884375810623169,
      "learning_rate": 1.5321933074610006e-05,
      "loss": 1.0544,
      "step": 60830
    },
    {
      "epoch": 2.4651040294969713,
      "grad_norm": 0.5888127088546753,
      "learning_rate": 1.5299354606236004e-05,
      "loss": 1.0372,
      "step": 60840
    },
    {
      "epoch": 2.4655092076740748,
      "grad_norm": 0.5982387661933899,
      "learning_rate": 1.527679140791003e-05,
      "loss": 1.0355,
      "step": 60850
    },
    {
      "epoch": 2.4659143858511783,
      "grad_norm": 0.6076526045799255,
      "learning_rate": 1.5254243483699892e-05,
      "loss": 1.0706,
      "step": 60860
    },
    {
      "epoch": 2.4663195640282813,
      "grad_norm": 0.656713604927063,
      "learning_rate": 1.523171083767052e-05,
      "loss": 1.0791,
      "step": 60870
    },
    {
      "epoch": 2.466724742205385,
      "grad_norm": 0.663959801197052,
      "learning_rate": 1.5209193473884232e-05,
      "loss": 1.0387,
      "step": 60880
    },
    {
      "epoch": 2.4671299203824884,
      "grad_norm": 0.6185634136199951,
      "learning_rate": 1.518669139640041e-05,
      "loss": 1.0259,
      "step": 60890
    },
    {
      "epoch": 2.4675350985595914,
      "grad_norm": 0.7060873508453369,
      "learning_rate": 1.5164204609275912e-05,
      "loss": 1.0518,
      "step": 60900
    },
    {
      "epoch": 2.467940276736695,
      "grad_norm": 0.7138689160346985,
      "learning_rate": 1.5141733116564627e-05,
      "loss": 1.0612,
      "step": 60910
    },
    {
      "epoch": 2.4683454549137984,
      "grad_norm": 0.5810047388076782,
      "learning_rate": 1.5119276922317827e-05,
      "loss": 1.0608,
      "step": 60920
    },
    {
      "epoch": 2.4687506330909015,
      "grad_norm": 0.6215168833732605,
      "learning_rate": 1.5096836030583916e-05,
      "loss": 1.0399,
      "step": 60930
    },
    {
      "epoch": 2.469155811268005,
      "grad_norm": 0.5580259561538696,
      "learning_rate": 1.5074410445408626e-05,
      "loss": 1.0703,
      "step": 60940
    },
    {
      "epoch": 2.4695609894451085,
      "grad_norm": 0.7175332307815552,
      "learning_rate": 1.5052000170834901e-05,
      "loss": 1.0658,
      "step": 60950
    },
    {
      "epoch": 2.469966167622212,
      "grad_norm": 0.6258244514465332,
      "learning_rate": 1.5029605210902931e-05,
      "loss": 1.0523,
      "step": 60960
    },
    {
      "epoch": 2.470371345799315,
      "grad_norm": 0.51904296875,
      "learning_rate": 1.5007225569650107e-05,
      "loss": 1.0413,
      "step": 60970
    },
    {
      "epoch": 2.4707765239764186,
      "grad_norm": 0.666019082069397,
      "learning_rate": 1.4984861251111126e-05,
      "loss": 1.046,
      "step": 60980
    },
    {
      "epoch": 2.471181702153522,
      "grad_norm": 0.634087324142456,
      "learning_rate": 1.496251225931783e-05,
      "loss": 1.034,
      "step": 60990
    },
    {
      "epoch": 2.4715868803306256,
      "grad_norm": 0.6459301114082336,
      "learning_rate": 1.4940178598299392e-05,
      "loss": 1.0504,
      "step": 61000
    },
    {
      "epoch": 2.4719920585077286,
      "grad_norm": 0.5848231315612793,
      "learning_rate": 1.4917860272082174e-05,
      "loss": 1.0228,
      "step": 61010
    },
    {
      "epoch": 2.472397236684832,
      "grad_norm": 0.59383225440979,
      "learning_rate": 1.4895557284689799e-05,
      "loss": 1.0608,
      "step": 61020
    },
    {
      "epoch": 2.4728024148619356,
      "grad_norm": 0.6304118633270264,
      "learning_rate": 1.4873269640143062e-05,
      "loss": 1.0636,
      "step": 61030
    },
    {
      "epoch": 2.4732075930390387,
      "grad_norm": 0.5940285325050354,
      "learning_rate": 1.4850997342460094e-05,
      "loss": 1.0213,
      "step": 61040
    },
    {
      "epoch": 2.473612771216142,
      "grad_norm": 0.6594524383544922,
      "learning_rate": 1.4828740395656137e-05,
      "loss": 1.0423,
      "step": 61050
    },
    {
      "epoch": 2.4740179493932457,
      "grad_norm": 0.7412016987800598,
      "learning_rate": 1.4806498803743774e-05,
      "loss": 1.0634,
      "step": 61060
    },
    {
      "epoch": 2.474423127570349,
      "grad_norm": 0.5877138376235962,
      "learning_rate": 1.478427257073276e-05,
      "loss": 1.0385,
      "step": 61070
    },
    {
      "epoch": 2.4748283057474523,
      "grad_norm": 0.6860798001289368,
      "learning_rate": 1.4762061700630115e-05,
      "loss": 1.0558,
      "step": 61080
    },
    {
      "epoch": 2.4752334839245558,
      "grad_norm": 0.6328216791152954,
      "learning_rate": 1.4739866197440044e-05,
      "loss": 1.0298,
      "step": 61090
    },
    {
      "epoch": 2.4756386621016593,
      "grad_norm": 0.7107235193252563,
      "learning_rate": 1.4717686065164016e-05,
      "loss": 1.0523,
      "step": 61100
    },
    {
      "epoch": 2.476043840278763,
      "grad_norm": 0.5914685130119324,
      "learning_rate": 1.469552130780073e-05,
      "loss": 1.0255,
      "step": 61110
    },
    {
      "epoch": 2.476449018455866,
      "grad_norm": 0.7013688087463379,
      "learning_rate": 1.4673371929346125e-05,
      "loss": 1.0128,
      "step": 61120
    },
    {
      "epoch": 2.4768541966329694,
      "grad_norm": 0.7178220748901367,
      "learning_rate": 1.4651237933793305e-05,
      "loss": 1.0364,
      "step": 61130
    },
    {
      "epoch": 2.477259374810073,
      "grad_norm": 0.7215100526809692,
      "learning_rate": 1.4629119325132657e-05,
      "loss": 1.0426,
      "step": 61140
    },
    {
      "epoch": 2.477664552987176,
      "grad_norm": 0.6187329888343811,
      "learning_rate": 1.4607016107351824e-05,
      "loss": 1.0438,
      "step": 61150
    },
    {
      "epoch": 2.4780697311642794,
      "grad_norm": 0.6190857887268066,
      "learning_rate": 1.458492828443555e-05,
      "loss": 1.1233,
      "step": 61160
    },
    {
      "epoch": 2.478474909341383,
      "grad_norm": 0.6592904925346375,
      "learning_rate": 1.4562855860365943e-05,
      "loss": 1.0624,
      "step": 61170
    },
    {
      "epoch": 2.4788800875184864,
      "grad_norm": 0.5857892036437988,
      "learning_rate": 1.454079883912225e-05,
      "loss": 1.0474,
      "step": 61180
    },
    {
      "epoch": 2.4792852656955895,
      "grad_norm": 0.6507061719894409,
      "learning_rate": 1.4518757224681001e-05,
      "loss": 1.0262,
      "step": 61190
    },
    {
      "epoch": 2.479690443872693,
      "grad_norm": 0.6098853945732117,
      "learning_rate": 1.4496731021015864e-05,
      "loss": 1.0356,
      "step": 61200
    },
    {
      "epoch": 2.4800956220497965,
      "grad_norm": 0.6730479001998901,
      "learning_rate": 1.4474720232097838e-05,
      "loss": 1.048,
      "step": 61210
    },
    {
      "epoch": 2.4805008002269,
      "grad_norm": 0.7337656021118164,
      "learning_rate": 1.4452724861894994e-05,
      "loss": 1.0268,
      "step": 61220
    },
    {
      "epoch": 2.480905978404003,
      "grad_norm": 0.6185232996940613,
      "learning_rate": 1.443074491437283e-05,
      "loss": 1.0533,
      "step": 61230
    },
    {
      "epoch": 2.4813111565811066,
      "grad_norm": 0.6603759527206421,
      "learning_rate": 1.4408780393493849e-05,
      "loss": 1.0402,
      "step": 61240
    },
    {
      "epoch": 2.48171633475821,
      "grad_norm": 0.6457160711288452,
      "learning_rate": 1.438683130321794e-05,
      "loss": 1.0413,
      "step": 61250
    },
    {
      "epoch": 2.482121512935313,
      "grad_norm": 0.6620184183120728,
      "learning_rate": 1.4364897647502085e-05,
      "loss": 1.0337,
      "step": 61260
    },
    {
      "epoch": 2.4825266911124166,
      "grad_norm": 0.7049242258071899,
      "learning_rate": 1.4342979430300585e-05,
      "loss": 1.0519,
      "step": 61270
    },
    {
      "epoch": 2.48293186928952,
      "grad_norm": 0.7185577750205994,
      "learning_rate": 1.4321076655564846e-05,
      "loss": 1.0743,
      "step": 61280
    },
    {
      "epoch": 2.483337047466623,
      "grad_norm": 0.6402226090431213,
      "learning_rate": 1.4299189327243656e-05,
      "loss": 1.0586,
      "step": 61290
    },
    {
      "epoch": 2.4837422256437267,
      "grad_norm": 0.6443219780921936,
      "learning_rate": 1.4277317449282834e-05,
      "loss": 1.067,
      "step": 61300
    },
    {
      "epoch": 2.48414740382083,
      "grad_norm": 0.7008201479911804,
      "learning_rate": 1.425546102562555e-05,
      "loss": 1.0596,
      "step": 61310
    },
    {
      "epoch": 2.4845525819979337,
      "grad_norm": 0.6514986157417297,
      "learning_rate": 1.4233620060212083e-05,
      "loss": 1.0158,
      "step": 61320
    },
    {
      "epoch": 2.484957760175037,
      "grad_norm": 0.5854817628860474,
      "learning_rate": 1.421179455698004e-05,
      "loss": 1.0518,
      "step": 61330
    },
    {
      "epoch": 2.4853629383521403,
      "grad_norm": 0.6582303047180176,
      "learning_rate": 1.4189984519864097e-05,
      "loss": 1.0044,
      "step": 61340
    },
    {
      "epoch": 2.485768116529244,
      "grad_norm": 0.6829269528388977,
      "learning_rate": 1.4168189952796318e-05,
      "loss": 1.0554,
      "step": 61350
    },
    {
      "epoch": 2.4861732947063473,
      "grad_norm": 0.6435067057609558,
      "learning_rate": 1.4146410859705817e-05,
      "loss": 1.0847,
      "step": 61360
    },
    {
      "epoch": 2.4865784728834504,
      "grad_norm": 0.7191896438598633,
      "learning_rate": 1.4124647244519029e-05,
      "loss": 1.0202,
      "step": 61370
    },
    {
      "epoch": 2.486983651060554,
      "grad_norm": 0.709791898727417,
      "learning_rate": 1.4102899111159507e-05,
      "loss": 1.0263,
      "step": 61380
    },
    {
      "epoch": 2.4873888292376574,
      "grad_norm": 0.6120791435241699,
      "learning_rate": 1.408116646354808e-05,
      "loss": 1.0413,
      "step": 61390
    },
    {
      "epoch": 2.4877940074147604,
      "grad_norm": 0.6652215719223022,
      "learning_rate": 1.4059449305602767e-05,
      "loss": 1.0125,
      "step": 61400
    },
    {
      "epoch": 2.488199185591864,
      "grad_norm": 0.6380071640014648,
      "learning_rate": 1.4037747641238818e-05,
      "loss": 1.0735,
      "step": 61410
    },
    {
      "epoch": 2.4886043637689674,
      "grad_norm": 0.6762730479240417,
      "learning_rate": 1.4016061474368613e-05,
      "loss": 1.0623,
      "step": 61420
    },
    {
      "epoch": 2.489009541946071,
      "grad_norm": 0.639022707939148,
      "learning_rate": 1.3994390808901815e-05,
      "loss": 1.0209,
      "step": 61430
    },
    {
      "epoch": 2.489414720123174,
      "grad_norm": 0.6502436995506287,
      "learning_rate": 1.3972735648745294e-05,
      "loss": 1.0683,
      "step": 61440
    },
    {
      "epoch": 2.4898198983002775,
      "grad_norm": 0.6319053769111633,
      "learning_rate": 1.3951095997803054e-05,
      "loss": 1.0185,
      "step": 61450
    },
    {
      "epoch": 2.490225076477381,
      "grad_norm": 0.6188423037528992,
      "learning_rate": 1.3929471859976351e-05,
      "loss": 1.0642,
      "step": 61460
    },
    {
      "epoch": 2.4906302546544845,
      "grad_norm": 0.6963823437690735,
      "learning_rate": 1.3907863239163666e-05,
      "loss": 1.0496,
      "step": 61470
    },
    {
      "epoch": 2.4910354328315876,
      "grad_norm": 0.6357050538063049,
      "learning_rate": 1.3886270139260649e-05,
      "loss": 1.0269,
      "step": 61480
    },
    {
      "epoch": 2.491440611008691,
      "grad_norm": 0.6979972124099731,
      "learning_rate": 1.3864692564160143e-05,
      "loss": 1.0846,
      "step": 61490
    },
    {
      "epoch": 2.4918457891857946,
      "grad_norm": 0.6899288296699524,
      "learning_rate": 1.3843130517752235e-05,
      "loss": 1.0322,
      "step": 61500
    },
    {
      "epoch": 2.4922509673628976,
      "grad_norm": 0.5828148126602173,
      "learning_rate": 1.3821584003924127e-05,
      "loss": 1.0411,
      "step": 61510
    },
    {
      "epoch": 2.492656145540001,
      "grad_norm": 0.57685786485672,
      "learning_rate": 1.3800053026560356e-05,
      "loss": 1.0282,
      "step": 61520
    },
    {
      "epoch": 2.4930613237171046,
      "grad_norm": 0.5957061648368835,
      "learning_rate": 1.3778537589542529e-05,
      "loss": 1.0495,
      "step": 61530
    },
    {
      "epoch": 2.493466501894208,
      "grad_norm": 0.7835922837257385,
      "learning_rate": 1.3757037696749541e-05,
      "loss": 1.0485,
      "step": 61540
    },
    {
      "epoch": 2.493871680071311,
      "grad_norm": 0.6438522338867188,
      "learning_rate": 1.3735553352057407e-05,
      "loss": 1.0778,
      "step": 61550
    },
    {
      "epoch": 2.4942768582484147,
      "grad_norm": 0.6191057562828064,
      "learning_rate": 1.3714084559339402e-05,
      "loss": 1.0499,
      "step": 61560
    },
    {
      "epoch": 2.4946820364255182,
      "grad_norm": 0.7150356769561768,
      "learning_rate": 1.3692631322465976e-05,
      "loss": 1.051,
      "step": 61570
    },
    {
      "epoch": 2.4950872146026217,
      "grad_norm": 0.6939385533332825,
      "learning_rate": 1.3671193645304781e-05,
      "loss": 1.0411,
      "step": 61580
    },
    {
      "epoch": 2.495492392779725,
      "grad_norm": 0.6938865184783936,
      "learning_rate": 1.3649771531720623e-05,
      "loss": 1.0351,
      "step": 61590
    },
    {
      "epoch": 2.4958975709568283,
      "grad_norm": 0.7250855565071106,
      "learning_rate": 1.3628364985575582e-05,
      "loss": 1.0246,
      "step": 61600
    },
    {
      "epoch": 2.496302749133932,
      "grad_norm": 0.6673681735992432,
      "learning_rate": 1.3606974010728835e-05,
      "loss": 1.0492,
      "step": 61610
    },
    {
      "epoch": 2.496707927311035,
      "grad_norm": 0.6511656045913696,
      "learning_rate": 1.358559861103683e-05,
      "loss": 1.0528,
      "step": 61620
    },
    {
      "epoch": 2.4971131054881384,
      "grad_norm": 0.7362741827964783,
      "learning_rate": 1.3564238790353179e-05,
      "loss": 1.0574,
      "step": 61630
    },
    {
      "epoch": 2.497518283665242,
      "grad_norm": 0.6365575194358826,
      "learning_rate": 1.3542894552528696e-05,
      "loss": 1.0642,
      "step": 61640
    },
    {
      "epoch": 2.497923461842345,
      "grad_norm": 0.7268562912940979,
      "learning_rate": 1.3521565901411326e-05,
      "loss": 1.0458,
      "step": 61650
    },
    {
      "epoch": 2.4983286400194484,
      "grad_norm": 0.6559151411056519,
      "learning_rate": 1.3500252840846317e-05,
      "loss": 1.0656,
      "step": 61660
    },
    {
      "epoch": 2.498733818196552,
      "grad_norm": 0.6674041748046875,
      "learning_rate": 1.3478955374675994e-05,
      "loss": 1.0302,
      "step": 61670
    },
    {
      "epoch": 2.4991389963736554,
      "grad_norm": 0.7046427726745605,
      "learning_rate": 1.345767350673992e-05,
      "loss": 1.0691,
      "step": 61680
    },
    {
      "epoch": 2.4995441745507585,
      "grad_norm": 0.5789068937301636,
      "learning_rate": 1.343640724087487e-05,
      "loss": 1.0768,
      "step": 61690
    },
    {
      "epoch": 2.499949352727862,
      "grad_norm": 0.7282841205596924,
      "learning_rate": 1.341515658091479e-05,
      "loss": 1.035,
      "step": 61700
    },
    {
      "epoch": 2.5003545309049655,
      "grad_norm": 0.6825752258300781,
      "learning_rate": 1.3393921530690756e-05,
      "loss": 1.1065,
      "step": 61710
    },
    {
      "epoch": 2.500759709082069,
      "grad_norm": 0.6858643293380737,
      "learning_rate": 1.3372702094031108e-05,
      "loss": 1.0301,
      "step": 61720
    },
    {
      "epoch": 2.501164887259172,
      "grad_norm": 0.6341392397880554,
      "learning_rate": 1.3351498274761332e-05,
      "loss": 1.086,
      "step": 61730
    },
    {
      "epoch": 2.5015700654362756,
      "grad_norm": 0.6368936896324158,
      "learning_rate": 1.3330310076704133e-05,
      "loss": 1.0317,
      "step": 61740
    },
    {
      "epoch": 2.501975243613379,
      "grad_norm": 0.5928543210029602,
      "learning_rate": 1.3309137503679325e-05,
      "loss": 1.0291,
      "step": 61750
    },
    {
      "epoch": 2.502380421790482,
      "grad_norm": 0.6084482073783875,
      "learning_rate": 1.3287980559503987e-05,
      "loss": 1.0612,
      "step": 61760
    },
    {
      "epoch": 2.5027855999675857,
      "grad_norm": 0.6444761157035828,
      "learning_rate": 1.3266839247992346e-05,
      "loss": 1.051,
      "step": 61770
    },
    {
      "epoch": 2.503190778144689,
      "grad_norm": 0.6693594455718994,
      "learning_rate": 1.3245713572955798e-05,
      "loss": 1.0516,
      "step": 61780
    },
    {
      "epoch": 2.5035959563217927,
      "grad_norm": 0.5518051981925964,
      "learning_rate": 1.3224603538202929e-05,
      "loss": 1.036,
      "step": 61790
    },
    {
      "epoch": 2.504001134498896,
      "grad_norm": 0.6228415369987488,
      "learning_rate": 1.3203509147539527e-05,
      "loss": 1.0161,
      "step": 61800
    },
    {
      "epoch": 2.5044063126759992,
      "grad_norm": 0.680917501449585,
      "learning_rate": 1.3182430404768553e-05,
      "loss": 1.0686,
      "step": 61810
    },
    {
      "epoch": 2.5048114908531027,
      "grad_norm": 0.7012894153594971,
      "learning_rate": 1.3161367313690099e-05,
      "loss": 1.0189,
      "step": 61820
    },
    {
      "epoch": 2.5052166690302062,
      "grad_norm": 0.6442646980285645,
      "learning_rate": 1.3140319878101515e-05,
      "loss": 1.0354,
      "step": 61830
    },
    {
      "epoch": 2.5056218472073093,
      "grad_norm": 0.6483091115951538,
      "learning_rate": 1.3119288101797211e-05,
      "loss": 1.035,
      "step": 61840
    },
    {
      "epoch": 2.506027025384413,
      "grad_norm": 0.6168128848075867,
      "learning_rate": 1.3098271988568955e-05,
      "loss": 1.0397,
      "step": 61850
    },
    {
      "epoch": 2.5064322035615163,
      "grad_norm": 0.6936774253845215,
      "learning_rate": 1.3077271542205515e-05,
      "loss": 1.0267,
      "step": 61860
    },
    {
      "epoch": 2.5068373817386194,
      "grad_norm": 0.7322385311126709,
      "learning_rate": 1.3056286766492932e-05,
      "loss": 1.0895,
      "step": 61870
    },
    {
      "epoch": 2.507242559915723,
      "grad_norm": 0.6403207778930664,
      "learning_rate": 1.3035317665214374e-05,
      "loss": 1.0912,
      "step": 61880
    },
    {
      "epoch": 2.5076477380928264,
      "grad_norm": 0.5705447793006897,
      "learning_rate": 1.3014364242150235e-05,
      "loss": 1.0386,
      "step": 61890
    },
    {
      "epoch": 2.5080529162699294,
      "grad_norm": 0.60397869348526,
      "learning_rate": 1.2993426501077987e-05,
      "loss": 1.0576,
      "step": 61900
    },
    {
      "epoch": 2.508458094447033,
      "grad_norm": 0.7136611938476562,
      "learning_rate": 1.2972504445772426e-05,
      "loss": 1.0084,
      "step": 61910
    },
    {
      "epoch": 2.5088632726241364,
      "grad_norm": 0.7953967452049255,
      "learning_rate": 1.2951598080005378e-05,
      "loss": 1.0802,
      "step": 61920
    },
    {
      "epoch": 2.50926845080124,
      "grad_norm": 0.6959277391433716,
      "learning_rate": 1.2930707407545917e-05,
      "loss": 1.0487,
      "step": 61930
    },
    {
      "epoch": 2.5096736289783435,
      "grad_norm": 0.6529216170310974,
      "learning_rate": 1.2909832432160252e-05,
      "loss": 1.0679,
      "step": 61940
    },
    {
      "epoch": 2.5100788071554465,
      "grad_norm": 0.6839214563369751,
      "learning_rate": 1.2888973157611794e-05,
      "loss": 1.0548,
      "step": 61950
    },
    {
      "epoch": 2.51048398533255,
      "grad_norm": 0.6548508405685425,
      "learning_rate": 1.286812958766106e-05,
      "loss": 1.0628,
      "step": 61960
    },
    {
      "epoch": 2.5108891635096535,
      "grad_norm": 0.7983167171478271,
      "learning_rate": 1.2847301726065852e-05,
      "loss": 1.0267,
      "step": 61970
    },
    {
      "epoch": 2.5112943416867566,
      "grad_norm": 0.6706383228302002,
      "learning_rate": 1.2826489576581024e-05,
      "loss": 1.0452,
      "step": 61980
    },
    {
      "epoch": 2.51169951986386,
      "grad_norm": 0.7226294279098511,
      "learning_rate": 1.2805693142958674e-05,
      "loss": 1.0265,
      "step": 61990
    },
    {
      "epoch": 2.5121046980409636,
      "grad_norm": 0.6772617697715759,
      "learning_rate": 1.2784912428947992e-05,
      "loss": 1.0481,
      "step": 62000
    },
    {
      "epoch": 2.5125098762180667,
      "grad_norm": 0.6398630142211914,
      "learning_rate": 1.2764147438295403e-05,
      "loss": 1.009,
      "step": 62010
    },
    {
      "epoch": 2.51291505439517,
      "grad_norm": 0.6157912611961365,
      "learning_rate": 1.2743398174744481e-05,
      "loss": 1.0422,
      "step": 62020
    },
    {
      "epoch": 2.5133202325722737,
      "grad_norm": 0.776006281375885,
      "learning_rate": 1.2722664642035964e-05,
      "loss": 1.0628,
      "step": 62030
    },
    {
      "epoch": 2.513725410749377,
      "grad_norm": 0.6272456049919128,
      "learning_rate": 1.2701946843907708e-05,
      "loss": 1.0764,
      "step": 62040
    },
    {
      "epoch": 2.5141305889264807,
      "grad_norm": 0.5541890263557434,
      "learning_rate": 1.2681244784094815e-05,
      "loss": 1.0341,
      "step": 62050
    },
    {
      "epoch": 2.5145357671035837,
      "grad_norm": 0.6391535997390747,
      "learning_rate": 1.2660558466329453e-05,
      "loss": 1.0268,
      "step": 62060
    },
    {
      "epoch": 2.5149409452806872,
      "grad_norm": 0.6593660116195679,
      "learning_rate": 1.263988789434104e-05,
      "loss": 1.0916,
      "step": 62070
    },
    {
      "epoch": 2.5153461234577907,
      "grad_norm": 0.7259678840637207,
      "learning_rate": 1.2619233071856107e-05,
      "loss": 1.0593,
      "step": 62080
    },
    {
      "epoch": 2.515751301634894,
      "grad_norm": 0.7168288230895996,
      "learning_rate": 1.2598594002598396e-05,
      "loss": 1.07,
      "step": 62090
    },
    {
      "epoch": 2.5161564798119973,
      "grad_norm": 0.6792853474617004,
      "learning_rate": 1.257797069028871e-05,
      "loss": 1.0328,
      "step": 62100
    },
    {
      "epoch": 2.516561657989101,
      "grad_norm": 0.6541315317153931,
      "learning_rate": 1.2557363138645095e-05,
      "loss": 1.0121,
      "step": 62110
    },
    {
      "epoch": 2.516966836166204,
      "grad_norm": 0.6771706938743591,
      "learning_rate": 1.253677135138277e-05,
      "loss": 1.0969,
      "step": 62120
    },
    {
      "epoch": 2.5173720143433074,
      "grad_norm": 0.6614355444908142,
      "learning_rate": 1.251619533221402e-05,
      "loss": 1.0107,
      "step": 62130
    },
    {
      "epoch": 2.517777192520411,
      "grad_norm": 0.6040178537368774,
      "learning_rate": 1.2495635084848356e-05,
      "loss": 1.0496,
      "step": 62140
    },
    {
      "epoch": 2.5181823706975144,
      "grad_norm": 0.6817635893821716,
      "learning_rate": 1.2475090612992446e-05,
      "loss": 1.0446,
      "step": 62150
    },
    {
      "epoch": 2.518587548874618,
      "grad_norm": 0.7022846341133118,
      "learning_rate": 1.2454561920350128e-05,
      "loss": 1.0182,
      "step": 62160
    },
    {
      "epoch": 2.518992727051721,
      "grad_norm": 0.7933214902877808,
      "learning_rate": 1.2434049010622295e-05,
      "loss": 1.0469,
      "step": 62170
    },
    {
      "epoch": 2.5193979052288245,
      "grad_norm": 0.6394073963165283,
      "learning_rate": 1.2413551887507124e-05,
      "loss": 1.0393,
      "step": 62180
    },
    {
      "epoch": 2.519803083405928,
      "grad_norm": 0.6491120457649231,
      "learning_rate": 1.2393070554699859e-05,
      "loss": 1.0615,
      "step": 62190
    },
    {
      "epoch": 2.520208261583031,
      "grad_norm": 0.6958550214767456,
      "learning_rate": 1.2372605015892958e-05,
      "loss": 1.0519,
      "step": 62200
    },
    {
      "epoch": 2.5206134397601345,
      "grad_norm": 0.631353497505188,
      "learning_rate": 1.2352155274775967e-05,
      "loss": 1.0133,
      "step": 62210
    },
    {
      "epoch": 2.521018617937238,
      "grad_norm": 0.6332811117172241,
      "learning_rate": 1.2331721335035639e-05,
      "loss": 1.0788,
      "step": 62220
    },
    {
      "epoch": 2.521423796114341,
      "grad_norm": 0.6609264016151428,
      "learning_rate": 1.2311303200355838e-05,
      "loss": 1.1079,
      "step": 62230
    },
    {
      "epoch": 2.5218289742914446,
      "grad_norm": 0.6767718195915222,
      "learning_rate": 1.229090087441761e-05,
      "loss": 1.0347,
      "step": 62240
    },
    {
      "epoch": 2.522234152468548,
      "grad_norm": 0.6615316271781921,
      "learning_rate": 1.2270514360899133e-05,
      "loss": 1.1022,
      "step": 62250
    },
    {
      "epoch": 2.522639330645651,
      "grad_norm": 0.658566951751709,
      "learning_rate": 1.2250143663475765e-05,
      "loss": 1.0584,
      "step": 62260
    },
    {
      "epoch": 2.5230445088227547,
      "grad_norm": 0.59205561876297,
      "learning_rate": 1.222978878581994e-05,
      "loss": 1.0388,
      "step": 62270
    },
    {
      "epoch": 2.523449686999858,
      "grad_norm": 0.6718086004257202,
      "learning_rate": 1.220944973160133e-05,
      "loss": 1.0288,
      "step": 62280
    },
    {
      "epoch": 2.5238548651769617,
      "grad_norm": 0.6535348892211914,
      "learning_rate": 1.2189126504486682e-05,
      "loss": 1.053,
      "step": 62290
    },
    {
      "epoch": 2.524260043354065,
      "grad_norm": 0.6352729201316833,
      "learning_rate": 1.2168819108139929e-05,
      "loss": 1.0508,
      "step": 62300
    },
    {
      "epoch": 2.5246652215311682,
      "grad_norm": 0.7574026584625244,
      "learning_rate": 1.2148527546222133e-05,
      "loss": 1.0753,
      "step": 62310
    },
    {
      "epoch": 2.5250703997082717,
      "grad_norm": 0.6643089652061462,
      "learning_rate": 1.2128251822391546e-05,
      "loss": 1.0348,
      "step": 62320
    },
    {
      "epoch": 2.5254755778853752,
      "grad_norm": 0.6315234303474426,
      "learning_rate": 1.2107991940303476e-05,
      "loss": 1.0427,
      "step": 62330
    },
    {
      "epoch": 2.5258807560624783,
      "grad_norm": 0.6793181896209717,
      "learning_rate": 1.2087747903610469e-05,
      "loss": 1.0335,
      "step": 62340
    },
    {
      "epoch": 2.526285934239582,
      "grad_norm": 0.5886202454566956,
      "learning_rate": 1.2067519715962116e-05,
      "loss": 1.0496,
      "step": 62350
    },
    {
      "epoch": 2.5266911124166853,
      "grad_norm": 0.7031931281089783,
      "learning_rate": 1.2047307381005279e-05,
      "loss": 1.0916,
      "step": 62360
    },
    {
      "epoch": 2.5270962905937884,
      "grad_norm": 0.7037198543548584,
      "learning_rate": 1.202711090238382e-05,
      "loss": 1.0232,
      "step": 62370
    },
    {
      "epoch": 2.527501468770892,
      "grad_norm": 0.7104637026786804,
      "learning_rate": 1.2006930283738882e-05,
      "loss": 1.099,
      "step": 62380
    },
    {
      "epoch": 2.5279066469479954,
      "grad_norm": 0.6663996577262878,
      "learning_rate": 1.198676552870861e-05,
      "loss": 1.052,
      "step": 62390
    },
    {
      "epoch": 2.528311825125099,
      "grad_norm": 0.6384869813919067,
      "learning_rate": 1.1966616640928385e-05,
      "loss": 1.0685,
      "step": 62400
    },
    {
      "epoch": 2.5287170033022024,
      "grad_norm": 0.6742812395095825,
      "learning_rate": 1.1946483624030702e-05,
      "loss": 1.0565,
      "step": 62410
    },
    {
      "epoch": 2.5291221814793055,
      "grad_norm": 0.7073874473571777,
      "learning_rate": 1.1926366481645213e-05,
      "loss": 1.0015,
      "step": 62420
    },
    {
      "epoch": 2.529527359656409,
      "grad_norm": 0.587647020816803,
      "learning_rate": 1.1906265217398649e-05,
      "loss": 1.0197,
      "step": 62430
    },
    {
      "epoch": 2.5299325378335125,
      "grad_norm": 0.7354785799980164,
      "learning_rate": 1.1886179834914934e-05,
      "loss": 1.0782,
      "step": 62440
    },
    {
      "epoch": 2.5303377160106155,
      "grad_norm": 0.6905028223991394,
      "learning_rate": 1.1866110337815129e-05,
      "loss": 1.0304,
      "step": 62450
    },
    {
      "epoch": 2.530742894187719,
      "grad_norm": 0.6644706726074219,
      "learning_rate": 1.1846056729717381e-05,
      "loss": 1.0458,
      "step": 62460
    },
    {
      "epoch": 2.5311480723648225,
      "grad_norm": 0.6274743676185608,
      "learning_rate": 1.1826019014237022e-05,
      "loss": 1.0661,
      "step": 62470
    },
    {
      "epoch": 2.5315532505419256,
      "grad_norm": 0.5376157760620117,
      "learning_rate": 1.1805997194986507e-05,
      "loss": 1.0749,
      "step": 62480
    },
    {
      "epoch": 2.531958428719029,
      "grad_norm": 0.6396746635437012,
      "learning_rate": 1.1785991275575426e-05,
      "loss": 1.0445,
      "step": 62490
    },
    {
      "epoch": 2.5323636068961326,
      "grad_norm": 0.5994840264320374,
      "learning_rate": 1.1766001259610482e-05,
      "loss": 1.0426,
      "step": 62500
    },
    {
      "epoch": 2.532768785073236,
      "grad_norm": 0.652032196521759,
      "learning_rate": 1.1746027150695538e-05,
      "loss": 1.0684,
      "step": 62510
    },
    {
      "epoch": 2.5331739632503396,
      "grad_norm": 0.638837993144989,
      "learning_rate": 1.1726068952431536e-05,
      "loss": 1.0833,
      "step": 62520
    },
    {
      "epoch": 2.5335791414274427,
      "grad_norm": 0.7287196516990662,
      "learning_rate": 1.1706126668416673e-05,
      "loss": 1.0555,
      "step": 62530
    },
    {
      "epoch": 2.533984319604546,
      "grad_norm": 0.7407642602920532,
      "learning_rate": 1.168620030224612e-05,
      "loss": 1.0895,
      "step": 62540
    },
    {
      "epoch": 2.5343894977816497,
      "grad_norm": 0.6686701774597168,
      "learning_rate": 1.1666289857512302e-05,
      "loss": 1.0472,
      "step": 62550
    },
    {
      "epoch": 2.5347946759587527,
      "grad_norm": 0.6867257952690125,
      "learning_rate": 1.1646395337804683e-05,
      "loss": 1.0337,
      "step": 62560
    },
    {
      "epoch": 2.5351998541358562,
      "grad_norm": 0.7794269919395447,
      "learning_rate": 1.1626516746709936e-05,
      "loss": 1.0458,
      "step": 62570
    },
    {
      "epoch": 2.5356050323129598,
      "grad_norm": 0.6391009092330933,
      "learning_rate": 1.1606654087811764e-05,
      "loss": 1.0299,
      "step": 62580
    },
    {
      "epoch": 2.536010210490063,
      "grad_norm": 0.6790600419044495,
      "learning_rate": 1.1586807364691143e-05,
      "loss": 1.089,
      "step": 62590
    },
    {
      "epoch": 2.5364153886671663,
      "grad_norm": 0.6624394655227661,
      "learning_rate": 1.1566976580926026e-05,
      "loss": 1.0678,
      "step": 62600
    },
    {
      "epoch": 2.53682056684427,
      "grad_norm": 0.6111233234405518,
      "learning_rate": 1.1547161740091594e-05,
      "loss": 1.0786,
      "step": 62610
    },
    {
      "epoch": 2.537225745021373,
      "grad_norm": 0.6683362126350403,
      "learning_rate": 1.152736284576008e-05,
      "loss": 1.0593,
      "step": 62620
    },
    {
      "epoch": 2.5376309231984764,
      "grad_norm": 0.6552515625953674,
      "learning_rate": 1.1507579901500909e-05,
      "loss": 1.0271,
      "step": 62630
    },
    {
      "epoch": 2.53803610137558,
      "grad_norm": 0.6662560701370239,
      "learning_rate": 1.1487812910880592e-05,
      "loss": 1.0692,
      "step": 62640
    },
    {
      "epoch": 2.5384412795526834,
      "grad_norm": 0.5998287200927734,
      "learning_rate": 1.1468061877462788e-05,
      "loss": 1.0711,
      "step": 62650
    },
    {
      "epoch": 2.538846457729787,
      "grad_norm": 0.6301242113113403,
      "learning_rate": 1.1448326804808229e-05,
      "loss": 1.078,
      "step": 62660
    },
    {
      "epoch": 2.53925163590689,
      "grad_norm": 0.6042320132255554,
      "learning_rate": 1.1428607696474847e-05,
      "loss": 1.054,
      "step": 62670
    },
    {
      "epoch": 2.5396568140839935,
      "grad_norm": 0.6775391697883606,
      "learning_rate": 1.1408904556017608e-05,
      "loss": 1.0429,
      "step": 62680
    },
    {
      "epoch": 2.540061992261097,
      "grad_norm": 0.6637089848518372,
      "learning_rate": 1.138921738698866e-05,
      "loss": 1.0427,
      "step": 62690
    },
    {
      "epoch": 2.5404671704382,
      "grad_norm": 0.6609033942222595,
      "learning_rate": 1.1369546192937264e-05,
      "loss": 1.0507,
      "step": 62700
    },
    {
      "epoch": 2.5408723486153035,
      "grad_norm": 0.7481523752212524,
      "learning_rate": 1.1349890977409816e-05,
      "loss": 1.0821,
      "step": 62710
    },
    {
      "epoch": 2.541277526792407,
      "grad_norm": 0.6330344080924988,
      "learning_rate": 1.133025174394976e-05,
      "loss": 1.0381,
      "step": 62720
    },
    {
      "epoch": 2.54168270496951,
      "grad_norm": 0.627997100353241,
      "learning_rate": 1.131062849609773e-05,
      "loss": 1.0587,
      "step": 62730
    },
    {
      "epoch": 2.5420878831466136,
      "grad_norm": 0.6913468241691589,
      "learning_rate": 1.1291021237391475e-05,
      "loss": 1.0267,
      "step": 62740
    },
    {
      "epoch": 2.542493061323717,
      "grad_norm": 0.6898131370544434,
      "learning_rate": 1.12714299713658e-05,
      "loss": 1.0463,
      "step": 62750
    },
    {
      "epoch": 2.5428982395008206,
      "grad_norm": 0.6714458465576172,
      "learning_rate": 1.125185470155269e-05,
      "loss": 1.0307,
      "step": 62760
    },
    {
      "epoch": 2.543303417677924,
      "grad_norm": 0.7312777638435364,
      "learning_rate": 1.1232295431481222e-05,
      "loss": 1.0313,
      "step": 62770
    },
    {
      "epoch": 2.543708595855027,
      "grad_norm": 0.6506624221801758,
      "learning_rate": 1.1212752164677609e-05,
      "loss": 1.0216,
      "step": 62780
    },
    {
      "epoch": 2.5441137740321307,
      "grad_norm": 0.6198391318321228,
      "learning_rate": 1.1193224904665134e-05,
      "loss": 1.054,
      "step": 62790
    },
    {
      "epoch": 2.544518952209234,
      "grad_norm": 0.6521852016448975,
      "learning_rate": 1.1173713654964235e-05,
      "loss": 1.0295,
      "step": 62800
    },
    {
      "epoch": 2.5449241303863372,
      "grad_norm": 0.7267647385597229,
      "learning_rate": 1.1154218419092444e-05,
      "loss": 1.048,
      "step": 62810
    },
    {
      "epoch": 2.5453293085634408,
      "grad_norm": 0.6052403450012207,
      "learning_rate": 1.1134739200564426e-05,
      "loss": 1.0303,
      "step": 62820
    },
    {
      "epoch": 2.5457344867405443,
      "grad_norm": 0.6459693312644958,
      "learning_rate": 1.1115276002891927e-05,
      "loss": 1.0422,
      "step": 62830
    },
    {
      "epoch": 2.5461396649176473,
      "grad_norm": 0.6386662125587463,
      "learning_rate": 1.1095828829583843e-05,
      "loss": 1.0392,
      "step": 62840
    },
    {
      "epoch": 2.546544843094751,
      "grad_norm": 0.6758667826652527,
      "learning_rate": 1.1076397684146123e-05,
      "loss": 1.0469,
      "step": 62850
    },
    {
      "epoch": 2.5469500212718543,
      "grad_norm": 0.6420110464096069,
      "learning_rate": 1.1056982570081874e-05,
      "loss": 1.0525,
      "step": 62860
    },
    {
      "epoch": 2.547355199448958,
      "grad_norm": 0.6401052474975586,
      "learning_rate": 1.1037583490891334e-05,
      "loss": 1.0699,
      "step": 62870
    },
    {
      "epoch": 2.5477603776260613,
      "grad_norm": 0.6530700922012329,
      "learning_rate": 1.1018200450071803e-05,
      "loss": 1.051,
      "step": 62880
    },
    {
      "epoch": 2.5481655558031644,
      "grad_norm": 0.6893059015274048,
      "learning_rate": 1.0998833451117685e-05,
      "loss": 1.0502,
      "step": 62890
    },
    {
      "epoch": 2.548570733980268,
      "grad_norm": 0.6946951150894165,
      "learning_rate": 1.0979482497520554e-05,
      "loss": 1.0035,
      "step": 62900
    },
    {
      "epoch": 2.5489759121573714,
      "grad_norm": 0.592856764793396,
      "learning_rate": 1.0960147592768988e-05,
      "loss": 1.0357,
      "step": 62910
    },
    {
      "epoch": 2.5493810903344745,
      "grad_norm": 0.7096565961837769,
      "learning_rate": 1.0940828740348773e-05,
      "loss": 1.0881,
      "step": 62920
    },
    {
      "epoch": 2.549786268511578,
      "grad_norm": 0.8211702108383179,
      "learning_rate": 1.0921525943742762e-05,
      "loss": 1.0419,
      "step": 62930
    },
    {
      "epoch": 2.5501914466886815,
      "grad_norm": 0.6682546138763428,
      "learning_rate": 1.090223920643092e-05,
      "loss": 1.0646,
      "step": 62940
    },
    {
      "epoch": 2.5505966248657845,
      "grad_norm": 0.6187719702720642,
      "learning_rate": 1.088296853189027e-05,
      "loss": 1.0595,
      "step": 62950
    },
    {
      "epoch": 2.551001803042888,
      "grad_norm": 0.6306130886077881,
      "learning_rate": 1.0863713923595032e-05,
      "loss": 1.0413,
      "step": 62960
    },
    {
      "epoch": 2.5514069812199915,
      "grad_norm": 0.6462347507476807,
      "learning_rate": 1.0844475385016394e-05,
      "loss": 1.0546,
      "step": 62970
    },
    {
      "epoch": 2.5518121593970946,
      "grad_norm": 0.6334856152534485,
      "learning_rate": 1.082525291962283e-05,
      "loss": 1.0405,
      "step": 62980
    },
    {
      "epoch": 2.552217337574198,
      "grad_norm": 0.6949588060379028,
      "learning_rate": 1.0806046530879743e-05,
      "loss": 1.0564,
      "step": 62990
    },
    {
      "epoch": 2.5526225157513016,
      "grad_norm": 0.6425763368606567,
      "learning_rate": 1.0786856222249753e-05,
      "loss": 1.059,
      "step": 63000
    },
    {
      "epoch": 2.553027693928405,
      "grad_norm": 0.6409955024719238,
      "learning_rate": 1.0767681997192492e-05,
      "loss": 1.0314,
      "step": 63010
    },
    {
      "epoch": 2.5534328721055086,
      "grad_norm": 0.6697434782981873,
      "learning_rate": 1.0748523859164794e-05,
      "loss": 1.0387,
      "step": 63020
    },
    {
      "epoch": 2.5538380502826117,
      "grad_norm": 0.6463661193847656,
      "learning_rate": 1.0729381811620454e-05,
      "loss": 1.0505,
      "step": 63030
    },
    {
      "epoch": 2.554243228459715,
      "grad_norm": 0.6385862827301025,
      "learning_rate": 1.0710255858010554e-05,
      "loss": 1.0508,
      "step": 63040
    },
    {
      "epoch": 2.5546484066368187,
      "grad_norm": 0.6122366189956665,
      "learning_rate": 1.0691146001783081e-05,
      "loss": 1.0479,
      "step": 63050
    },
    {
      "epoch": 2.5550535848139218,
      "grad_norm": 0.6900201439857483,
      "learning_rate": 1.067205224638328e-05,
      "loss": 1.0585,
      "step": 63060
    },
    {
      "epoch": 2.5554587629910253,
      "grad_norm": 0.6562058925628662,
      "learning_rate": 1.0652974595253363e-05,
      "loss": 1.039,
      "step": 63070
    },
    {
      "epoch": 2.5558639411681288,
      "grad_norm": 0.6418850421905518,
      "learning_rate": 1.0633913051832723e-05,
      "loss": 1.0531,
      "step": 63080
    },
    {
      "epoch": 2.556269119345232,
      "grad_norm": 0.7037691473960876,
      "learning_rate": 1.0614867619557811e-05,
      "loss": 1.0594,
      "step": 63090
    },
    {
      "epoch": 2.5566742975223353,
      "grad_norm": 0.5960628390312195,
      "learning_rate": 1.0595838301862227e-05,
      "loss": 1.0196,
      "step": 63100
    },
    {
      "epoch": 2.557079475699439,
      "grad_norm": 0.6772860288619995,
      "learning_rate": 1.0576825102176568e-05,
      "loss": 1.0505,
      "step": 63110
    },
    {
      "epoch": 2.5574846538765423,
      "grad_norm": 0.6383720636367798,
      "learning_rate": 1.0557828023928607e-05,
      "loss": 1.0334,
      "step": 63120
    },
    {
      "epoch": 2.557889832053646,
      "grad_norm": 0.6520067453384399,
      "learning_rate": 1.0538847070543211e-05,
      "loss": 1.0593,
      "step": 63130
    },
    {
      "epoch": 2.558295010230749,
      "grad_norm": 0.6111291646957397,
      "learning_rate": 1.051988224544228e-05,
      "loss": 1.0347,
      "step": 63140
    },
    {
      "epoch": 2.5587001884078524,
      "grad_norm": 0.6984622478485107,
      "learning_rate": 1.0500933552044845e-05,
      "loss": 1.0188,
      "step": 63150
    },
    {
      "epoch": 2.559105366584956,
      "grad_norm": 0.7472353577613831,
      "learning_rate": 1.0482000993767027e-05,
      "loss": 1.0469,
      "step": 63160
    },
    {
      "epoch": 2.559510544762059,
      "grad_norm": 0.6526617407798767,
      "learning_rate": 1.0463084574022064e-05,
      "loss": 1.0596,
      "step": 63170
    },
    {
      "epoch": 2.5599157229391625,
      "grad_norm": 0.6292341947555542,
      "learning_rate": 1.0444184296220205e-05,
      "loss": 1.0842,
      "step": 63180
    },
    {
      "epoch": 2.560320901116266,
      "grad_norm": 0.7004529237747192,
      "learning_rate": 1.0425300163768902e-05,
      "loss": 1.0659,
      "step": 63190
    },
    {
      "epoch": 2.560726079293369,
      "grad_norm": 0.6709743738174438,
      "learning_rate": 1.0406432180072545e-05,
      "loss": 1.0567,
      "step": 63200
    },
    {
      "epoch": 2.5611312574704725,
      "grad_norm": 0.6055924892425537,
      "learning_rate": 1.038758034853281e-05,
      "loss": 1.0505,
      "step": 63210
    },
    {
      "epoch": 2.561536435647576,
      "grad_norm": 0.7864584922790527,
      "learning_rate": 1.036874467254828e-05,
      "loss": 1.1001,
      "step": 63220
    },
    {
      "epoch": 2.5619416138246796,
      "grad_norm": 0.6296778917312622,
      "learning_rate": 1.0349925155514728e-05,
      "loss": 1.0541,
      "step": 63230
    },
    {
      "epoch": 2.562346792001783,
      "grad_norm": 0.6446629166603088,
      "learning_rate": 1.0331121800824972e-05,
      "loss": 1.0419,
      "step": 63240
    },
    {
      "epoch": 2.562751970178886,
      "grad_norm": 0.5892048478126526,
      "learning_rate": 1.0312334611868923e-05,
      "loss": 1.0512,
      "step": 63250
    },
    {
      "epoch": 2.5631571483559896,
      "grad_norm": 0.6184515953063965,
      "learning_rate": 1.0293563592033595e-05,
      "loss": 1.0231,
      "step": 63260
    },
    {
      "epoch": 2.563562326533093,
      "grad_norm": 0.6294946670532227,
      "learning_rate": 1.027480874470309e-05,
      "loss": 1.0374,
      "step": 63270
    },
    {
      "epoch": 2.563967504710196,
      "grad_norm": 0.6747177243232727,
      "learning_rate": 1.0256070073258539e-05,
      "loss": 1.0778,
      "step": 63280
    },
    {
      "epoch": 2.5643726828872997,
      "grad_norm": 0.6005906462669373,
      "learning_rate": 1.0237347581078249e-05,
      "loss": 1.0195,
      "step": 63290
    },
    {
      "epoch": 2.564777861064403,
      "grad_norm": 0.6721743941307068,
      "learning_rate": 1.021864127153751e-05,
      "loss": 1.0509,
      "step": 63300
    },
    {
      "epoch": 2.5651830392415063,
      "grad_norm": 0.6471602916717529,
      "learning_rate": 1.0199951148008757e-05,
      "loss": 1.0224,
      "step": 63310
    },
    {
      "epoch": 2.5655882174186098,
      "grad_norm": 0.6746918559074402,
      "learning_rate": 1.0181277213861507e-05,
      "loss": 1.0518,
      "step": 63320
    },
    {
      "epoch": 2.5659933955957133,
      "grad_norm": 0.6532706618309021,
      "learning_rate": 1.0162619472462354e-05,
      "loss": 1.0444,
      "step": 63330
    },
    {
      "epoch": 2.5663985737728163,
      "grad_norm": 0.6472551822662354,
      "learning_rate": 1.0143977927174919e-05,
      "loss": 1.056,
      "step": 63340
    },
    {
      "epoch": 2.56680375194992,
      "grad_norm": 0.6260112524032593,
      "learning_rate": 1.0125352581360003e-05,
      "loss": 1.0417,
      "step": 63350
    },
    {
      "epoch": 2.5672089301270233,
      "grad_norm": 0.7527186274528503,
      "learning_rate": 1.0106743438375388e-05,
      "loss": 1.053,
      "step": 63360
    },
    {
      "epoch": 2.567614108304127,
      "grad_norm": 0.6101062893867493,
      "learning_rate": 1.0088150501575988e-05,
      "loss": 1.0386,
      "step": 63370
    },
    {
      "epoch": 2.5680192864812303,
      "grad_norm": 0.7048624753952026,
      "learning_rate": 1.0069573774313778e-05,
      "loss": 1.0534,
      "step": 63380
    },
    {
      "epoch": 2.5684244646583334,
      "grad_norm": 0.6812266111373901,
      "learning_rate": 1.0051013259937859e-05,
      "loss": 1.0596,
      "step": 63390
    },
    {
      "epoch": 2.568829642835437,
      "grad_norm": 0.6004383563995361,
      "learning_rate": 1.0032468961794317e-05,
      "loss": 1.0338,
      "step": 63400
    },
    {
      "epoch": 2.5692348210125404,
      "grad_norm": 0.6658854484558105,
      "learning_rate": 1.0013940883226391e-05,
      "loss": 1.0326,
      "step": 63410
    },
    {
      "epoch": 2.5696399991896435,
      "grad_norm": 0.6994752883911133,
      "learning_rate": 9.995429027574355e-06,
      "loss": 1.0273,
      "step": 63420
    },
    {
      "epoch": 2.570045177366747,
      "grad_norm": 0.6766693592071533,
      "learning_rate": 9.976933398175604e-06,
      "loss": 1.0162,
      "step": 63430
    },
    {
      "epoch": 2.5704503555438505,
      "grad_norm": 0.7036154270172119,
      "learning_rate": 9.958453998364548e-06,
      "loss": 1.0046,
      "step": 63440
    },
    {
      "epoch": 2.5708555337209535,
      "grad_norm": 0.6658051609992981,
      "learning_rate": 9.9399908314727e-06,
      "loss": 1.0008,
      "step": 63450
    },
    {
      "epoch": 2.571260711898057,
      "grad_norm": 0.5886639356613159,
      "learning_rate": 9.921543900828668e-06,
      "loss": 1.0477,
      "step": 63460
    },
    {
      "epoch": 2.5716658900751606,
      "grad_norm": 0.638578474521637,
      "learning_rate": 9.903113209758096e-06,
      "loss": 1.0681,
      "step": 63470
    },
    {
      "epoch": 2.572071068252264,
      "grad_norm": 0.646210789680481,
      "learning_rate": 9.884698761583711e-06,
      "loss": 1.0176,
      "step": 63480
    },
    {
      "epoch": 2.5724762464293676,
      "grad_norm": 0.7032479047775269,
      "learning_rate": 9.86630055962532e-06,
      "loss": 1.0513,
      "step": 63490
    },
    {
      "epoch": 2.5728814246064706,
      "grad_norm": 0.6498873233795166,
      "learning_rate": 9.847918607199824e-06,
      "loss": 1.0528,
      "step": 63500
    },
    {
      "epoch": 2.573286602783574,
      "grad_norm": 0.6717117428779602,
      "learning_rate": 9.829552907621131e-06,
      "loss": 1.0628,
      "step": 63510
    },
    {
      "epoch": 2.5736917809606776,
      "grad_norm": 0.6922509670257568,
      "learning_rate": 9.811203464200291e-06,
      "loss": 1.0451,
      "step": 63520
    },
    {
      "epoch": 2.5740969591377807,
      "grad_norm": 0.6946414113044739,
      "learning_rate": 9.792870280245359e-06,
      "loss": 1.041,
      "step": 63530
    },
    {
      "epoch": 2.574502137314884,
      "grad_norm": 0.6538294553756714,
      "learning_rate": 9.774553359061489e-06,
      "loss": 1.0293,
      "step": 63540
    },
    {
      "epoch": 2.5749073154919877,
      "grad_norm": 0.7274805903434753,
      "learning_rate": 9.756252703950907e-06,
      "loss": 1.0686,
      "step": 63550
    },
    {
      "epoch": 2.5753124936690908,
      "grad_norm": 0.7142536640167236,
      "learning_rate": 9.73796831821293e-06,
      "loss": 1.0472,
      "step": 63560
    },
    {
      "epoch": 2.5757176718461943,
      "grad_norm": 0.6753360033035278,
      "learning_rate": 9.719700205143866e-06,
      "loss": 1.101,
      "step": 63570
    },
    {
      "epoch": 2.5761228500232978,
      "grad_norm": 0.6210930943489075,
      "learning_rate": 9.70144836803719e-06,
      "loss": 1.046,
      "step": 63580
    },
    {
      "epoch": 2.5765280282004013,
      "grad_norm": 0.6975363492965698,
      "learning_rate": 9.683212810183318e-06,
      "loss": 1.0647,
      "step": 63590
    },
    {
      "epoch": 2.576933206377505,
      "grad_norm": 0.718923032283783,
      "learning_rate": 9.664993534869882e-06,
      "loss": 1.0931,
      "step": 63600
    },
    {
      "epoch": 2.577338384554608,
      "grad_norm": 0.6226449012756348,
      "learning_rate": 9.646790545381445e-06,
      "loss": 1.0701,
      "step": 63610
    },
    {
      "epoch": 2.5777435627317113,
      "grad_norm": 0.6675795316696167,
      "learning_rate": 9.628603844999717e-06,
      "loss": 1.0644,
      "step": 63620
    },
    {
      "epoch": 2.578148740908815,
      "grad_norm": 0.6072681546211243,
      "learning_rate": 9.610433437003419e-06,
      "loss": 1.0552,
      "step": 63630
    },
    {
      "epoch": 2.578553919085918,
      "grad_norm": 0.625006914138794,
      "learning_rate": 9.592279324668397e-06,
      "loss": 1.0353,
      "step": 63640
    },
    {
      "epoch": 2.5789590972630214,
      "grad_norm": 0.6306703090667725,
      "learning_rate": 9.574141511267443e-06,
      "loss": 1.0442,
      "step": 63650
    },
    {
      "epoch": 2.579364275440125,
      "grad_norm": 0.6155623197555542,
      "learning_rate": 9.556020000070586e-06,
      "loss": 1.0852,
      "step": 63660
    },
    {
      "epoch": 2.579769453617228,
      "grad_norm": 0.6230387687683105,
      "learning_rate": 9.537914794344749e-06,
      "loss": 1.0479,
      "step": 63670
    },
    {
      "epoch": 2.5801746317943315,
      "grad_norm": 0.6416307091712952,
      "learning_rate": 9.51982589735403e-06,
      "loss": 1.0511,
      "step": 63680
    },
    {
      "epoch": 2.580579809971435,
      "grad_norm": 0.5774282813072205,
      "learning_rate": 9.501753312359507e-06,
      "loss": 1.04,
      "step": 63690
    },
    {
      "epoch": 2.580984988148538,
      "grad_norm": 0.69566410779953,
      "learning_rate": 9.483697042619354e-06,
      "loss": 1.0368,
      "step": 63700
    },
    {
      "epoch": 2.5813901663256416,
      "grad_norm": 0.668522298336029,
      "learning_rate": 9.465657091388813e-06,
      "loss": 1.0332,
      "step": 63710
    },
    {
      "epoch": 2.581795344502745,
      "grad_norm": 0.6623801589012146,
      "learning_rate": 9.447633461920191e-06,
      "loss": 1.0484,
      "step": 63720
    },
    {
      "epoch": 2.5822005226798486,
      "grad_norm": 0.6671329140663147,
      "learning_rate": 9.42962615746279e-06,
      "loss": 1.0253,
      "step": 63730
    },
    {
      "epoch": 2.582605700856952,
      "grad_norm": 0.7117037773132324,
      "learning_rate": 9.411635181263034e-06,
      "loss": 1.0584,
      "step": 63740
    },
    {
      "epoch": 2.583010879034055,
      "grad_norm": 0.7121086120605469,
      "learning_rate": 9.393660536564408e-06,
      "loss": 1.0577,
      "step": 63750
    },
    {
      "epoch": 2.5834160572111586,
      "grad_norm": 0.6177279949188232,
      "learning_rate": 9.375702226607386e-06,
      "loss": 1.0622,
      "step": 63760
    },
    {
      "epoch": 2.583821235388262,
      "grad_norm": 0.630750298500061,
      "learning_rate": 9.357760254629543e-06,
      "loss": 1.0485,
      "step": 63770
    },
    {
      "epoch": 2.584226413565365,
      "grad_norm": 0.6413726210594177,
      "learning_rate": 9.339834623865529e-06,
      "loss": 1.0273,
      "step": 63780
    },
    {
      "epoch": 2.5846315917424687,
      "grad_norm": 0.6252331137657166,
      "learning_rate": 9.321925337547022e-06,
      "loss": 1.0222,
      "step": 63790
    },
    {
      "epoch": 2.585036769919572,
      "grad_norm": 0.6834114789962769,
      "learning_rate": 9.304032398902718e-06,
      "loss": 1.0401,
      "step": 63800
    },
    {
      "epoch": 2.5854419480966753,
      "grad_norm": 0.6251487135887146,
      "learning_rate": 9.286155811158449e-06,
      "loss": 1.0438,
      "step": 63810
    },
    {
      "epoch": 2.5858471262737788,
      "grad_norm": 0.7588444948196411,
      "learning_rate": 9.268295577536978e-06,
      "loss": 1.0634,
      "step": 63820
    },
    {
      "epoch": 2.5862523044508823,
      "grad_norm": 0.7019191980361938,
      "learning_rate": 9.250451701258289e-06,
      "loss": 1.0369,
      "step": 63830
    },
    {
      "epoch": 2.586657482627986,
      "grad_norm": 0.6282879710197449,
      "learning_rate": 9.232624185539251e-06,
      "loss": 1.0463,
      "step": 63840
    },
    {
      "epoch": 2.5870626608050893,
      "grad_norm": 0.6331921219825745,
      "learning_rate": 9.214813033593895e-06,
      "loss": 1.0237,
      "step": 63850
    },
    {
      "epoch": 2.5874678389821923,
      "grad_norm": 0.6534949541091919,
      "learning_rate": 9.197018248633226e-06,
      "loss": 1.0478,
      "step": 63860
    },
    {
      "epoch": 2.587873017159296,
      "grad_norm": 0.5678619742393494,
      "learning_rate": 9.17923983386535e-06,
      "loss": 1.0106,
      "step": 63870
    },
    {
      "epoch": 2.5882781953363994,
      "grad_norm": 0.6136121153831482,
      "learning_rate": 9.161477792495398e-06,
      "loss": 1.02,
      "step": 63880
    },
    {
      "epoch": 2.5886833735135024,
      "grad_norm": 0.702462911605835,
      "learning_rate": 9.14373212772559e-06,
      "loss": 1.0685,
      "step": 63890
    },
    {
      "epoch": 2.589088551690606,
      "grad_norm": 0.6370589137077332,
      "learning_rate": 9.12600284275511e-06,
      "loss": 1.0541,
      "step": 63900
    },
    {
      "epoch": 2.5894937298677094,
      "grad_norm": 0.7284108400344849,
      "learning_rate": 9.10828994078028e-06,
      "loss": 1.0589,
      "step": 63910
    },
    {
      "epoch": 2.5898989080448125,
      "grad_norm": 0.687861979007721,
      "learning_rate": 9.090593424994399e-06,
      "loss": 1.0658,
      "step": 63920
    },
    {
      "epoch": 2.590304086221916,
      "grad_norm": 0.6021081209182739,
      "learning_rate": 9.072913298587838e-06,
      "loss": 1.0406,
      "step": 63930
    },
    {
      "epoch": 2.5907092643990195,
      "grad_norm": 0.6238572001457214,
      "learning_rate": 9.055249564748025e-06,
      "loss": 1.0093,
      "step": 63940
    },
    {
      "epoch": 2.591114442576123,
      "grad_norm": 0.6446263790130615,
      "learning_rate": 9.037602226659447e-06,
      "loss": 1.0804,
      "step": 63950
    },
    {
      "epoch": 2.5915196207532265,
      "grad_norm": 0.6099786162376404,
      "learning_rate": 9.019971287503571e-06,
      "loss": 1.0821,
      "step": 63960
    },
    {
      "epoch": 2.5919247989303296,
      "grad_norm": 0.6503088474273682,
      "learning_rate": 9.002356750458984e-06,
      "loss": 1.0658,
      "step": 63970
    },
    {
      "epoch": 2.592329977107433,
      "grad_norm": 0.5218167901039124,
      "learning_rate": 8.98475861870124e-06,
      "loss": 1.0284,
      "step": 63980
    },
    {
      "epoch": 2.5927351552845366,
      "grad_norm": 0.6116967797279358,
      "learning_rate": 8.967176895402996e-06,
      "loss": 1.0574,
      "step": 63990
    },
    {
      "epoch": 2.5931403334616396,
      "grad_norm": 0.6642605662345886,
      "learning_rate": 8.94961158373393e-06,
      "loss": 1.04,
      "step": 64000
    },
    {
      "epoch": 2.593545511638743,
      "grad_norm": 0.7253957986831665,
      "learning_rate": 8.932062686860787e-06,
      "loss": 1.0411,
      "step": 64010
    },
    {
      "epoch": 2.5939506898158466,
      "grad_norm": 0.5908827185630798,
      "learning_rate": 8.914530207947269e-06,
      "loss": 1.0043,
      "step": 64020
    },
    {
      "epoch": 2.5943558679929497,
      "grad_norm": 0.6512472033500671,
      "learning_rate": 8.897014150154237e-06,
      "loss": 1.0556,
      "step": 64030
    },
    {
      "epoch": 2.594761046170053,
      "grad_norm": 0.5881099104881287,
      "learning_rate": 8.879514516639454e-06,
      "loss": 1.0567,
      "step": 64040
    },
    {
      "epoch": 2.5951662243471567,
      "grad_norm": 0.673605740070343,
      "learning_rate": 8.862031310557883e-06,
      "loss": 1.0762,
      "step": 64050
    },
    {
      "epoch": 2.5955714025242598,
      "grad_norm": 0.6863631010055542,
      "learning_rate": 8.844564535061395e-06,
      "loss": 1.055,
      "step": 64060
    },
    {
      "epoch": 2.5959765807013633,
      "grad_norm": 0.730025053024292,
      "learning_rate": 8.827114193298958e-06,
      "loss": 1.0527,
      "step": 64070
    },
    {
      "epoch": 2.596381758878467,
      "grad_norm": 0.6387914419174194,
      "learning_rate": 8.809680288416555e-06,
      "loss": 1.0495,
      "step": 64080
    },
    {
      "epoch": 2.5967869370555703,
      "grad_norm": 0.681715190410614,
      "learning_rate": 8.792262823557207e-06,
      "loss": 1.0437,
      "step": 64090
    },
    {
      "epoch": 2.597192115232674,
      "grad_norm": 0.5814065337181091,
      "learning_rate": 8.774861801861e-06,
      "loss": 1.0578,
      "step": 64100
    },
    {
      "epoch": 2.597597293409777,
      "grad_norm": 0.6233329772949219,
      "learning_rate": 8.757477226465039e-06,
      "loss": 1.0293,
      "step": 64110
    },
    {
      "epoch": 2.5980024715868804,
      "grad_norm": 0.685470461845398,
      "learning_rate": 8.740109100503425e-06,
      "loss": 1.0356,
      "step": 64120
    },
    {
      "epoch": 2.598407649763984,
      "grad_norm": 0.6811968088150024,
      "learning_rate": 8.722757427107353e-06,
      "loss": 1.025,
      "step": 64130
    },
    {
      "epoch": 2.598812827941087,
      "grad_norm": 0.6374936699867249,
      "learning_rate": 8.705422209405033e-06,
      "loss": 1.0142,
      "step": 64140
    },
    {
      "epoch": 2.5992180061181904,
      "grad_norm": 0.6665769815444946,
      "learning_rate": 8.688103450521679e-06,
      "loss": 1.0318,
      "step": 64150
    },
    {
      "epoch": 2.599623184295294,
      "grad_norm": 0.6762329339981079,
      "learning_rate": 8.670801153579577e-06,
      "loss": 1.0492,
      "step": 64160
    },
    {
      "epoch": 2.600028362472397,
      "grad_norm": 0.6924728155136108,
      "learning_rate": 8.653515321698025e-06,
      "loss": 1.0575,
      "step": 64170
    },
    {
      "epoch": 2.6004335406495005,
      "grad_norm": 0.6844394207000732,
      "learning_rate": 8.636245957993372e-06,
      "loss": 1.0354,
      "step": 64180
    },
    {
      "epoch": 2.600838718826604,
      "grad_norm": 0.6112249493598938,
      "learning_rate": 8.61899306557895e-06,
      "loss": 1.0596,
      "step": 64190
    },
    {
      "epoch": 2.6012438970037075,
      "grad_norm": 0.6828764081001282,
      "learning_rate": 8.601756647565195e-06,
      "loss": 1.0997,
      "step": 64200
    },
    {
      "epoch": 2.601649075180811,
      "grad_norm": 0.7811608910560608,
      "learning_rate": 8.584536707059476e-06,
      "loss": 1.0577,
      "step": 64210
    },
    {
      "epoch": 2.602054253357914,
      "grad_norm": 0.6192735433578491,
      "learning_rate": 8.56733324716632e-06,
      "loss": 1.025,
      "step": 64220
    },
    {
      "epoch": 2.6024594315350176,
      "grad_norm": 0.656994104385376,
      "learning_rate": 8.550146270987147e-06,
      "loss": 1.0882,
      "step": 64230
    },
    {
      "epoch": 2.602864609712121,
      "grad_norm": 0.5723156929016113,
      "learning_rate": 8.532975781620512e-06,
      "loss": 0.9986,
      "step": 64240
    },
    {
      "epoch": 2.603269787889224,
      "grad_norm": 0.6488220691680908,
      "learning_rate": 8.515821782161926e-06,
      "loss": 1.0361,
      "step": 64250
    },
    {
      "epoch": 2.6036749660663276,
      "grad_norm": 0.5762472748756409,
      "learning_rate": 8.498684275703983e-06,
      "loss": 1.0513,
      "step": 64260
    },
    {
      "epoch": 2.604080144243431,
      "grad_norm": 0.7161347270011902,
      "learning_rate": 8.481563265336222e-06,
      "loss": 1.0557,
      "step": 64270
    },
    {
      "epoch": 2.604485322420534,
      "grad_norm": 0.6740996241569519,
      "learning_rate": 8.464458754145332e-06,
      "loss": 1.0581,
      "step": 64280
    },
    {
      "epoch": 2.6048905005976377,
      "grad_norm": 0.606200098991394,
      "learning_rate": 8.44737074521491e-06,
      "loss": 1.1051,
      "step": 64290
    },
    {
      "epoch": 2.605295678774741,
      "grad_norm": 0.6960146427154541,
      "learning_rate": 8.430299241625672e-06,
      "loss": 1.0483,
      "step": 64300
    },
    {
      "epoch": 2.6057008569518447,
      "grad_norm": 0.5631593465805054,
      "learning_rate": 8.413244246455254e-06,
      "loss": 1.037,
      "step": 64310
    },
    {
      "epoch": 2.6061060351289482,
      "grad_norm": 0.6377202868461609,
      "learning_rate": 8.396205762778397e-06,
      "loss": 1.0049,
      "step": 64320
    },
    {
      "epoch": 2.6065112133060513,
      "grad_norm": 0.6213030815124512,
      "learning_rate": 8.379183793666856e-06,
      "loss": 1.0343,
      "step": 64330
    },
    {
      "epoch": 2.606916391483155,
      "grad_norm": 0.6792376637458801,
      "learning_rate": 8.362178342189397e-06,
      "loss": 1.0112,
      "step": 64340
    },
    {
      "epoch": 2.6073215696602583,
      "grad_norm": 0.6555320620536804,
      "learning_rate": 8.345189411411791e-06,
      "loss": 1.0078,
      "step": 64350
    },
    {
      "epoch": 2.6077267478373614,
      "grad_norm": 0.6578645706176758,
      "learning_rate": 8.328217004396854e-06,
      "loss": 1.0422,
      "step": 64360
    },
    {
      "epoch": 2.608131926014465,
      "grad_norm": 0.6310391426086426,
      "learning_rate": 8.311261124204395e-06,
      "loss": 1.0534,
      "step": 64370
    },
    {
      "epoch": 2.6085371041915684,
      "grad_norm": 0.6190862059593201,
      "learning_rate": 8.29432177389129e-06,
      "loss": 1.0366,
      "step": 64380
    },
    {
      "epoch": 2.6089422823686714,
      "grad_norm": 0.6551217436790466,
      "learning_rate": 8.277398956511384e-06,
      "loss": 1.0682,
      "step": 64390
    },
    {
      "epoch": 2.609347460545775,
      "grad_norm": 0.6227602958679199,
      "learning_rate": 8.260492675115604e-06,
      "loss": 1.0641,
      "step": 64400
    },
    {
      "epoch": 2.6097526387228784,
      "grad_norm": 0.67998206615448,
      "learning_rate": 8.243602932751815e-06,
      "loss": 1.0497,
      "step": 64410
    },
    {
      "epoch": 2.6101578168999815,
      "grad_norm": 0.6539530158042908,
      "learning_rate": 8.226729732464954e-06,
      "loss": 1.0548,
      "step": 64420
    },
    {
      "epoch": 2.610562995077085,
      "grad_norm": 0.6113176345825195,
      "learning_rate": 8.209873077296993e-06,
      "loss": 1.0347,
      "step": 64430
    },
    {
      "epoch": 2.6109681732541885,
      "grad_norm": 0.7026630640029907,
      "learning_rate": 8.193032970286851e-06,
      "loss": 1.0324,
      "step": 64440
    },
    {
      "epoch": 2.611373351431292,
      "grad_norm": 0.6272042393684387,
      "learning_rate": 8.176209414470525e-06,
      "loss": 1.0456,
      "step": 64450
    },
    {
      "epoch": 2.6117785296083955,
      "grad_norm": 0.6670408248901367,
      "learning_rate": 8.159402412881012e-06,
      "loss": 1.0581,
      "step": 64460
    },
    {
      "epoch": 2.6121837077854986,
      "grad_norm": 0.7318409085273743,
      "learning_rate": 8.14261196854833e-06,
      "loss": 1.0559,
      "step": 64470
    },
    {
      "epoch": 2.612588885962602,
      "grad_norm": 0.5690757632255554,
      "learning_rate": 8.125838084499482e-06,
      "loss": 1.0238,
      "step": 64480
    },
    {
      "epoch": 2.6129940641397056,
      "grad_norm": 0.7260243892669678,
      "learning_rate": 8.109080763758503e-06,
      "loss": 1.0354,
      "step": 64490
    },
    {
      "epoch": 2.6133992423168086,
      "grad_norm": 0.6771280765533447,
      "learning_rate": 8.092340009346466e-06,
      "loss": 1.05,
      "step": 64500
    },
    {
      "epoch": 2.613804420493912,
      "grad_norm": 0.6482203602790833,
      "learning_rate": 8.075615824281434e-06,
      "loss": 1.0602,
      "step": 64510
    },
    {
      "epoch": 2.6142095986710157,
      "grad_norm": 0.7163315415382385,
      "learning_rate": 8.058908211578475e-06,
      "loss": 1.0621,
      "step": 64520
    },
    {
      "epoch": 2.6146147768481187,
      "grad_norm": 0.6158546805381775,
      "learning_rate": 8.042217174249699e-06,
      "loss": 1.0571,
      "step": 64530
    },
    {
      "epoch": 2.615019955025222,
      "grad_norm": 0.643213152885437,
      "learning_rate": 8.02554271530418e-06,
      "loss": 1.0445,
      "step": 64540
    },
    {
      "epoch": 2.6154251332023257,
      "grad_norm": 0.6476795077323914,
      "learning_rate": 8.008884837748031e-06,
      "loss": 1.0479,
      "step": 64550
    },
    {
      "epoch": 2.6158303113794292,
      "grad_norm": 0.6520179510116577,
      "learning_rate": 7.99224354458441e-06,
      "loss": 1.0468,
      "step": 64560
    },
    {
      "epoch": 2.6162354895565327,
      "grad_norm": 0.6652622222900391,
      "learning_rate": 7.975618838813448e-06,
      "loss": 1.0518,
      "step": 64570
    },
    {
      "epoch": 2.616640667733636,
      "grad_norm": 0.578407347202301,
      "learning_rate": 7.959010723432248e-06,
      "loss": 1.0651,
      "step": 64580
    },
    {
      "epoch": 2.6170458459107393,
      "grad_norm": 0.6812475323677063,
      "learning_rate": 7.942419201435013e-06,
      "loss": 1.0523,
      "step": 64590
    },
    {
      "epoch": 2.617451024087843,
      "grad_norm": 0.7072286605834961,
      "learning_rate": 7.925844275812866e-06,
      "loss": 1.0687,
      "step": 64600
    },
    {
      "epoch": 2.617856202264946,
      "grad_norm": 0.6304755210876465,
      "learning_rate": 7.909285949553991e-06,
      "loss": 1.0602,
      "step": 64610
    },
    {
      "epoch": 2.6182613804420494,
      "grad_norm": 0.5797163248062134,
      "learning_rate": 7.892744225643579e-06,
      "loss": 1.0704,
      "step": 64620
    },
    {
      "epoch": 2.618666558619153,
      "grad_norm": 0.6200985312461853,
      "learning_rate": 7.876219107063799e-06,
      "loss": 1.0381,
      "step": 64630
    },
    {
      "epoch": 2.619071736796256,
      "grad_norm": 0.7104570865631104,
      "learning_rate": 7.859710596793845e-06,
      "loss": 1.0463,
      "step": 64640
    },
    {
      "epoch": 2.6194769149733594,
      "grad_norm": 0.6576548218727112,
      "learning_rate": 7.843218697809917e-06,
      "loss": 1.059,
      "step": 64650
    },
    {
      "epoch": 2.619882093150463,
      "grad_norm": 0.6742052435874939,
      "learning_rate": 7.826743413085192e-06,
      "loss": 1.0594,
      "step": 64660
    },
    {
      "epoch": 2.6202872713275664,
      "grad_norm": 0.6988142728805542,
      "learning_rate": 7.810284745589925e-06,
      "loss": 1.01,
      "step": 64670
    },
    {
      "epoch": 2.62069244950467,
      "grad_norm": 0.5744436979293823,
      "learning_rate": 7.793842698291276e-06,
      "loss": 1.0353,
      "step": 64680
    },
    {
      "epoch": 2.621097627681773,
      "grad_norm": 0.6603493094444275,
      "learning_rate": 7.777417274153509e-06,
      "loss": 1.0581,
      "step": 64690
    },
    {
      "epoch": 2.6215028058588765,
      "grad_norm": 0.6489821672439575,
      "learning_rate": 7.761008476137788e-06,
      "loss": 1.0442,
      "step": 64700
    },
    {
      "epoch": 2.62190798403598,
      "grad_norm": 0.6674945950508118,
      "learning_rate": 7.744616307202357e-06,
      "loss": 1.0472,
      "step": 64710
    },
    {
      "epoch": 2.622313162213083,
      "grad_norm": 0.6469969153404236,
      "learning_rate": 7.728240770302441e-06,
      "loss": 1.0768,
      "step": 64720
    },
    {
      "epoch": 2.6227183403901866,
      "grad_norm": 0.6207972168922424,
      "learning_rate": 7.711881868390291e-06,
      "loss": 1.0355,
      "step": 64730
    },
    {
      "epoch": 2.62312351856729,
      "grad_norm": 0.7144344449043274,
      "learning_rate": 7.69553960441508e-06,
      "loss": 1.0657,
      "step": 64740
    },
    {
      "epoch": 2.623528696744393,
      "grad_norm": 0.6371684074401855,
      "learning_rate": 7.679213981323053e-06,
      "loss": 1.0516,
      "step": 64750
    },
    {
      "epoch": 2.6239338749214967,
      "grad_norm": 0.6930067539215088,
      "learning_rate": 7.662905002057453e-06,
      "loss": 1.029,
      "step": 64760
    },
    {
      "epoch": 2.6243390530986,
      "grad_norm": 0.6952108144760132,
      "learning_rate": 7.646612669558472e-06,
      "loss": 1.0049,
      "step": 64770
    },
    {
      "epoch": 2.624744231275703,
      "grad_norm": 0.6577498316764832,
      "learning_rate": 7.630336986763353e-06,
      "loss": 1.0433,
      "step": 64780
    },
    {
      "epoch": 2.6251494094528067,
      "grad_norm": 0.644213855266571,
      "learning_rate": 7.6140779566063135e-06,
      "loss": 1.0306,
      "step": 64790
    },
    {
      "epoch": 2.6255545876299102,
      "grad_norm": 0.6772760152816772,
      "learning_rate": 7.5978355820185865e-06,
      "loss": 1.019,
      "step": 64800
    },
    {
      "epoch": 2.6259597658070137,
      "grad_norm": 0.6066564321517944,
      "learning_rate": 7.581609865928363e-06,
      "loss": 1.0214,
      "step": 64810
    },
    {
      "epoch": 2.6263649439841172,
      "grad_norm": 0.6227908134460449,
      "learning_rate": 7.565400811260892e-06,
      "loss": 1.0373,
      "step": 64820
    },
    {
      "epoch": 2.6267701221612203,
      "grad_norm": 0.6441375613212585,
      "learning_rate": 7.5492084209383254e-06,
      "loss": 1.0598,
      "step": 64830
    },
    {
      "epoch": 2.627175300338324,
      "grad_norm": 0.6656818389892578,
      "learning_rate": 7.533032697879927e-06,
      "loss": 1.0478,
      "step": 64840
    },
    {
      "epoch": 2.6275804785154273,
      "grad_norm": 0.6837387084960938,
      "learning_rate": 7.516873645001865e-06,
      "loss": 1.0568,
      "step": 64850
    },
    {
      "epoch": 2.6279856566925304,
      "grad_norm": 0.7047874927520752,
      "learning_rate": 7.500731265217364e-06,
      "loss": 1.046,
      "step": 64860
    },
    {
      "epoch": 2.628390834869634,
      "grad_norm": 0.6862103939056396,
      "learning_rate": 7.484605561436575e-06,
      "loss": 1.0407,
      "step": 64870
    },
    {
      "epoch": 2.6287960130467374,
      "grad_norm": 0.6609403491020203,
      "learning_rate": 7.468496536566716e-06,
      "loss": 1.0312,
      "step": 64880
    },
    {
      "epoch": 2.6292011912238404,
      "grad_norm": 0.7305052876472473,
      "learning_rate": 7.452404193511908e-06,
      "loss": 1.0275,
      "step": 64890
    },
    {
      "epoch": 2.629606369400944,
      "grad_norm": 0.6743596196174622,
      "learning_rate": 7.436328535173398e-06,
      "loss": 1.0479,
      "step": 64900
    },
    {
      "epoch": 2.6300115475780474,
      "grad_norm": 0.6514437794685364,
      "learning_rate": 7.420269564449278e-06,
      "loss": 1.0652,
      "step": 64910
    },
    {
      "epoch": 2.630416725755151,
      "grad_norm": 0.6548585295677185,
      "learning_rate": 7.404227284234755e-06,
      "loss": 1.0601,
      "step": 64920
    },
    {
      "epoch": 2.6308219039322545,
      "grad_norm": 0.619291365146637,
      "learning_rate": 7.388201697421926e-06,
      "loss": 1.0633,
      "step": 64930
    },
    {
      "epoch": 2.6312270821093575,
      "grad_norm": 0.6096026301383972,
      "learning_rate": 7.372192806899947e-06,
      "loss": 1.0469,
      "step": 64940
    },
    {
      "epoch": 2.631632260286461,
      "grad_norm": 0.6029451489448547,
      "learning_rate": 7.356200615554931e-06,
      "loss": 1.0581,
      "step": 64950
    },
    {
      "epoch": 2.6320374384635645,
      "grad_norm": 0.6524321436882019,
      "learning_rate": 7.3402251262700285e-06,
      "loss": 1.0758,
      "step": 64960
    },
    {
      "epoch": 2.6324426166406676,
      "grad_norm": 0.6970962882041931,
      "learning_rate": 7.3242663419252896e-06,
      "loss": 1.0558,
      "step": 64970
    },
    {
      "epoch": 2.632847794817771,
      "grad_norm": 0.5899817943572998,
      "learning_rate": 7.308324265397836e-06,
      "loss": 1.0721,
      "step": 64980
    },
    {
      "epoch": 2.6332529729948746,
      "grad_norm": 0.6159383058547974,
      "learning_rate": 7.292398899561725e-06,
      "loss": 1.0325,
      "step": 64990
    },
    {
      "epoch": 2.6336581511719777,
      "grad_norm": 0.6719132661819458,
      "learning_rate": 7.276490247288037e-06,
      "loss": 1.0315,
      "step": 65000
    },
    {
      "epoch": 2.634063329349081,
      "grad_norm": 0.7300938963890076,
      "learning_rate": 7.260598311444822e-06,
      "loss": 1.056,
      "step": 65010
    },
    {
      "epoch": 2.6344685075261847,
      "grad_norm": 0.8211233615875244,
      "learning_rate": 7.244723094897121e-06,
      "loss": 1.0546,
      "step": 65020
    },
    {
      "epoch": 2.634873685703288,
      "grad_norm": 0.6553299427032471,
      "learning_rate": 7.2288646005069575e-06,
      "loss": 1.0846,
      "step": 65030
    },
    {
      "epoch": 2.6352788638803917,
      "grad_norm": 0.6653985977172852,
      "learning_rate": 7.213022831133342e-06,
      "loss": 1.0554,
      "step": 65040
    },
    {
      "epoch": 2.6356840420574947,
      "grad_norm": 0.6012342572212219,
      "learning_rate": 7.197197789632248e-06,
      "loss": 1.0277,
      "step": 65050
    },
    {
      "epoch": 2.6360892202345982,
      "grad_norm": 0.5777649879455566,
      "learning_rate": 7.181389478856681e-06,
      "loss": 1.0415,
      "step": 65060
    },
    {
      "epoch": 2.6364943984117017,
      "grad_norm": 0.7567587494850159,
      "learning_rate": 7.165597901656596e-06,
      "loss": 1.0631,
      "step": 65070
    },
    {
      "epoch": 2.636899576588805,
      "grad_norm": 0.7103394865989685,
      "learning_rate": 7.1498230608789465e-06,
      "loss": 0.9924,
      "step": 65080
    },
    {
      "epoch": 2.6373047547659083,
      "grad_norm": 0.6149234175682068,
      "learning_rate": 7.1340649593676365e-06,
      "loss": 1.0429,
      "step": 65090
    },
    {
      "epoch": 2.637709932943012,
      "grad_norm": 0.6325362324714661,
      "learning_rate": 7.118323599963595e-06,
      "loss": 1.0645,
      "step": 65100
    },
    {
      "epoch": 2.638115111120115,
      "grad_norm": 0.6262956857681274,
      "learning_rate": 7.102598985504705e-06,
      "loss": 1.0189,
      "step": 65110
    },
    {
      "epoch": 2.6385202892972184,
      "grad_norm": 0.6298450231552124,
      "learning_rate": 7.0868911188258665e-06,
      "loss": 1.0675,
      "step": 65120
    },
    {
      "epoch": 2.638925467474322,
      "grad_norm": 0.6284121870994568,
      "learning_rate": 7.071200002758904e-06,
      "loss": 1.0385,
      "step": 65130
    },
    {
      "epoch": 2.639330645651425,
      "grad_norm": 0.6881783604621887,
      "learning_rate": 7.055525640132654e-06,
      "loss": 1.0368,
      "step": 65140
    },
    {
      "epoch": 2.6397358238285284,
      "grad_norm": 0.6276208758354187,
      "learning_rate": 7.039868033772956e-06,
      "loss": 1.0321,
      "step": 65150
    },
    {
      "epoch": 2.640141002005632,
      "grad_norm": 0.658777117729187,
      "learning_rate": 7.024227186502574e-06,
      "loss": 1.047,
      "step": 65160
    },
    {
      "epoch": 2.6405461801827355,
      "grad_norm": 0.5930315256118774,
      "learning_rate": 7.008603101141287e-06,
      "loss": 1.0613,
      "step": 65170
    },
    {
      "epoch": 2.640951358359839,
      "grad_norm": 0.7628633975982666,
      "learning_rate": 6.99299578050584e-06,
      "loss": 1.0736,
      "step": 65180
    },
    {
      "epoch": 2.641356536536942,
      "grad_norm": 0.5257205963134766,
      "learning_rate": 6.977405227409994e-06,
      "loss": 1.0266,
      "step": 65190
    },
    {
      "epoch": 2.6417617147140455,
      "grad_norm": 0.6445432901382446,
      "learning_rate": 6.961831444664413e-06,
      "loss": 1.0544,
      "step": 65200
    },
    {
      "epoch": 2.642166892891149,
      "grad_norm": 0.7142376899719238,
      "learning_rate": 6.946274435076816e-06,
      "loss": 1.0421,
      "step": 65210
    },
    {
      "epoch": 2.642572071068252,
      "grad_norm": 0.6976451277732849,
      "learning_rate": 6.930734201451816e-06,
      "loss": 1.0126,
      "step": 65220
    },
    {
      "epoch": 2.6429772492453556,
      "grad_norm": 0.5936594009399414,
      "learning_rate": 6.915210746591072e-06,
      "loss": 1.0242,
      "step": 65230
    },
    {
      "epoch": 2.643382427422459,
      "grad_norm": 0.618010938167572,
      "learning_rate": 6.8997040732931785e-06,
      "loss": 1.0326,
      "step": 65240
    },
    {
      "epoch": 2.643787605599562,
      "grad_norm": 0.599492609500885,
      "learning_rate": 6.884214184353755e-06,
      "loss": 1.0267,
      "step": 65250
    },
    {
      "epoch": 2.6441927837766657,
      "grad_norm": 0.7091574668884277,
      "learning_rate": 6.868741082565311e-06,
      "loss": 1.009,
      "step": 65260
    },
    {
      "epoch": 2.644597961953769,
      "grad_norm": 0.5891917943954468,
      "learning_rate": 6.853284770717416e-06,
      "loss": 1.0219,
      "step": 65270
    },
    {
      "epoch": 2.6450031401308727,
      "grad_norm": 0.670443058013916,
      "learning_rate": 6.8378452515965195e-06,
      "loss": 1.0644,
      "step": 65280
    },
    {
      "epoch": 2.645408318307976,
      "grad_norm": 0.6814293265342712,
      "learning_rate": 6.822422527986161e-06,
      "loss": 1.0473,
      "step": 65290
    },
    {
      "epoch": 2.6458134964850792,
      "grad_norm": 0.6541982889175415,
      "learning_rate": 6.807016602666749e-06,
      "loss": 1.072,
      "step": 65300
    },
    {
      "epoch": 2.6462186746621827,
      "grad_norm": 0.6670882701873779,
      "learning_rate": 6.791627478415719e-06,
      "loss": 1.0473,
      "step": 65310
    },
    {
      "epoch": 2.6466238528392863,
      "grad_norm": 0.7101354002952576,
      "learning_rate": 6.77625515800745e-06,
      "loss": 1.0675,
      "step": 65320
    },
    {
      "epoch": 2.6470290310163893,
      "grad_norm": 0.6664820313453674,
      "learning_rate": 6.760899644213325e-06,
      "loss": 1.031,
      "step": 65330
    },
    {
      "epoch": 2.647434209193493,
      "grad_norm": 0.6574360728263855,
      "learning_rate": 6.74556093980162e-06,
      "loss": 1.0111,
      "step": 65340
    },
    {
      "epoch": 2.6478393873705963,
      "grad_norm": 0.6648122668266296,
      "learning_rate": 6.730239047537712e-06,
      "loss": 1.0287,
      "step": 65350
    },
    {
      "epoch": 2.6482445655476994,
      "grad_norm": 0.6398398280143738,
      "learning_rate": 6.714933970183812e-06,
      "loss": 1.0588,
      "step": 65360
    },
    {
      "epoch": 2.648649743724803,
      "grad_norm": 0.5966915488243103,
      "learning_rate": 6.6996457104992025e-06,
      "loss": 1.0049,
      "step": 65370
    },
    {
      "epoch": 2.6490549219019064,
      "grad_norm": 0.635956346988678,
      "learning_rate": 6.684374271240046e-06,
      "loss": 1.056,
      "step": 65380
    },
    {
      "epoch": 2.64946010007901,
      "grad_norm": 0.6801062226295471,
      "learning_rate": 6.6691196551595396e-06,
      "loss": 1.0359,
      "step": 65390
    },
    {
      "epoch": 2.6498652782561134,
      "grad_norm": 0.6016404032707214,
      "learning_rate": 6.653881865007816e-06,
      "loss": 1.0435,
      "step": 65400
    },
    {
      "epoch": 2.6502704564332165,
      "grad_norm": 0.7044928669929504,
      "learning_rate": 6.638660903532013e-06,
      "loss": 1.0692,
      "step": 65410
    },
    {
      "epoch": 2.65067563461032,
      "grad_norm": 0.6855720281600952,
      "learning_rate": 6.623456773476166e-06,
      "loss": 1.0519,
      "step": 65420
    },
    {
      "epoch": 2.6510808127874235,
      "grad_norm": 0.6460932493209839,
      "learning_rate": 6.60826947758132e-06,
      "loss": 1.03,
      "step": 65430
    },
    {
      "epoch": 2.6514859909645265,
      "grad_norm": 0.6444928646087646,
      "learning_rate": 6.593099018585513e-06,
      "loss": 1.0098,
      "step": 65440
    },
    {
      "epoch": 2.65189116914163,
      "grad_norm": 0.661777913570404,
      "learning_rate": 6.577945399223673e-06,
      "loss": 1.0167,
      "step": 65450
    },
    {
      "epoch": 2.6522963473187335,
      "grad_norm": 0.6765967607498169,
      "learning_rate": 6.562808622227745e-06,
      "loss": 1.054,
      "step": 65460
    },
    {
      "epoch": 2.6527015254958366,
      "grad_norm": 0.6636807918548584,
      "learning_rate": 6.5476886903266255e-06,
      "loss": 1.0245,
      "step": 65470
    },
    {
      "epoch": 2.65310670367294,
      "grad_norm": 0.620265007019043,
      "learning_rate": 6.5325856062462e-06,
      "loss": 1.0384,
      "step": 65480
    },
    {
      "epoch": 2.6535118818500436,
      "grad_norm": 0.6507800817489624,
      "learning_rate": 6.517499372709257e-06,
      "loss": 1.0581,
      "step": 65490
    },
    {
      "epoch": 2.6539170600271467,
      "grad_norm": 0.6452245116233826,
      "learning_rate": 6.5024299924355996e-06,
      "loss": 1.0471,
      "step": 65500
    },
    {
      "epoch": 2.65432223820425,
      "grad_norm": 0.7252616286277771,
      "learning_rate": 6.487377468141942e-06,
      "loss": 1.034,
      "step": 65510
    },
    {
      "epoch": 2.6547274163813537,
      "grad_norm": 0.5927180647850037,
      "learning_rate": 6.472341802542048e-06,
      "loss": 1.0536,
      "step": 65520
    },
    {
      "epoch": 2.655132594558457,
      "grad_norm": 0.6541041135787964,
      "learning_rate": 6.457322998346549e-06,
      "loss": 1.0362,
      "step": 65530
    },
    {
      "epoch": 2.6555377727355607,
      "grad_norm": 0.7640507817268372,
      "learning_rate": 6.44232105826309e-06,
      "loss": 1.0448,
      "step": 65540
    },
    {
      "epoch": 2.6559429509126637,
      "grad_norm": 0.5923418998718262,
      "learning_rate": 6.42733598499623e-06,
      "loss": 1.0131,
      "step": 65550
    },
    {
      "epoch": 2.6563481290897673,
      "grad_norm": 0.7157917618751526,
      "learning_rate": 6.41236778124753e-06,
      "loss": 1.0518,
      "step": 65560
    },
    {
      "epoch": 2.6567533072668708,
      "grad_norm": 0.640005886554718,
      "learning_rate": 6.39741644971551e-06,
      "loss": 1.0397,
      "step": 65570
    },
    {
      "epoch": 2.657158485443974,
      "grad_norm": 0.6886975765228271,
      "learning_rate": 6.382481993095635e-06,
      "loss": 1.028,
      "step": 65580
    },
    {
      "epoch": 2.6575636636210773,
      "grad_norm": 0.6717981100082397,
      "learning_rate": 6.367564414080307e-06,
      "loss": 1.0222,
      "step": 65590
    },
    {
      "epoch": 2.657968841798181,
      "grad_norm": 0.660165011882782,
      "learning_rate": 6.35266371535892e-06,
      "loss": 1.0373,
      "step": 65600
    },
    {
      "epoch": 2.658374019975284,
      "grad_norm": 0.5947567224502563,
      "learning_rate": 6.337779899617802e-06,
      "loss": 1.013,
      "step": 65610
    },
    {
      "epoch": 2.6587791981523874,
      "grad_norm": 0.5883741974830627,
      "learning_rate": 6.32291296954024e-06,
      "loss": 1.054,
      "step": 65620
    },
    {
      "epoch": 2.659184376329491,
      "grad_norm": 0.6612681150436401,
      "learning_rate": 6.308062927806491e-06,
      "loss": 1.0575,
      "step": 65630
    },
    {
      "epoch": 2.6595895545065944,
      "grad_norm": 0.7099812030792236,
      "learning_rate": 6.2932297770937785e-06,
      "loss": 1.0282,
      "step": 65640
    },
    {
      "epoch": 2.659994732683698,
      "grad_norm": 0.7445299029350281,
      "learning_rate": 6.278413520076221e-06,
      "loss": 1.0641,
      "step": 65650
    },
    {
      "epoch": 2.660399910860801,
      "grad_norm": 0.6229289770126343,
      "learning_rate": 6.263614159424958e-06,
      "loss": 1.0236,
      "step": 65660
    },
    {
      "epoch": 2.6608050890379045,
      "grad_norm": 0.6828339695930481,
      "learning_rate": 6.248831697808044e-06,
      "loss": 1.0718,
      "step": 65670
    },
    {
      "epoch": 2.661210267215008,
      "grad_norm": 0.599912703037262,
      "learning_rate": 6.234066137890493e-06,
      "loss": 1.081,
      "step": 65680
    },
    {
      "epoch": 2.661615445392111,
      "grad_norm": 0.6871984601020813,
      "learning_rate": 6.219317482334286e-06,
      "loss": 1.0459,
      "step": 65690
    },
    {
      "epoch": 2.6620206235692145,
      "grad_norm": 0.7077010869979858,
      "learning_rate": 6.204585733798374e-06,
      "loss": 1.0075,
      "step": 65700
    },
    {
      "epoch": 2.662425801746318,
      "grad_norm": 0.6861071586608887,
      "learning_rate": 6.189870894938587e-06,
      "loss": 1.0321,
      "step": 65710
    },
    {
      "epoch": 2.662830979923421,
      "grad_norm": 0.6313314437866211,
      "learning_rate": 6.175172968407783e-06,
      "loss": 1.0303,
      "step": 65720
    },
    {
      "epoch": 2.6632361581005246,
      "grad_norm": 0.6299399137496948,
      "learning_rate": 6.160491956855729e-06,
      "loss": 1.0335,
      "step": 65730
    },
    {
      "epoch": 2.663641336277628,
      "grad_norm": 0.614443302154541,
      "learning_rate": 6.145827862929177e-06,
      "loss": 1.027,
      "step": 65740
    },
    {
      "epoch": 2.6640465144547316,
      "grad_norm": 0.7040666341781616,
      "learning_rate": 6.131180689271776e-06,
      "loss": 1.0427,
      "step": 65750
    },
    {
      "epoch": 2.664451692631835,
      "grad_norm": 0.6023210883140564,
      "learning_rate": 6.116550438524171e-06,
      "loss": 1.001,
      "step": 65760
    },
    {
      "epoch": 2.664856870808938,
      "grad_norm": 0.8013294339179993,
      "learning_rate": 6.101937113323952e-06,
      "loss": 1.0646,
      "step": 65770
    },
    {
      "epoch": 2.6652620489860417,
      "grad_norm": 0.6472370028495789,
      "learning_rate": 6.087340716305623e-06,
      "loss": 1.0744,
      "step": 65780
    },
    {
      "epoch": 2.665667227163145,
      "grad_norm": 0.6240482330322266,
      "learning_rate": 6.072761250100667e-06,
      "loss": 1.0346,
      "step": 65790
    },
    {
      "epoch": 2.6660724053402483,
      "grad_norm": 0.6232341527938843,
      "learning_rate": 6.058198717337516e-06,
      "loss": 1.0341,
      "step": 65800
    },
    {
      "epoch": 2.6664775835173518,
      "grad_norm": 0.7262073755264282,
      "learning_rate": 6.043653120641557e-06,
      "loss": 1.0087,
      "step": 65810
    },
    {
      "epoch": 2.6668827616944553,
      "grad_norm": 0.7206834554672241,
      "learning_rate": 6.029124462635072e-06,
      "loss": 1.0562,
      "step": 65820
    },
    {
      "epoch": 2.6672879398715583,
      "grad_norm": 0.6887161135673523,
      "learning_rate": 6.014612745937343e-06,
      "loss": 1.0487,
      "step": 65830
    },
    {
      "epoch": 2.667693118048662,
      "grad_norm": 0.7064412236213684,
      "learning_rate": 6.000117973164565e-06,
      "loss": 1.0241,
      "step": 65840
    },
    {
      "epoch": 2.6680982962257653,
      "grad_norm": 0.7087885737419128,
      "learning_rate": 5.9856401469299054e-06,
      "loss": 1.0557,
      "step": 65850
    },
    {
      "epoch": 2.6685034744028684,
      "grad_norm": 0.6839293241500854,
      "learning_rate": 5.971179269843463e-06,
      "loss": 1.0524,
      "step": 65860
    },
    {
      "epoch": 2.668908652579972,
      "grad_norm": 0.6871784329414368,
      "learning_rate": 5.956735344512299e-06,
      "loss": 1.0484,
      "step": 65870
    },
    {
      "epoch": 2.6693138307570754,
      "grad_norm": 0.6715488433837891,
      "learning_rate": 5.942308373540362e-06,
      "loss": 1.0183,
      "step": 65880
    },
    {
      "epoch": 2.669719008934179,
      "grad_norm": 0.6156019568443298,
      "learning_rate": 5.9278983595286164e-06,
      "loss": 1.0515,
      "step": 65890
    },
    {
      "epoch": 2.6701241871112824,
      "grad_norm": 0.630997359752655,
      "learning_rate": 5.913505305074885e-06,
      "loss": 1.0408,
      "step": 65900
    },
    {
      "epoch": 2.6705293652883855,
      "grad_norm": 0.6676200032234192,
      "learning_rate": 5.899129212774057e-06,
      "loss": 1.0339,
      "step": 65910
    },
    {
      "epoch": 2.670934543465489,
      "grad_norm": 0.6519805788993835,
      "learning_rate": 5.88477008521785e-06,
      "loss": 1.0512,
      "step": 65920
    },
    {
      "epoch": 2.6713397216425925,
      "grad_norm": 0.6331145763397217,
      "learning_rate": 5.8704279249949815e-06,
      "loss": 1.0675,
      "step": 65930
    },
    {
      "epoch": 2.6717448998196955,
      "grad_norm": 0.6575809121131897,
      "learning_rate": 5.85610273469106e-06,
      "loss": 1.0465,
      "step": 65940
    },
    {
      "epoch": 2.672150077996799,
      "grad_norm": 0.6230703592300415,
      "learning_rate": 5.84179451688871e-06,
      "loss": 1.0158,
      "step": 65950
    },
    {
      "epoch": 2.6725552561739025,
      "grad_norm": 0.6456611156463623,
      "learning_rate": 5.827503274167389e-06,
      "loss": 1.0408,
      "step": 65960
    },
    {
      "epoch": 2.6729604343510056,
      "grad_norm": 0.6308913230895996,
      "learning_rate": 5.813229009103649e-06,
      "loss": 1.0565,
      "step": 65970
    },
    {
      "epoch": 2.673365612528109,
      "grad_norm": 0.6037288308143616,
      "learning_rate": 5.798971724270819e-06,
      "loss": 1.0415,
      "step": 65980
    },
    {
      "epoch": 2.6737707907052126,
      "grad_norm": 0.649295449256897,
      "learning_rate": 5.7847314222392755e-06,
      "loss": 1.0185,
      "step": 65990
    },
    {
      "epoch": 2.674175968882316,
      "grad_norm": 0.7145460844039917,
      "learning_rate": 5.770508105576278e-06,
      "loss": 1.0182,
      "step": 66000
    },
    {
      "epoch": 2.6745811470594196,
      "grad_norm": 0.6902170181274414,
      "learning_rate": 5.75630177684604e-06,
      "loss": 1.0493,
      "step": 66010
    },
    {
      "epoch": 2.6749863252365227,
      "grad_norm": 0.6736736297607422,
      "learning_rate": 5.742112438609726e-06,
      "loss": 1.0695,
      "step": 66020
    },
    {
      "epoch": 2.675391503413626,
      "grad_norm": 0.6110474467277527,
      "learning_rate": 5.727940093425444e-06,
      "loss": 1.0566,
      "step": 66030
    },
    {
      "epoch": 2.6757966815907297,
      "grad_norm": 0.6682555079460144,
      "learning_rate": 5.713784743848172e-06,
      "loss": 1.0235,
      "step": 66040
    },
    {
      "epoch": 2.6762018597678328,
      "grad_norm": 0.6701818108558655,
      "learning_rate": 5.6996463924299134e-06,
      "loss": 1.0712,
      "step": 66050
    },
    {
      "epoch": 2.6766070379449363,
      "grad_norm": 0.6798429489135742,
      "learning_rate": 5.6855250417195525e-06,
      "loss": 1.0592,
      "step": 66060
    },
    {
      "epoch": 2.6770122161220398,
      "grad_norm": 0.6558367013931274,
      "learning_rate": 5.6714206942629185e-06,
      "loss": 1.0425,
      "step": 66070
    },
    {
      "epoch": 2.677417394299143,
      "grad_norm": 0.7204040884971619,
      "learning_rate": 5.657333352602778e-06,
      "loss": 1.033,
      "step": 66080
    },
    {
      "epoch": 2.6778225724762463,
      "grad_norm": 0.573173463344574,
      "learning_rate": 5.643263019278844e-06,
      "loss": 1.0341,
      "step": 66090
    },
    {
      "epoch": 2.67822775065335,
      "grad_norm": 0.6122177243232727,
      "learning_rate": 5.629209696827753e-06,
      "loss": 1.0644,
      "step": 66100
    },
    {
      "epoch": 2.6786329288304533,
      "grad_norm": 0.6455460786819458,
      "learning_rate": 5.615173387783057e-06,
      "loss": 1.0833,
      "step": 66110
    },
    {
      "epoch": 2.679038107007557,
      "grad_norm": 0.8222653865814209,
      "learning_rate": 5.601154094675276e-06,
      "loss": 1.0674,
      "step": 66120
    },
    {
      "epoch": 2.67944328518466,
      "grad_norm": 0.6883975267410278,
      "learning_rate": 5.587151820031811e-06,
      "loss": 1.0412,
      "step": 66130
    },
    {
      "epoch": 2.6798484633617634,
      "grad_norm": 0.6949438452720642,
      "learning_rate": 5.573166566377086e-06,
      "loss": 1.0071,
      "step": 66140
    },
    {
      "epoch": 2.680253641538867,
      "grad_norm": 0.7525830864906311,
      "learning_rate": 5.559198336232341e-06,
      "loss": 1.03,
      "step": 66150
    },
    {
      "epoch": 2.68065881971597,
      "grad_norm": 0.6960757970809937,
      "learning_rate": 5.545247132115839e-06,
      "loss": 1.0064,
      "step": 66160
    },
    {
      "epoch": 2.6810639978930735,
      "grad_norm": 0.6672601103782654,
      "learning_rate": 5.531312956542723e-06,
      "loss": 1.035,
      "step": 66170
    },
    {
      "epoch": 2.681469176070177,
      "grad_norm": 0.6604568362236023,
      "learning_rate": 5.5173958120250725e-06,
      "loss": 1.0495,
      "step": 66180
    },
    {
      "epoch": 2.68187435424728,
      "grad_norm": 0.6941354870796204,
      "learning_rate": 5.503495701071926e-06,
      "loss": 1.0282,
      "step": 66190
    },
    {
      "epoch": 2.6822795324243836,
      "grad_norm": 0.6655781865119934,
      "learning_rate": 5.489612626189245e-06,
      "loss": 1.032,
      "step": 66200
    },
    {
      "epoch": 2.682684710601487,
      "grad_norm": 0.7044093012809753,
      "learning_rate": 5.47574658987986e-06,
      "loss": 1.0568,
      "step": 66210
    },
    {
      "epoch": 2.68308988877859,
      "grad_norm": 0.66450035572052,
      "learning_rate": 5.461897594643617e-06,
      "loss": 1.0833,
      "step": 66220
    },
    {
      "epoch": 2.6834950669556936,
      "grad_norm": 0.6407907605171204,
      "learning_rate": 5.448065642977218e-06,
      "loss": 1.0276,
      "step": 66230
    },
    {
      "epoch": 2.683900245132797,
      "grad_norm": 0.5988668203353882,
      "learning_rate": 5.434250737374324e-06,
      "loss": 1.0622,
      "step": 66240
    },
    {
      "epoch": 2.6843054233099006,
      "grad_norm": 0.649024486541748,
      "learning_rate": 5.420452880325544e-06,
      "loss": 1.0245,
      "step": 66250
    },
    {
      "epoch": 2.684710601487004,
      "grad_norm": 0.6162046790122986,
      "learning_rate": 5.4066720743183865e-06,
      "loss": 1.0566,
      "step": 66260
    },
    {
      "epoch": 2.685115779664107,
      "grad_norm": 0.6972870230674744,
      "learning_rate": 5.392908321837276e-06,
      "loss": 1.0385,
      "step": 66270
    },
    {
      "epoch": 2.6855209578412107,
      "grad_norm": 0.6430394649505615,
      "learning_rate": 5.379161625363593e-06,
      "loss": 1.0523,
      "step": 66280
    },
    {
      "epoch": 2.685926136018314,
      "grad_norm": 0.6262328624725342,
      "learning_rate": 5.365431987375602e-06,
      "loss": 1.0523,
      "step": 66290
    },
    {
      "epoch": 2.6863313141954173,
      "grad_norm": 0.6829304099082947,
      "learning_rate": 5.351719410348533e-06,
      "loss": 1.0412,
      "step": 66300
    },
    {
      "epoch": 2.6867364923725208,
      "grad_norm": 0.6154037117958069,
      "learning_rate": 5.338023896754529e-06,
      "loss": 1.0671,
      "step": 66310
    },
    {
      "epoch": 2.6871416705496243,
      "grad_norm": 0.5792384147644043,
      "learning_rate": 5.324345449062662e-06,
      "loss": 1.0387,
      "step": 66320
    },
    {
      "epoch": 2.6875468487267273,
      "grad_norm": 0.6974734663963318,
      "learning_rate": 5.310684069738881e-06,
      "loss": 1.0846,
      "step": 66330
    },
    {
      "epoch": 2.687952026903831,
      "grad_norm": 0.687290370464325,
      "learning_rate": 5.297039761246137e-06,
      "loss": 1.0625,
      "step": 66340
    },
    {
      "epoch": 2.6883572050809343,
      "grad_norm": 0.6020795702934265,
      "learning_rate": 5.283412526044207e-06,
      "loss": 1.0132,
      "step": 66350
    },
    {
      "epoch": 2.688762383258038,
      "grad_norm": 0.6055883765220642,
      "learning_rate": 5.2698023665899045e-06,
      "loss": 1.035,
      "step": 66360
    },
    {
      "epoch": 2.6891675614351414,
      "grad_norm": 0.506350040435791,
      "learning_rate": 5.256209285336866e-06,
      "loss": 1.0285,
      "step": 66370
    },
    {
      "epoch": 2.6895727396122444,
      "grad_norm": 0.6959723234176636,
      "learning_rate": 5.2426332847357096e-06,
      "loss": 1.0361,
      "step": 66380
    },
    {
      "epoch": 2.689977917789348,
      "grad_norm": 0.6567620038986206,
      "learning_rate": 5.229074367233921e-06,
      "loss": 1.0089,
      "step": 66390
    },
    {
      "epoch": 2.6903830959664514,
      "grad_norm": 0.7189040184020996,
      "learning_rate": 5.215532535275958e-06,
      "loss": 1.0597,
      "step": 66400
    },
    {
      "epoch": 2.6907882741435545,
      "grad_norm": 0.6959870457649231,
      "learning_rate": 5.202007791303165e-06,
      "loss": 1.047,
      "step": 66410
    },
    {
      "epoch": 2.691193452320658,
      "grad_norm": 0.7129644751548767,
      "learning_rate": 5.188500137753838e-06,
      "loss": 1.0986,
      "step": 66420
    },
    {
      "epoch": 2.6915986304977615,
      "grad_norm": 0.6493264436721802,
      "learning_rate": 5.1750095770631505e-06,
      "loss": 1.0471,
      "step": 66430
    },
    {
      "epoch": 2.6920038086748646,
      "grad_norm": 0.6645280718803406,
      "learning_rate": 5.161536111663212e-06,
      "loss": 1.0255,
      "step": 66440
    },
    {
      "epoch": 2.692408986851968,
      "grad_norm": 0.7266080379486084,
      "learning_rate": 5.148079743983081e-06,
      "loss": 1.0312,
      "step": 66450
    },
    {
      "epoch": 2.6928141650290716,
      "grad_norm": 0.6587215662002563,
      "learning_rate": 5.1346404764486825e-06,
      "loss": 1.0434,
      "step": 66460
    },
    {
      "epoch": 2.693219343206175,
      "grad_norm": 0.7853602766990662,
      "learning_rate": 5.1212183114828785e-06,
      "loss": 1.0741,
      "step": 66470
    },
    {
      "epoch": 2.6936245213832786,
      "grad_norm": 0.659125030040741,
      "learning_rate": 5.107813251505455e-06,
      "loss": 1.0144,
      "step": 66480
    },
    {
      "epoch": 2.6940296995603816,
      "grad_norm": 0.6175633072853088,
      "learning_rate": 5.094425298933136e-06,
      "loss": 1.0565,
      "step": 66490
    },
    {
      "epoch": 2.694434877737485,
      "grad_norm": 0.6446332335472107,
      "learning_rate": 5.081054456179502e-06,
      "loss": 1.0503,
      "step": 66500
    },
    {
      "epoch": 2.6948400559145886,
      "grad_norm": 0.6175461411476135,
      "learning_rate": 5.067700725655111e-06,
      "loss": 1.0745,
      "step": 66510
    },
    {
      "epoch": 2.6952452340916917,
      "grad_norm": 0.5953205823898315,
      "learning_rate": 5.054364109767362e-06,
      "loss": 1.0321,
      "step": 66520
    },
    {
      "epoch": 2.695650412268795,
      "grad_norm": 0.6995067000389099,
      "learning_rate": 5.041044610920664e-06,
      "loss": 1.0439,
      "step": 66530
    },
    {
      "epoch": 2.6960555904458987,
      "grad_norm": 0.7024220824241638,
      "learning_rate": 5.027742231516264e-06,
      "loss": 1.0395,
      "step": 66540
    },
    {
      "epoch": 2.6964607686230018,
      "grad_norm": 0.6799652576446533,
      "learning_rate": 5.014456973952375e-06,
      "loss": 1.0255,
      "step": 66550
    },
    {
      "epoch": 2.6968659468001053,
      "grad_norm": 0.729877233505249,
      "learning_rate": 5.00118884062405e-06,
      "loss": 1.0394,
      "step": 66560
    },
    {
      "epoch": 2.697271124977209,
      "grad_norm": 0.7372061014175415,
      "learning_rate": 4.987937833923351e-06,
      "loss": 1.0367,
      "step": 66570
    },
    {
      "epoch": 2.697676303154312,
      "grad_norm": 0.7451594471931458,
      "learning_rate": 4.9747039562391466e-06,
      "loss": 1.0468,
      "step": 66580
    },
    {
      "epoch": 2.6980814813314153,
      "grad_norm": 0.6229259371757507,
      "learning_rate": 4.961487209957349e-06,
      "loss": 1.037,
      "step": 66590
    },
    {
      "epoch": 2.698486659508519,
      "grad_norm": 0.7005883455276489,
      "learning_rate": 4.948287597460643e-06,
      "loss": 1.0521,
      "step": 66600
    },
    {
      "epoch": 2.6988918376856224,
      "grad_norm": 0.707958459854126,
      "learning_rate": 4.935105121128725e-06,
      "loss": 1.0539,
      "step": 66610
    },
    {
      "epoch": 2.699297015862726,
      "grad_norm": 0.6686216592788696,
      "learning_rate": 4.921939783338137e-06,
      "loss": 1.0542,
      "step": 66620
    },
    {
      "epoch": 2.699702194039829,
      "grad_norm": 0.6808463335037231,
      "learning_rate": 4.90879158646238e-06,
      "loss": 1.0355,
      "step": 66630
    },
    {
      "epoch": 2.7001073722169324,
      "grad_norm": 0.6651437878608704,
      "learning_rate": 4.895660532871837e-06,
      "loss": 1.0486,
      "step": 66640
    },
    {
      "epoch": 2.700512550394036,
      "grad_norm": 0.6869820952415466,
      "learning_rate": 4.882546624933826e-06,
      "loss": 1.0642,
      "step": 66650
    },
    {
      "epoch": 2.700917728571139,
      "grad_norm": 0.7676206231117249,
      "learning_rate": 4.869449865012521e-06,
      "loss": 1.0445,
      "step": 66660
    },
    {
      "epoch": 2.7013229067482425,
      "grad_norm": 0.6602201461791992,
      "learning_rate": 4.856370255469078e-06,
      "loss": 1.0316,
      "step": 66670
    },
    {
      "epoch": 2.701728084925346,
      "grad_norm": 0.6641952395439148,
      "learning_rate": 4.843307798661489e-06,
      "loss": 1.0386,
      "step": 66680
    },
    {
      "epoch": 2.702133263102449,
      "grad_norm": 0.7012605667114258,
      "learning_rate": 4.830262496944693e-06,
      "loss": 1.0522,
      "step": 66690
    },
    {
      "epoch": 2.7025384412795526,
      "grad_norm": 0.6613057851791382,
      "learning_rate": 4.817234352670541e-06,
      "loss": 1.0391,
      "step": 66700
    },
    {
      "epoch": 2.702943619456656,
      "grad_norm": 0.6813157200813293,
      "learning_rate": 4.804223368187788e-06,
      "loss": 1.0195,
      "step": 66710
    },
    {
      "epoch": 2.7033487976337596,
      "grad_norm": 0.6018661260604858,
      "learning_rate": 4.791229545842068e-06,
      "loss": 1.0168,
      "step": 66720
    },
    {
      "epoch": 2.703753975810863,
      "grad_norm": 0.6393193602561951,
      "learning_rate": 4.778252887975942e-06,
      "loss": 1.0362,
      "step": 66730
    },
    {
      "epoch": 2.704159153987966,
      "grad_norm": 0.7010689377784729,
      "learning_rate": 4.765293396928905e-06,
      "loss": 1.0746,
      "step": 66740
    },
    {
      "epoch": 2.7045643321650696,
      "grad_norm": 0.7537323236465454,
      "learning_rate": 4.7523510750372755e-06,
      "loss": 1.0407,
      "step": 66750
    },
    {
      "epoch": 2.704969510342173,
      "grad_norm": 0.58720463514328,
      "learning_rate": 4.739425924634366e-06,
      "loss": 1.0447,
      "step": 66760
    },
    {
      "epoch": 2.705374688519276,
      "grad_norm": 0.5998979210853577,
      "learning_rate": 4.726517948050347e-06,
      "loss": 1.0676,
      "step": 66770
    },
    {
      "epoch": 2.7057798666963797,
      "grad_norm": 0.5920282006263733,
      "learning_rate": 4.7136271476123095e-06,
      "loss": 1.073,
      "step": 66780
    },
    {
      "epoch": 2.706185044873483,
      "grad_norm": 0.6740338802337646,
      "learning_rate": 4.700753525644219e-06,
      "loss": 1.0079,
      "step": 66790
    },
    {
      "epoch": 2.7065902230505863,
      "grad_norm": 0.6261231303215027,
      "learning_rate": 4.687897084466975e-06,
      "loss": 1.0068,
      "step": 66800
    },
    {
      "epoch": 2.70699540122769,
      "grad_norm": 0.7150638699531555,
      "learning_rate": 4.675057826398366e-06,
      "loss": 1.0525,
      "step": 66810
    },
    {
      "epoch": 2.7074005794047933,
      "grad_norm": 0.7257791757583618,
      "learning_rate": 4.6622357537531084e-06,
      "loss": 1.0517,
      "step": 66820
    },
    {
      "epoch": 2.707805757581897,
      "grad_norm": 0.6070414781570435,
      "learning_rate": 4.6494308688427635e-06,
      "loss": 1.0413,
      "step": 66830
    },
    {
      "epoch": 2.7082109357590003,
      "grad_norm": 0.7082441449165344,
      "learning_rate": 4.636643173975863e-06,
      "loss": 1.0444,
      "step": 66840
    },
    {
      "epoch": 2.7086161139361034,
      "grad_norm": 0.6144518852233887,
      "learning_rate": 4.623872671457774e-06,
      "loss": 1.0897,
      "step": 66850
    },
    {
      "epoch": 2.709021292113207,
      "grad_norm": 0.6007370948791504,
      "learning_rate": 4.61111936359081e-06,
      "loss": 1.0262,
      "step": 66860
    },
    {
      "epoch": 2.7094264702903104,
      "grad_norm": 0.7250626087188721,
      "learning_rate": 4.598383252674165e-06,
      "loss": 1.0939,
      "step": 66870
    },
    {
      "epoch": 2.7098316484674134,
      "grad_norm": 0.6852951645851135,
      "learning_rate": 4.585664341003948e-06,
      "loss": 1.0554,
      "step": 66880
    },
    {
      "epoch": 2.710236826644517,
      "grad_norm": 0.7064920663833618,
      "learning_rate": 4.5729626308731455e-06,
      "loss": 1.0727,
      "step": 66890
    },
    {
      "epoch": 2.7106420048216204,
      "grad_norm": 0.7586604952812195,
      "learning_rate": 4.560278124571671e-06,
      "loss": 1.0104,
      "step": 66900
    },
    {
      "epoch": 2.7110471829987235,
      "grad_norm": 0.6387265920639038,
      "learning_rate": 4.547610824386295e-06,
      "loss": 1.0417,
      "step": 66910
    },
    {
      "epoch": 2.711452361175827,
      "grad_norm": 0.731669008731842,
      "learning_rate": 4.5349607326007234e-06,
      "loss": 1.0251,
      "step": 66920
    },
    {
      "epoch": 2.7118575393529305,
      "grad_norm": 0.6375449299812317,
      "learning_rate": 4.522327851495545e-06,
      "loss": 1.0248,
      "step": 66930
    },
    {
      "epoch": 2.7122627175300336,
      "grad_norm": 0.7223088145256042,
      "learning_rate": 4.5097121833482585e-06,
      "loss": 1.0476,
      "step": 66940
    },
    {
      "epoch": 2.712667895707137,
      "grad_norm": 0.6839558482170105,
      "learning_rate": 4.497113730433234e-06,
      "loss": 1.037,
      "step": 66950
    },
    {
      "epoch": 2.7130730738842406,
      "grad_norm": 0.5944827795028687,
      "learning_rate": 4.484532495021765e-06,
      "loss": 1.0434,
      "step": 66960
    },
    {
      "epoch": 2.713478252061344,
      "grad_norm": 0.7026113271713257,
      "learning_rate": 4.4719684793819935e-06,
      "loss": 1.0221,
      "step": 66970
    },
    {
      "epoch": 2.7138834302384476,
      "grad_norm": 0.6223257780075073,
      "learning_rate": 4.4594216857790524e-06,
      "loss": 1.0778,
      "step": 66980
    },
    {
      "epoch": 2.7142886084155506,
      "grad_norm": 0.707642674446106,
      "learning_rate": 4.446892116474866e-06,
      "loss": 1.0674,
      "step": 66990
    },
    {
      "epoch": 2.714693786592654,
      "grad_norm": 0.6836246252059937,
      "learning_rate": 4.434379773728314e-06,
      "loss": 1.0521,
      "step": 67000
    },
    {
      "epoch": 2.7150989647697577,
      "grad_norm": 0.7846552133560181,
      "learning_rate": 4.421884659795139e-06,
      "loss": 1.027,
      "step": 67010
    },
    {
      "epoch": 2.7155041429468607,
      "grad_norm": 0.7537200450897217,
      "learning_rate": 4.409406776928005e-06,
      "loss": 1.0519,
      "step": 67020
    },
    {
      "epoch": 2.715909321123964,
      "grad_norm": 0.6757401823997498,
      "learning_rate": 4.396946127376445e-06,
      "loss": 1.1311,
      "step": 67030
    },
    {
      "epoch": 2.7163144993010677,
      "grad_norm": 0.6449024677276611,
      "learning_rate": 4.384502713386918e-06,
      "loss": 1.0166,
      "step": 67040
    },
    {
      "epoch": 2.716719677478171,
      "grad_norm": 0.684986412525177,
      "learning_rate": 4.37207653720273e-06,
      "loss": 1.0295,
      "step": 67050
    },
    {
      "epoch": 2.7171248556552743,
      "grad_norm": 0.7085080146789551,
      "learning_rate": 4.359667601064122e-06,
      "loss": 1.0536,
      "step": 67060
    },
    {
      "epoch": 2.717530033832378,
      "grad_norm": 0.638027012348175,
      "learning_rate": 4.347275907208204e-06,
      "loss": 1.047,
      "step": 67070
    },
    {
      "epoch": 2.7179352120094813,
      "grad_norm": 0.6686566472053528,
      "learning_rate": 4.334901457868967e-06,
      "loss": 1.0543,
      "step": 67080
    },
    {
      "epoch": 2.718340390186585,
      "grad_norm": 0.7276633977890015,
      "learning_rate": 4.322544255277317e-06,
      "loss": 1.0714,
      "step": 67090
    },
    {
      "epoch": 2.718745568363688,
      "grad_norm": 0.6984259486198425,
      "learning_rate": 4.3102043016610475e-06,
      "loss": 1.0255,
      "step": 67100
    },
    {
      "epoch": 2.7191507465407914,
      "grad_norm": 0.5963259935379028,
      "learning_rate": 4.2978815992448576e-06,
      "loss": 1.0238,
      "step": 67110
    },
    {
      "epoch": 2.719555924717895,
      "grad_norm": 0.6529956459999084,
      "learning_rate": 4.28557615025027e-06,
      "loss": 1.07,
      "step": 67120
    },
    {
      "epoch": 2.719961102894998,
      "grad_norm": 0.6592189073562622,
      "learning_rate": 4.273287956895777e-06,
      "loss": 1.0241,
      "step": 67130
    },
    {
      "epoch": 2.7203662810721014,
      "grad_norm": 0.6308000087738037,
      "learning_rate": 4.261017021396685e-06,
      "loss": 1.0765,
      "step": 67140
    },
    {
      "epoch": 2.720771459249205,
      "grad_norm": 0.7041938900947571,
      "learning_rate": 4.24876334596529e-06,
      "loss": 0.9971,
      "step": 67150
    },
    {
      "epoch": 2.721176637426308,
      "grad_norm": 0.6363931894302368,
      "learning_rate": 4.236526932810659e-06,
      "loss": 1.077,
      "step": 67160
    },
    {
      "epoch": 2.7215818156034115,
      "grad_norm": 0.6264490485191345,
      "learning_rate": 4.224307784138848e-06,
      "loss": 1.0396,
      "step": 67170
    },
    {
      "epoch": 2.721986993780515,
      "grad_norm": 0.608869731426239,
      "learning_rate": 4.21210590215273e-06,
      "loss": 1.1015,
      "step": 67180
    },
    {
      "epoch": 2.7223921719576185,
      "grad_norm": 0.625548243522644,
      "learning_rate": 4.1999212890521e-06,
      "loss": 1.079,
      "step": 67190
    },
    {
      "epoch": 2.722797350134722,
      "grad_norm": 0.7099867463111877,
      "learning_rate": 4.187753947033613e-06,
      "loss": 1.04,
      "step": 67200
    },
    {
      "epoch": 2.723202528311825,
      "grad_norm": 0.6100059747695923,
      "learning_rate": 4.175603878290857e-06,
      "loss": 1.0841,
      "step": 67210
    },
    {
      "epoch": 2.7236077064889286,
      "grad_norm": 0.6267779469490051,
      "learning_rate": 4.163471085014259e-06,
      "loss": 1.0281,
      "step": 67220
    },
    {
      "epoch": 2.724012884666032,
      "grad_norm": 0.6180202960968018,
      "learning_rate": 4.151355569391169e-06,
      "loss": 1.054,
      "step": 67230
    },
    {
      "epoch": 2.724418062843135,
      "grad_norm": 0.6676654815673828,
      "learning_rate": 4.139257333605784e-06,
      "loss": 1.0292,
      "step": 67240
    },
    {
      "epoch": 2.7248232410202387,
      "grad_norm": 0.6255014538764954,
      "learning_rate": 4.127176379839193e-06,
      "loss": 1.0723,
      "step": 67250
    },
    {
      "epoch": 2.725228419197342,
      "grad_norm": 0.6802026629447937,
      "learning_rate": 4.1151127102694085e-06,
      "loss": 1.0709,
      "step": 67260
    },
    {
      "epoch": 2.725633597374445,
      "grad_norm": 0.6473041772842407,
      "learning_rate": 4.103066327071303e-06,
      "loss": 1.0304,
      "step": 67270
    },
    {
      "epoch": 2.7260387755515487,
      "grad_norm": 0.6989057064056396,
      "learning_rate": 4.091037232416605e-06,
      "loss": 1.0538,
      "step": 67280
    },
    {
      "epoch": 2.7264439537286522,
      "grad_norm": 0.6305609941482544,
      "learning_rate": 4.0790254284739596e-06,
      "loss": 1.0459,
      "step": 67290
    },
    {
      "epoch": 2.7268491319057553,
      "grad_norm": 0.688043475151062,
      "learning_rate": 4.067030917408887e-06,
      "loss": 1.0679,
      "step": 67300
    },
    {
      "epoch": 2.727254310082859,
      "grad_norm": 0.6570233106613159,
      "learning_rate": 4.055053701383782e-06,
      "loss": 1.0305,
      "step": 67310
    },
    {
      "epoch": 2.7276594882599623,
      "grad_norm": 0.7097117900848389,
      "learning_rate": 4.043093782557939e-06,
      "loss": 1.0178,
      "step": 67320
    },
    {
      "epoch": 2.728064666437066,
      "grad_norm": 0.6307719945907593,
      "learning_rate": 4.031151163087521e-06,
      "loss": 1.07,
      "step": 67330
    },
    {
      "epoch": 2.7284698446141693,
      "grad_norm": 0.6955611705780029,
      "learning_rate": 4.0192258451255625e-06,
      "loss": 1.0325,
      "step": 67340
    },
    {
      "epoch": 2.7288750227912724,
      "grad_norm": 0.6827651858329773,
      "learning_rate": 4.00731783082201e-06,
      "loss": 1.0446,
      "step": 67350
    },
    {
      "epoch": 2.729280200968376,
      "grad_norm": 0.65016770362854,
      "learning_rate": 3.995427122323647e-06,
      "loss": 1.0677,
      "step": 67360
    },
    {
      "epoch": 2.7296853791454794,
      "grad_norm": 0.6599376797676086,
      "learning_rate": 3.983553721774169e-06,
      "loss": 1.0345,
      "step": 67370
    },
    {
      "epoch": 2.7300905573225824,
      "grad_norm": 0.6516991853713989,
      "learning_rate": 3.971697631314142e-06,
      "loss": 1.0594,
      "step": 67380
    },
    {
      "epoch": 2.730495735499686,
      "grad_norm": 0.6584638953208923,
      "learning_rate": 3.959858853081033e-06,
      "loss": 1.0368,
      "step": 67390
    },
    {
      "epoch": 2.7309009136767894,
      "grad_norm": 0.6251571774482727,
      "learning_rate": 3.948037389209136e-06,
      "loss": 1.0356,
      "step": 67400
    },
    {
      "epoch": 2.7313060918538925,
      "grad_norm": 0.6709654927253723,
      "learning_rate": 3.936233241829667e-06,
      "loss": 1.0547,
      "step": 67410
    },
    {
      "epoch": 2.731711270030996,
      "grad_norm": 0.6091697216033936,
      "learning_rate": 3.924446413070715e-06,
      "loss": 1.0558,
      "step": 67420
    },
    {
      "epoch": 2.7321164482080995,
      "grad_norm": 0.6689293384552002,
      "learning_rate": 3.912676905057244e-06,
      "loss": 1.0485,
      "step": 67430
    },
    {
      "epoch": 2.732521626385203,
      "grad_norm": 0.6800949573516846,
      "learning_rate": 3.90092471991107e-06,
      "loss": 1.0457,
      "step": 67440
    },
    {
      "epoch": 2.7329268045623065,
      "grad_norm": 0.8280810713768005,
      "learning_rate": 3.889189859750919e-06,
      "loss": 1.0493,
      "step": 67450
    },
    {
      "epoch": 2.7333319827394096,
      "grad_norm": 0.6356002688407898,
      "learning_rate": 3.877472326692389e-06,
      "loss": 1.0546,
      "step": 67460
    },
    {
      "epoch": 2.733737160916513,
      "grad_norm": 0.7119672894477844,
      "learning_rate": 3.865772122847921e-06,
      "loss": 1.0068,
      "step": 67470
    },
    {
      "epoch": 2.7341423390936166,
      "grad_norm": 0.6917499899864197,
      "learning_rate": 3.854089250326876e-06,
      "loss": 1.0279,
      "step": 67480
    },
    {
      "epoch": 2.7345475172707197,
      "grad_norm": 0.765984833240509,
      "learning_rate": 3.842423711235477e-06,
      "loss": 1.0061,
      "step": 67490
    },
    {
      "epoch": 2.734952695447823,
      "grad_norm": 0.6268441081047058,
      "learning_rate": 3.83077550767682e-06,
      "loss": 1.0419,
      "step": 67500
    },
    {
      "epoch": 2.7353578736249267,
      "grad_norm": 0.6561225652694702,
      "learning_rate": 3.819144641750849e-06,
      "loss": 1.0617,
      "step": 67510
    },
    {
      "epoch": 2.7357630518020297,
      "grad_norm": 0.7512331008911133,
      "learning_rate": 3.8075311155544304e-06,
      "loss": 1.0792,
      "step": 67520
    },
    {
      "epoch": 2.7361682299791332,
      "grad_norm": 0.6323046684265137,
      "learning_rate": 3.7959349311812665e-06,
      "loss": 1.0374,
      "step": 67530
    },
    {
      "epoch": 2.7365734081562367,
      "grad_norm": 0.5921147465705872,
      "learning_rate": 3.784356090721941e-06,
      "loss": 1.0302,
      "step": 67540
    },
    {
      "epoch": 2.7369785863333402,
      "grad_norm": 0.6422343850135803,
      "learning_rate": 3.7727945962639287e-06,
      "loss": 1.0411,
      "step": 67550
    },
    {
      "epoch": 2.7373837645104437,
      "grad_norm": 0.6591508388519287,
      "learning_rate": 3.7612504498915734e-06,
      "loss": 1.0489,
      "step": 67560
    },
    {
      "epoch": 2.737788942687547,
      "grad_norm": 0.7512255311012268,
      "learning_rate": 3.7497236536860658e-06,
      "loss": 1.0497,
      "step": 67570
    },
    {
      "epoch": 2.7381941208646503,
      "grad_norm": 0.6059693694114685,
      "learning_rate": 3.7382142097255102e-06,
      "loss": 1.052,
      "step": 67580
    },
    {
      "epoch": 2.738599299041754,
      "grad_norm": 0.6192132830619812,
      "learning_rate": 3.7267221200848134e-06,
      "loss": 1.0536,
      "step": 67590
    },
    {
      "epoch": 2.739004477218857,
      "grad_norm": 0.699518084526062,
      "learning_rate": 3.7152473868358404e-06,
      "loss": 0.9846,
      "step": 67600
    },
    {
      "epoch": 2.7394096553959604,
      "grad_norm": 0.6421318650245667,
      "learning_rate": 3.70379001204727e-06,
      "loss": 1.0387,
      "step": 67610
    },
    {
      "epoch": 2.739814833573064,
      "grad_norm": 0.6858587861061096,
      "learning_rate": 3.692349997784672e-06,
      "loss": 1.0261,
      "step": 67620
    },
    {
      "epoch": 2.740220011750167,
      "grad_norm": 0.656526505947113,
      "learning_rate": 3.6809273461104744e-06,
      "loss": 1.0861,
      "step": 67630
    },
    {
      "epoch": 2.7406251899272704,
      "grad_norm": 0.680861234664917,
      "learning_rate": 3.6695220590839963e-06,
      "loss": 1.0673,
      "step": 67640
    },
    {
      "epoch": 2.741030368104374,
      "grad_norm": 0.6942551732063293,
      "learning_rate": 3.658134138761371e-06,
      "loss": 1.0419,
      "step": 67650
    },
    {
      "epoch": 2.741435546281477,
      "grad_norm": 0.663364589214325,
      "learning_rate": 3.6467635871957007e-06,
      "loss": 1.0386,
      "step": 67660
    },
    {
      "epoch": 2.7418407244585805,
      "grad_norm": 0.6744933128356934,
      "learning_rate": 3.6354104064368566e-06,
      "loss": 1.0448,
      "step": 67670
    },
    {
      "epoch": 2.742245902635684,
      "grad_norm": 0.662519097328186,
      "learning_rate": 3.624074598531635e-06,
      "loss": 1.0332,
      "step": 67680
    },
    {
      "epoch": 2.7426510808127875,
      "grad_norm": 0.669949471950531,
      "learning_rate": 3.6127561655236786e-06,
      "loss": 1.0226,
      "step": 67690
    },
    {
      "epoch": 2.743056258989891,
      "grad_norm": 0.6850124597549438,
      "learning_rate": 3.6014551094535e-06,
      "loss": 1.0789,
      "step": 67700
    },
    {
      "epoch": 2.743461437166994,
      "grad_norm": 0.6848585605621338,
      "learning_rate": 3.590171432358491e-06,
      "loss": 1.0123,
      "step": 67710
    },
    {
      "epoch": 2.7438666153440976,
      "grad_norm": 0.69454425573349,
      "learning_rate": 3.578905136272903e-06,
      "loss": 1.0711,
      "step": 67720
    },
    {
      "epoch": 2.744271793521201,
      "grad_norm": 0.6316027641296387,
      "learning_rate": 3.567656223227833e-06,
      "loss": 1.0326,
      "step": 67730
    },
    {
      "epoch": 2.744676971698304,
      "grad_norm": 0.6289641261100769,
      "learning_rate": 3.5564246952512815e-06,
      "loss": 1.0702,
      "step": 67740
    },
    {
      "epoch": 2.7450821498754077,
      "grad_norm": 0.631451427936554,
      "learning_rate": 3.5452105543681056e-06,
      "loss": 1.0352,
      "step": 67750
    },
    {
      "epoch": 2.745487328052511,
      "grad_norm": 0.6449522376060486,
      "learning_rate": 3.5340138026000003e-06,
      "loss": 1.0284,
      "step": 67760
    },
    {
      "epoch": 2.7458925062296142,
      "grad_norm": 0.7029479146003723,
      "learning_rate": 3.52283444196555e-06,
      "loss": 1.0641,
      "step": 67770
    },
    {
      "epoch": 2.7462976844067177,
      "grad_norm": 0.6681315302848816,
      "learning_rate": 3.5116724744801987e-06,
      "loss": 1.018,
      "step": 67780
    },
    {
      "epoch": 2.7467028625838212,
      "grad_norm": 0.6388664841651917,
      "learning_rate": 3.5005279021562585e-06,
      "loss": 1.0665,
      "step": 67790
    },
    {
      "epoch": 2.7471080407609247,
      "grad_norm": 0.702818751335144,
      "learning_rate": 3.4894007270028894e-06,
      "loss": 1.0928,
      "step": 67800
    },
    {
      "epoch": 2.7475132189380282,
      "grad_norm": 0.593262255191803,
      "learning_rate": 3.478290951026153e-06,
      "loss": 1.0386,
      "step": 67810
    },
    {
      "epoch": 2.7479183971151313,
      "grad_norm": 0.6139265298843384,
      "learning_rate": 3.467198576228903e-06,
      "loss": 1.0472,
      "step": 67820
    },
    {
      "epoch": 2.748323575292235,
      "grad_norm": 0.767494797706604,
      "learning_rate": 3.4561236046109504e-06,
      "loss": 1.0522,
      "step": 67830
    },
    {
      "epoch": 2.7487287534693383,
      "grad_norm": 0.6260986328125,
      "learning_rate": 3.445066038168887e-06,
      "loss": 1.012,
      "step": 67840
    },
    {
      "epoch": 2.7491339316464414,
      "grad_norm": 0.5756550431251526,
      "learning_rate": 3.43402587889623e-06,
      "loss": 1.0321,
      "step": 67850
    },
    {
      "epoch": 2.749539109823545,
      "grad_norm": 0.700505256652832,
      "learning_rate": 3.4230031287832865e-06,
      "loss": 1.0221,
      "step": 67860
    },
    {
      "epoch": 2.7499442880006484,
      "grad_norm": 0.627779483795166,
      "learning_rate": 3.4119977898172784e-06,
      "loss": 1.0614,
      "step": 67870
    },
    {
      "epoch": 2.7503494661777514,
      "grad_norm": 0.6679654121398926,
      "learning_rate": 3.401009863982296e-06,
      "loss": 1.0501,
      "step": 67880
    },
    {
      "epoch": 2.750754644354855,
      "grad_norm": 0.6628422737121582,
      "learning_rate": 3.3900393532592667e-06,
      "loss": 1.0483,
      "step": 67890
    },
    {
      "epoch": 2.7511598225319585,
      "grad_norm": 0.63641756772995,
      "learning_rate": 3.379086259625963e-06,
      "loss": 1.0588,
      "step": 67900
    },
    {
      "epoch": 2.751565000709062,
      "grad_norm": 0.6537044644355774,
      "learning_rate": 3.3681505850570616e-06,
      "loss": 1.0297,
      "step": 67910
    },
    {
      "epoch": 2.7519701788861655,
      "grad_norm": 0.6475873589515686,
      "learning_rate": 3.3572323315240405e-06,
      "loss": 1.0343,
      "step": 67920
    },
    {
      "epoch": 2.7523753570632685,
      "grad_norm": 0.7202847003936768,
      "learning_rate": 3.346331500995292e-06,
      "loss": 1.072,
      "step": 67930
    },
    {
      "epoch": 2.752780535240372,
      "grad_norm": 0.7343586683273315,
      "learning_rate": 3.3354480954360445e-06,
      "loss": 1.0345,
      "step": 67940
    },
    {
      "epoch": 2.7531857134174755,
      "grad_norm": 0.5969061255455017,
      "learning_rate": 3.3245821168083945e-06,
      "loss": 1.0264,
      "step": 67950
    },
    {
      "epoch": 2.7535908915945786,
      "grad_norm": 0.6820617914199829,
      "learning_rate": 3.3137335670712643e-06,
      "loss": 1.0602,
      "step": 67960
    },
    {
      "epoch": 2.753996069771682,
      "grad_norm": 0.595834493637085,
      "learning_rate": 3.3029024481804893e-06,
      "loss": 1.0374,
      "step": 67970
    },
    {
      "epoch": 2.7544012479487856,
      "grad_norm": 0.6496195197105408,
      "learning_rate": 3.2920887620887075e-06,
      "loss": 1.041,
      "step": 67980
    },
    {
      "epoch": 2.7548064261258887,
      "grad_norm": 0.6904287338256836,
      "learning_rate": 3.2812925107454372e-06,
      "loss": 1.028,
      "step": 67990
    },
    {
      "epoch": 2.755211604302992,
      "grad_norm": 0.6721323132514954,
      "learning_rate": 3.270513696097055e-06,
      "loss": 1.0757,
      "step": 68000
    },
    {
      "epoch": 2.7556167824800957,
      "grad_norm": 0.691513180732727,
      "learning_rate": 3.2597523200868174e-06,
      "loss": 1.0452,
      "step": 68010
    },
    {
      "epoch": 2.7560219606571987,
      "grad_norm": 1.107867956161499,
      "learning_rate": 3.2490083846547836e-06,
      "loss": 1.0061,
      "step": 68020
    },
    {
      "epoch": 2.7564271388343022,
      "grad_norm": 0.7765364646911621,
      "learning_rate": 3.2382818917379042e-06,
      "loss": 1.0932,
      "step": 68030
    },
    {
      "epoch": 2.7568323170114057,
      "grad_norm": 0.6737365126609802,
      "learning_rate": 3.227572843269988e-06,
      "loss": 1.0723,
      "step": 68040
    },
    {
      "epoch": 2.7572374951885092,
      "grad_norm": 0.6435474753379822,
      "learning_rate": 3.216881241181702e-06,
      "loss": 1.0682,
      "step": 68050
    },
    {
      "epoch": 2.7576426733656128,
      "grad_norm": 0.6688570976257324,
      "learning_rate": 3.2062070874005147e-06,
      "loss": 1.0529,
      "step": 68060
    },
    {
      "epoch": 2.758047851542716,
      "grad_norm": 0.7069236636161804,
      "learning_rate": 3.1955503838508205e-06,
      "loss": 1.0701,
      "step": 68070
    },
    {
      "epoch": 2.7584530297198193,
      "grad_norm": 0.7137054204940796,
      "learning_rate": 3.1849111324538385e-06,
      "loss": 1.0709,
      "step": 68080
    },
    {
      "epoch": 2.758858207896923,
      "grad_norm": 0.6889622211456299,
      "learning_rate": 3.1742893351276117e-06,
      "loss": 1.0746,
      "step": 68090
    },
    {
      "epoch": 2.759263386074026,
      "grad_norm": 0.697225034236908,
      "learning_rate": 3.1636849937870973e-06,
      "loss": 1.033,
      "step": 68100
    },
    {
      "epoch": 2.7596685642511294,
      "grad_norm": 0.7015243768692017,
      "learning_rate": 3.1530981103440547e-06,
      "loss": 1.0368,
      "step": 68110
    },
    {
      "epoch": 2.760073742428233,
      "grad_norm": 0.6448512673377991,
      "learning_rate": 3.1425286867071248e-06,
      "loss": 1.0525,
      "step": 68120
    },
    {
      "epoch": 2.760478920605336,
      "grad_norm": 0.7433374524116516,
      "learning_rate": 3.1319767247817823e-06,
      "loss": 1.0362,
      "step": 68130
    },
    {
      "epoch": 2.7608840987824395,
      "grad_norm": 0.6560097932815552,
      "learning_rate": 3.1214422264703726e-06,
      "loss": 1.058,
      "step": 68140
    },
    {
      "epoch": 2.761289276959543,
      "grad_norm": 0.6843791604042053,
      "learning_rate": 3.110925193672054e-06,
      "loss": 1.0835,
      "step": 68150
    },
    {
      "epoch": 2.7616944551366465,
      "grad_norm": 0.6504501700401306,
      "learning_rate": 3.100425628282899e-06,
      "loss": 1.0528,
      "step": 68160
    },
    {
      "epoch": 2.76209963331375,
      "grad_norm": 0.695166826248169,
      "learning_rate": 3.0899435321957715e-06,
      "loss": 1.042,
      "step": 68170
    },
    {
      "epoch": 2.762504811490853,
      "grad_norm": 0.641570508480072,
      "learning_rate": 3.079478907300426e-06,
      "loss": 1.0566,
      "step": 68180
    },
    {
      "epoch": 2.7629099896679565,
      "grad_norm": 0.6837283372879028,
      "learning_rate": 3.0690317554834426e-06,
      "loss": 1.0263,
      "step": 68190
    },
    {
      "epoch": 2.76331516784506,
      "grad_norm": 0.725983738899231,
      "learning_rate": 3.0586020786282697e-06,
      "loss": 1.0537,
      "step": 68200
    },
    {
      "epoch": 2.763720346022163,
      "grad_norm": 0.6525148153305054,
      "learning_rate": 3.0481898786151706e-06,
      "loss": 1.0325,
      "step": 68210
    },
    {
      "epoch": 2.7641255241992666,
      "grad_norm": 0.6970521807670593,
      "learning_rate": 3.0377951573213216e-06,
      "loss": 1.0607,
      "step": 68220
    },
    {
      "epoch": 2.76453070237637,
      "grad_norm": 0.5749348998069763,
      "learning_rate": 3.0274179166206784e-06,
      "loss": 1.002,
      "step": 68230
    },
    {
      "epoch": 2.764935880553473,
      "grad_norm": 0.6475690007209778,
      "learning_rate": 3.0170581583841006e-06,
      "loss": 1.0341,
      "step": 68240
    },
    {
      "epoch": 2.7653410587305767,
      "grad_norm": 0.6521359086036682,
      "learning_rate": 3.00671588447925e-06,
      "loss": 1.0448,
      "step": 68250
    },
    {
      "epoch": 2.76574623690768,
      "grad_norm": 0.6607743501663208,
      "learning_rate": 2.9963910967706675e-06,
      "loss": 1.0013,
      "step": 68260
    },
    {
      "epoch": 2.7661514150847837,
      "grad_norm": 0.6495952606201172,
      "learning_rate": 2.986083797119721e-06,
      "loss": 1.0486,
      "step": 68270
    },
    {
      "epoch": 2.766556593261887,
      "grad_norm": 0.6330111622810364,
      "learning_rate": 2.9757939873846676e-06,
      "loss": 1.0378,
      "step": 68280
    },
    {
      "epoch": 2.7669617714389902,
      "grad_norm": 0.6583353281021118,
      "learning_rate": 2.9655216694205455e-06,
      "loss": 1.0232,
      "step": 68290
    },
    {
      "epoch": 2.7673669496160938,
      "grad_norm": 0.6272956132888794,
      "learning_rate": 2.9552668450792965e-06,
      "loss": 1.0388,
      "step": 68300
    },
    {
      "epoch": 2.7677721277931973,
      "grad_norm": 0.6670888066291809,
      "learning_rate": 2.945029516209674e-06,
      "loss": 1.099,
      "step": 68310
    },
    {
      "epoch": 2.7681773059703003,
      "grad_norm": 0.6600485444068909,
      "learning_rate": 2.9348096846573135e-06,
      "loss": 1.0559,
      "step": 68320
    },
    {
      "epoch": 2.768582484147404,
      "grad_norm": 0.6829593181610107,
      "learning_rate": 2.924607352264619e-06,
      "loss": 1.0166,
      "step": 68330
    },
    {
      "epoch": 2.7689876623245073,
      "grad_norm": 0.6591202616691589,
      "learning_rate": 2.914422520870952e-06,
      "loss": 1.0188,
      "step": 68340
    },
    {
      "epoch": 2.7693928405016104,
      "grad_norm": 0.6258066892623901,
      "learning_rate": 2.9042551923124327e-06,
      "loss": 1.0226,
      "step": 68350
    },
    {
      "epoch": 2.769798018678714,
      "grad_norm": 0.705803394317627,
      "learning_rate": 2.894105368422051e-06,
      "loss": 1.0593,
      "step": 68360
    },
    {
      "epoch": 2.7702031968558174,
      "grad_norm": 0.7124722003936768,
      "learning_rate": 2.883973051029654e-06,
      "loss": 1.0103,
      "step": 68370
    },
    {
      "epoch": 2.7706083750329205,
      "grad_norm": 0.6710171103477478,
      "learning_rate": 2.8738582419619022e-06,
      "loss": 1.0676,
      "step": 68380
    },
    {
      "epoch": 2.771013553210024,
      "grad_norm": 0.5992341041564941,
      "learning_rate": 2.863760943042337e-06,
      "loss": 1.0598,
      "step": 68390
    },
    {
      "epoch": 2.7714187313871275,
      "grad_norm": 0.6116741299629211,
      "learning_rate": 2.853681156091337e-06,
      "loss": 1.1025,
      "step": 68400
    },
    {
      "epoch": 2.771823909564231,
      "grad_norm": 0.794705867767334,
      "learning_rate": 2.843618882926069e-06,
      "loss": 1.0105,
      "step": 68410
    },
    {
      "epoch": 2.7722290877413345,
      "grad_norm": 0.6641634702682495,
      "learning_rate": 2.8335741253606276e-06,
      "loss": 1.05,
      "step": 68420
    },
    {
      "epoch": 2.7726342659184375,
      "grad_norm": 0.6714094281196594,
      "learning_rate": 2.8235468852058966e-06,
      "loss": 1.0366,
      "step": 68430
    },
    {
      "epoch": 2.773039444095541,
      "grad_norm": 0.6991980671882629,
      "learning_rate": 2.8135371642695864e-06,
      "loss": 1.0715,
      "step": 68440
    },
    {
      "epoch": 2.7734446222726445,
      "grad_norm": 0.6753196716308594,
      "learning_rate": 2.803544964356308e-06,
      "loss": 1.0654,
      "step": 68450
    },
    {
      "epoch": 2.7738498004497476,
      "grad_norm": 0.5989596247673035,
      "learning_rate": 2.7935702872674665e-06,
      "loss": 1.0275,
      "step": 68460
    },
    {
      "epoch": 2.774254978626851,
      "grad_norm": 0.6517424583435059,
      "learning_rate": 2.783613134801333e-06,
      "loss": 1.0592,
      "step": 68470
    },
    {
      "epoch": 2.7746601568039546,
      "grad_norm": 0.6519949436187744,
      "learning_rate": 2.773673508752983e-06,
      "loss": 1.0338,
      "step": 68480
    },
    {
      "epoch": 2.7750653349810577,
      "grad_norm": 0.6830589175224304,
      "learning_rate": 2.7637514109143838e-06,
      "loss": 1.0475,
      "step": 68490
    },
    {
      "epoch": 2.775470513158161,
      "grad_norm": 0.7218981385231018,
      "learning_rate": 2.753846843074304e-06,
      "loss": 1.0497,
      "step": 68500
    },
    {
      "epoch": 2.7758756913352647,
      "grad_norm": 0.7056266069412231,
      "learning_rate": 2.7439598070183703e-06,
      "loss": 1.0147,
      "step": 68510
    },
    {
      "epoch": 2.776280869512368,
      "grad_norm": 0.6485363841056824,
      "learning_rate": 2.7340903045290354e-06,
      "loss": 1.0776,
      "step": 68520
    },
    {
      "epoch": 2.7766860476894717,
      "grad_norm": 0.6794052720069885,
      "learning_rate": 2.7242383373856094e-06,
      "loss": 1.0409,
      "step": 68530
    },
    {
      "epoch": 2.7770912258665748,
      "grad_norm": 0.6242119669914246,
      "learning_rate": 2.7144039073642048e-06,
      "loss": 1.0205,
      "step": 68540
    },
    {
      "epoch": 2.7774964040436783,
      "grad_norm": 0.654306948184967,
      "learning_rate": 2.704587016237825e-06,
      "loss": 1.0433,
      "step": 68550
    },
    {
      "epoch": 2.7779015822207818,
      "grad_norm": 0.648297905921936,
      "learning_rate": 2.6947876657762774e-06,
      "loss": 1.0292,
      "step": 68560
    },
    {
      "epoch": 2.778306760397885,
      "grad_norm": 0.5792853832244873,
      "learning_rate": 2.6850058577462145e-06,
      "loss": 1.0763,
      "step": 68570
    },
    {
      "epoch": 2.7787119385749883,
      "grad_norm": 0.6200276613235474,
      "learning_rate": 2.6752415939111155e-06,
      "loss": 1.0721,
      "step": 68580
    },
    {
      "epoch": 2.779117116752092,
      "grad_norm": 0.638606071472168,
      "learning_rate": 2.665494876031327e-06,
      "loss": 1.0061,
      "step": 68590
    },
    {
      "epoch": 2.779522294929195,
      "grad_norm": 0.616303026676178,
      "learning_rate": 2.6557657058639775e-06,
      "loss": 1.0499,
      "step": 68600
    },
    {
      "epoch": 2.7799274731062984,
      "grad_norm": 0.6343486905097961,
      "learning_rate": 2.6460540851630968e-06,
      "loss": 1.0589,
      "step": 68610
    },
    {
      "epoch": 2.780332651283402,
      "grad_norm": 0.6568349599838257,
      "learning_rate": 2.6363600156795065e-06,
      "loss": 1.0306,
      "step": 68620
    },
    {
      "epoch": 2.7807378294605054,
      "grad_norm": 0.6669828295707703,
      "learning_rate": 2.626683499160898e-06,
      "loss": 1.0402,
      "step": 68630
    },
    {
      "epoch": 2.781143007637609,
      "grad_norm": 0.6316933631896973,
      "learning_rate": 2.6170245373517534e-06,
      "loss": 1.0509,
      "step": 68640
    },
    {
      "epoch": 2.781548185814712,
      "grad_norm": 0.6286550760269165,
      "learning_rate": 2.607383131993424e-06,
      "loss": 1.0356,
      "step": 68650
    },
    {
      "epoch": 2.7819533639918155,
      "grad_norm": 0.727990984916687,
      "learning_rate": 2.597759284824075e-06,
      "loss": 1.0144,
      "step": 68660
    },
    {
      "epoch": 2.782358542168919,
      "grad_norm": 0.6877814531326294,
      "learning_rate": 2.5881529975787523e-06,
      "loss": 1.0837,
      "step": 68670
    },
    {
      "epoch": 2.782763720346022,
      "grad_norm": 0.7435288429260254,
      "learning_rate": 2.57856427198927e-06,
      "loss": 1.0557,
      "step": 68680
    },
    {
      "epoch": 2.7831688985231255,
      "grad_norm": 0.6797462105751038,
      "learning_rate": 2.5689931097843345e-06,
      "loss": 1.0353,
      "step": 68690
    },
    {
      "epoch": 2.783574076700229,
      "grad_norm": 0.663299560546875,
      "learning_rate": 2.559439512689432e-06,
      "loss": 0.9951,
      "step": 68700
    },
    {
      "epoch": 2.783979254877332,
      "grad_norm": 0.6236684918403625,
      "learning_rate": 2.5499034824269076e-06,
      "loss": 1.0448,
      "step": 68710
    },
    {
      "epoch": 2.7843844330544356,
      "grad_norm": 0.7189601063728333,
      "learning_rate": 2.540385020715963e-06,
      "loss": 1.0772,
      "step": 68720
    },
    {
      "epoch": 2.784789611231539,
      "grad_norm": 0.6414672136306763,
      "learning_rate": 2.5308841292726147e-06,
      "loss": 1.0343,
      "step": 68730
    },
    {
      "epoch": 2.785194789408642,
      "grad_norm": 0.804131805896759,
      "learning_rate": 2.521400809809682e-06,
      "loss": 1.0319,
      "step": 68740
    },
    {
      "epoch": 2.7855999675857457,
      "grad_norm": 0.6059761047363281,
      "learning_rate": 2.511935064036852e-06,
      "loss": 1.0377,
      "step": 68750
    },
    {
      "epoch": 2.786005145762849,
      "grad_norm": 0.6294721364974976,
      "learning_rate": 2.5024868936606492e-06,
      "loss": 1.0136,
      "step": 68760
    },
    {
      "epoch": 2.7864103239399527,
      "grad_norm": 0.6864262223243713,
      "learning_rate": 2.493056300384389e-06,
      "loss": 1.019,
      "step": 68770
    },
    {
      "epoch": 2.786815502117056,
      "grad_norm": 0.6535334587097168,
      "learning_rate": 2.483643285908255e-06,
      "loss": 1.0684,
      "step": 68780
    },
    {
      "epoch": 2.7872206802941593,
      "grad_norm": 0.6322632431983948,
      "learning_rate": 2.4742478519292455e-06,
      "loss": 1.021,
      "step": 68790
    },
    {
      "epoch": 2.7876258584712628,
      "grad_norm": 0.615077793598175,
      "learning_rate": 2.4648700001412172e-06,
      "loss": 1.0528,
      "step": 68800
    },
    {
      "epoch": 2.7880310366483663,
      "grad_norm": 0.607313334941864,
      "learning_rate": 2.455509732234795e-06,
      "loss": 1.0586,
      "step": 68810
    },
    {
      "epoch": 2.7884362148254693,
      "grad_norm": 0.6246482729911804,
      "learning_rate": 2.4461670498974964e-06,
      "loss": 1.0351,
      "step": 68820
    },
    {
      "epoch": 2.788841393002573,
      "grad_norm": 0.6801596283912659,
      "learning_rate": 2.436841954813618e-06,
      "loss": 1.0802,
      "step": 68830
    },
    {
      "epoch": 2.7892465711796763,
      "grad_norm": 0.6460620760917664,
      "learning_rate": 2.4275344486643594e-06,
      "loss": 1.0809,
      "step": 68840
    },
    {
      "epoch": 2.7896517493567794,
      "grad_norm": 0.605324387550354,
      "learning_rate": 2.418244533127656e-06,
      "loss": 1.0842,
      "step": 68850
    },
    {
      "epoch": 2.790056927533883,
      "grad_norm": 0.6161472201347351,
      "learning_rate": 2.408972209878335e-06,
      "loss": 1.0425,
      "step": 68860
    },
    {
      "epoch": 2.7904621057109864,
      "grad_norm": 0.7125236392021179,
      "learning_rate": 2.3997174805880373e-06,
      "loss": 0.9951,
      "step": 68870
    },
    {
      "epoch": 2.79086728388809,
      "grad_norm": 0.7656241059303284,
      "learning_rate": 2.390480346925228e-06,
      "loss": 1.0271,
      "step": 68880
    },
    {
      "epoch": 2.7912724620651934,
      "grad_norm": 0.6730086207389832,
      "learning_rate": 2.381260810555186e-06,
      "loss": 1.0324,
      "step": 68890
    },
    {
      "epoch": 2.7916776402422965,
      "grad_norm": 0.6119165420532227,
      "learning_rate": 2.3720588731400593e-06,
      "loss": 1.0305,
      "step": 68900
    },
    {
      "epoch": 2.7920828184194,
      "grad_norm": 0.6928576231002808,
      "learning_rate": 2.362874536338777e-06,
      "loss": 1.0318,
      "step": 68910
    },
    {
      "epoch": 2.7924879965965035,
      "grad_norm": 0.6536161303520203,
      "learning_rate": 2.3537078018071257e-06,
      "loss": 1.0521,
      "step": 68920
    },
    {
      "epoch": 2.7928931747736065,
      "grad_norm": 0.6375036239624023,
      "learning_rate": 2.344558671197694e-06,
      "loss": 1.0515,
      "step": 68930
    },
    {
      "epoch": 2.79329835295071,
      "grad_norm": 0.6042559146881104,
      "learning_rate": 2.3354271461599297e-06,
      "loss": 1.0865,
      "step": 68940
    },
    {
      "epoch": 2.7937035311278136,
      "grad_norm": 0.665297269821167,
      "learning_rate": 2.326313228340049e-06,
      "loss": 1.0551,
      "step": 68950
    },
    {
      "epoch": 2.7941087093049166,
      "grad_norm": 0.631813645362854,
      "learning_rate": 2.3172169193811933e-06,
      "loss": 1.0249,
      "step": 68960
    },
    {
      "epoch": 2.79451388748202,
      "grad_norm": 0.6033936142921448,
      "learning_rate": 2.3081382209232173e-06,
      "loss": 1.0593,
      "step": 68970
    },
    {
      "epoch": 2.7949190656591236,
      "grad_norm": 0.6532555222511292,
      "learning_rate": 2.2990771346028782e-06,
      "loss": 1.0524,
      "step": 68980
    },
    {
      "epoch": 2.795324243836227,
      "grad_norm": 0.6711127758026123,
      "learning_rate": 2.290033662053703e-06,
      "loss": 1.0471,
      "step": 68990
    },
    {
      "epoch": 2.7957294220133306,
      "grad_norm": 0.6589937210083008,
      "learning_rate": 2.2810078049061103e-06,
      "loss": 1.0042,
      "step": 69000
    },
    {
      "epoch": 2.7961346001904337,
      "grad_norm": 0.6727391481399536,
      "learning_rate": 2.271999564787275e-06,
      "loss": 1.0331,
      "step": 69010
    },
    {
      "epoch": 2.796539778367537,
      "grad_norm": 0.678297221660614,
      "learning_rate": 2.2630089433212543e-06,
      "loss": 1.0456,
      "step": 69020
    },
    {
      "epoch": 2.7969449565446407,
      "grad_norm": 0.7427763938903809,
      "learning_rate": 2.2540359421288626e-06,
      "loss": 1.1118,
      "step": 69030
    },
    {
      "epoch": 2.7973501347217438,
      "grad_norm": 0.6215517520904541,
      "learning_rate": 2.245080562827806e-06,
      "loss": 1.0092,
      "step": 69040
    },
    {
      "epoch": 2.7977553128988473,
      "grad_norm": 0.62956702709198,
      "learning_rate": 2.2361428070325706e-06,
      "loss": 1.0334,
      "step": 69050
    },
    {
      "epoch": 2.7981604910759508,
      "grad_norm": 0.6547031402587891,
      "learning_rate": 2.227222676354479e-06,
      "loss": 1.0672,
      "step": 69060
    },
    {
      "epoch": 2.798565669253054,
      "grad_norm": 0.6110846400260925,
      "learning_rate": 2.2183201724016667e-06,
      "loss": 1.0647,
      "step": 69070
    },
    {
      "epoch": 2.7989708474301573,
      "grad_norm": 0.7029847502708435,
      "learning_rate": 2.2094352967791167e-06,
      "loss": 1.0552,
      "step": 69080
    },
    {
      "epoch": 2.799376025607261,
      "grad_norm": 0.6890741586685181,
      "learning_rate": 2.200568051088625e-06,
      "loss": 1.0305,
      "step": 69090
    },
    {
      "epoch": 2.799781203784364,
      "grad_norm": 0.634532630443573,
      "learning_rate": 2.191718436928758e-06,
      "loss": 1.0992,
      "step": 69100
    },
    {
      "epoch": 2.8001863819614674,
      "grad_norm": 0.6948356032371521,
      "learning_rate": 2.1828864558949835e-06,
      "loss": 1.0474,
      "step": 69110
    },
    {
      "epoch": 2.800591560138571,
      "grad_norm": 0.6239469051361084,
      "learning_rate": 2.17407210957955e-06,
      "loss": 1.0383,
      "step": 69120
    },
    {
      "epoch": 2.8009967383156744,
      "grad_norm": 0.6278371810913086,
      "learning_rate": 2.1652753995715313e-06,
      "loss": 1.0529,
      "step": 69130
    },
    {
      "epoch": 2.801401916492778,
      "grad_norm": 0.6438855528831482,
      "learning_rate": 2.1564963274568027e-06,
      "loss": 1.0859,
      "step": 69140
    },
    {
      "epoch": 2.801807094669881,
      "grad_norm": 0.7399506568908691,
      "learning_rate": 2.1477348948180877e-06,
      "loss": 1.0492,
      "step": 69150
    },
    {
      "epoch": 2.8022122728469845,
      "grad_norm": 0.6966156959533691,
      "learning_rate": 2.138991103234922e-06,
      "loss": 1.0496,
      "step": 69160
    },
    {
      "epoch": 2.802617451024088,
      "grad_norm": 0.6751066446304321,
      "learning_rate": 2.130264954283656e-06,
      "loss": 1.0317,
      "step": 69170
    },
    {
      "epoch": 2.803022629201191,
      "grad_norm": 0.649691641330719,
      "learning_rate": 2.1215564495374653e-06,
      "loss": 1.0401,
      "step": 69180
    },
    {
      "epoch": 2.8034278073782946,
      "grad_norm": 0.6258114576339722,
      "learning_rate": 2.1128655905663487e-06,
      "loss": 1.0612,
      "step": 69190
    },
    {
      "epoch": 2.803832985555398,
      "grad_norm": 0.696506679058075,
      "learning_rate": 2.1041923789370977e-06,
      "loss": 1.0494,
      "step": 69200
    },
    {
      "epoch": 2.804238163732501,
      "grad_norm": 0.6723718643188477,
      "learning_rate": 2.0955368162133505e-06,
      "loss": 1.0379,
      "step": 69210
    },
    {
      "epoch": 2.8046433419096046,
      "grad_norm": 0.6514427661895752,
      "learning_rate": 2.086898903955559e-06,
      "loss": 1.0738,
      "step": 69220
    },
    {
      "epoch": 2.805048520086708,
      "grad_norm": 0.667366087436676,
      "learning_rate": 2.078278643720977e-06,
      "loss": 1.0417,
      "step": 69230
    },
    {
      "epoch": 2.8054536982638116,
      "grad_norm": 0.63231360912323,
      "learning_rate": 2.069676037063684e-06,
      "loss": 1.0359,
      "step": 69240
    },
    {
      "epoch": 2.805858876440915,
      "grad_norm": 0.6106957197189331,
      "learning_rate": 2.0610910855346055e-06,
      "loss": 1.0147,
      "step": 69250
    },
    {
      "epoch": 2.806264054618018,
      "grad_norm": 0.6847702860832214,
      "learning_rate": 2.0525237906814375e-06,
      "loss": 1.0215,
      "step": 69260
    },
    {
      "epoch": 2.8066692327951217,
      "grad_norm": 0.6943152546882629,
      "learning_rate": 2.0439741540487225e-06,
      "loss": 1.0309,
      "step": 69270
    },
    {
      "epoch": 2.807074410972225,
      "grad_norm": 0.6821515560150146,
      "learning_rate": 2.035442177177782e-06,
      "loss": 1.037,
      "step": 69280
    },
    {
      "epoch": 2.8074795891493283,
      "grad_norm": 0.6328322887420654,
      "learning_rate": 2.026927861606831e-06,
      "loss": 1.0831,
      "step": 69290
    },
    {
      "epoch": 2.8078847673264318,
      "grad_norm": 0.6787276864051819,
      "learning_rate": 2.018431208870819e-06,
      "loss": 1.0469,
      "step": 69300
    },
    {
      "epoch": 2.8082899455035353,
      "grad_norm": 0.7551531791687012,
      "learning_rate": 2.009952220501565e-06,
      "loss": 1.0192,
      "step": 69310
    },
    {
      "epoch": 2.8086951236806383,
      "grad_norm": 0.6845291256904602,
      "learning_rate": 2.001490898027658e-06,
      "loss": 1.0715,
      "step": 69320
    },
    {
      "epoch": 2.809100301857742,
      "grad_norm": 0.6368486285209656,
      "learning_rate": 1.993047242974544e-06,
      "loss": 1.0087,
      "step": 69330
    },
    {
      "epoch": 2.8095054800348453,
      "grad_norm": 0.6401746273040771,
      "learning_rate": 1.9846212568644497e-06,
      "loss": 1.0139,
      "step": 69340
    },
    {
      "epoch": 2.809910658211949,
      "grad_norm": 0.712479293346405,
      "learning_rate": 1.9762129412164598e-06,
      "loss": 1.0115,
      "step": 69350
    },
    {
      "epoch": 2.8103158363890524,
      "grad_norm": 0.7417783141136169,
      "learning_rate": 1.967822297546429e-06,
      "loss": 1.0506,
      "step": 69360
    },
    {
      "epoch": 2.8107210145661554,
      "grad_norm": 0.8579964637756348,
      "learning_rate": 1.9594493273670466e-06,
      "loss": 1.0283,
      "step": 69370
    },
    {
      "epoch": 2.811126192743259,
      "grad_norm": 0.6560613512992859,
      "learning_rate": 1.9510940321878166e-06,
      "loss": 1.0559,
      "step": 69380
    },
    {
      "epoch": 2.8115313709203624,
      "grad_norm": 0.6437401175498962,
      "learning_rate": 1.9427564135150346e-06,
      "loss": 1.0135,
      "step": 69390
    },
    {
      "epoch": 2.8119365490974655,
      "grad_norm": 0.6880306601524353,
      "learning_rate": 1.934436472851853e-06,
      "loss": 1.029,
      "step": 69400
    },
    {
      "epoch": 2.812341727274569,
      "grad_norm": 0.7221899032592773,
      "learning_rate": 1.9261342116982162e-06,
      "loss": 1.0688,
      "step": 69410
    },
    {
      "epoch": 2.8127469054516725,
      "grad_norm": 0.6212006211280823,
      "learning_rate": 1.91784963155085e-06,
      "loss": 1.0637,
      "step": 69420
    },
    {
      "epoch": 2.8131520836287756,
      "grad_norm": 0.5961959958076477,
      "learning_rate": 1.9095827339033257e-06,
      "loss": 1.0588,
      "step": 69430
    },
    {
      "epoch": 2.813557261805879,
      "grad_norm": 0.6268693804740906,
      "learning_rate": 1.9013335202460404e-06,
      "loss": 1.0378,
      "step": 69440
    },
    {
      "epoch": 2.8139624399829826,
      "grad_norm": 0.6483755707740784,
      "learning_rate": 1.8931019920661708e-06,
      "loss": 1.0636,
      "step": 69450
    },
    {
      "epoch": 2.8143676181600856,
      "grad_norm": 0.7445999383926392,
      "learning_rate": 1.8848881508477078e-06,
      "loss": 1.0327,
      "step": 69460
    },
    {
      "epoch": 2.814772796337189,
      "grad_norm": 0.6329090595245361,
      "learning_rate": 1.876691998071478e-06,
      "loss": 1.0344,
      "step": 69470
    },
    {
      "epoch": 2.8151779745142926,
      "grad_norm": 0.6659951210021973,
      "learning_rate": 1.8685135352150996e-06,
      "loss": 1.0332,
      "step": 69480
    },
    {
      "epoch": 2.815583152691396,
      "grad_norm": 0.6571592688560486,
      "learning_rate": 1.860352763753004e-06,
      "loss": 1.0405,
      "step": 69490
    },
    {
      "epoch": 2.8159883308684996,
      "grad_norm": 0.8239168524742126,
      "learning_rate": 1.8522096851564475e-06,
      "loss": 1.0607,
      "step": 69500
    },
    {
      "epoch": 2.8163935090456027,
      "grad_norm": 0.7895735502243042,
      "learning_rate": 1.8440843008934561e-06,
      "loss": 1.0836,
      "step": 69510
    },
    {
      "epoch": 2.816798687222706,
      "grad_norm": 0.6779717803001404,
      "learning_rate": 1.8359766124289357e-06,
      "loss": 1.0169,
      "step": 69520
    },
    {
      "epoch": 2.8172038653998097,
      "grad_norm": 0.6896478533744812,
      "learning_rate": 1.8278866212245393e-06,
      "loss": 1.0296,
      "step": 69530
    },
    {
      "epoch": 2.8176090435769128,
      "grad_norm": 0.6068835258483887,
      "learning_rate": 1.819814328738756e-06,
      "loss": 1.0706,
      "step": 69540
    },
    {
      "epoch": 2.8180142217540163,
      "grad_norm": 0.7148863673210144,
      "learning_rate": 1.8117597364268546e-06,
      "loss": 1.0432,
      "step": 69550
    },
    {
      "epoch": 2.81841939993112,
      "grad_norm": 0.675895094871521,
      "learning_rate": 1.803722845740985e-06,
      "loss": 1.0332,
      "step": 69560
    },
    {
      "epoch": 2.818824578108223,
      "grad_norm": 0.6620787382125854,
      "learning_rate": 1.7957036581299992e-06,
      "loss": 1.0482,
      "step": 69570
    },
    {
      "epoch": 2.8192297562853263,
      "grad_norm": 0.7438561320304871,
      "learning_rate": 1.7877021750396738e-06,
      "loss": 1.0676,
      "step": 69580
    },
    {
      "epoch": 2.81963493446243,
      "grad_norm": 0.6898619532585144,
      "learning_rate": 1.779718397912511e-06,
      "loss": 1.0855,
      "step": 69590
    },
    {
      "epoch": 2.8200401126395334,
      "grad_norm": 0.6694394946098328,
      "learning_rate": 1.7717523281878478e-06,
      "loss": 1.0749,
      "step": 69600
    },
    {
      "epoch": 2.820445290816637,
      "grad_norm": 0.6418788433074951,
      "learning_rate": 1.7638039673018136e-06,
      "loss": 1.0201,
      "step": 69610
    },
    {
      "epoch": 2.82085046899374,
      "grad_norm": 0.6671682596206665,
      "learning_rate": 1.7558733166873841e-06,
      "loss": 1.0516,
      "step": 69620
    },
    {
      "epoch": 2.8212556471708434,
      "grad_norm": 0.6646163463592529,
      "learning_rate": 1.7479603777742938e-06,
      "loss": 1.0636,
      "step": 69630
    },
    {
      "epoch": 2.821660825347947,
      "grad_norm": 0.6759927272796631,
      "learning_rate": 1.7400651519891341e-06,
      "loss": 1.0622,
      "step": 69640
    },
    {
      "epoch": 2.82206600352505,
      "grad_norm": 0.6833605170249939,
      "learning_rate": 1.7321876407552563e-06,
      "loss": 1.0414,
      "step": 69650
    },
    {
      "epoch": 2.8224711817021535,
      "grad_norm": 0.6718578934669495,
      "learning_rate": 1.7243278454928458e-06,
      "loss": 1.0457,
      "step": 69660
    },
    {
      "epoch": 2.822876359879257,
      "grad_norm": 0.6336293816566467,
      "learning_rate": 1.7164857676188805e-06,
      "loss": 1.0219,
      "step": 69670
    },
    {
      "epoch": 2.82328153805636,
      "grad_norm": 0.6957451701164246,
      "learning_rate": 1.708661408547152e-06,
      "loss": 1.0655,
      "step": 69680
    },
    {
      "epoch": 2.8236867162334636,
      "grad_norm": 0.695421040058136,
      "learning_rate": 1.700854769688265e-06,
      "loss": 1.0753,
      "step": 69690
    },
    {
      "epoch": 2.824091894410567,
      "grad_norm": 0.686269998550415,
      "learning_rate": 1.6930658524496156e-06,
      "loss": 1.0538,
      "step": 69700
    },
    {
      "epoch": 2.8244970725876706,
      "grad_norm": 0.6078731417655945,
      "learning_rate": 1.6852946582353922e-06,
      "loss": 1.0352,
      "step": 69710
    },
    {
      "epoch": 2.824902250764774,
      "grad_norm": 0.6727893948554993,
      "learning_rate": 1.6775411884466296e-06,
      "loss": 1.0468,
      "step": 69720
    },
    {
      "epoch": 2.825307428941877,
      "grad_norm": 0.6661391258239746,
      "learning_rate": 1.6698054444811318e-06,
      "loss": 1.0583,
      "step": 69730
    },
    {
      "epoch": 2.8257126071189806,
      "grad_norm": 0.7017878890037537,
      "learning_rate": 1.6620874277335275e-06,
      "loss": 1.0058,
      "step": 69740
    },
    {
      "epoch": 2.826117785296084,
      "grad_norm": 0.6524959802627563,
      "learning_rate": 1.654387139595226e-06,
      "loss": 1.0486,
      "step": 69750
    },
    {
      "epoch": 2.826522963473187,
      "grad_norm": 0.6768895983695984,
      "learning_rate": 1.6467045814544612e-06,
      "loss": 1.0326,
      "step": 69760
    },
    {
      "epoch": 2.8269281416502907,
      "grad_norm": 0.6400077939033508,
      "learning_rate": 1.639039754696281e-06,
      "loss": 1.0551,
      "step": 69770
    },
    {
      "epoch": 2.827333319827394,
      "grad_norm": 0.6403458118438721,
      "learning_rate": 1.6313926607024798e-06,
      "loss": 1.0416,
      "step": 69780
    },
    {
      "epoch": 2.8277384980044973,
      "grad_norm": 0.6919483542442322,
      "learning_rate": 1.6237633008517328e-06,
      "loss": 1.0113,
      "step": 69790
    },
    {
      "epoch": 2.828143676181601,
      "grad_norm": 0.7242262959480286,
      "learning_rate": 1.6161516765194617e-06,
      "loss": 1.0055,
      "step": 69800
    },
    {
      "epoch": 2.8285488543587043,
      "grad_norm": 0.6595095992088318,
      "learning_rate": 1.6085577890779247e-06,
      "loss": 1.0217,
      "step": 69810
    },
    {
      "epoch": 2.8289540325358074,
      "grad_norm": 0.6237406730651855,
      "learning_rate": 1.6009816398961486e-06,
      "loss": 1.0416,
      "step": 69820
    },
    {
      "epoch": 2.829359210712911,
      "grad_norm": 0.6861992478370667,
      "learning_rate": 1.5934232303399965e-06,
      "loss": 1.0088,
      "step": 69830
    },
    {
      "epoch": 2.8297643888900144,
      "grad_norm": 0.8565301895141602,
      "learning_rate": 1.585882561772112e-06,
      "loss": 1.0446,
      "step": 69840
    },
    {
      "epoch": 2.830169567067118,
      "grad_norm": 0.651001513004303,
      "learning_rate": 1.5783596355519402e-06,
      "loss": 1.0302,
      "step": 69850
    },
    {
      "epoch": 2.8305747452442214,
      "grad_norm": 0.7436246275901794,
      "learning_rate": 1.5708544530357305e-06,
      "loss": 1.0194,
      "step": 69860
    },
    {
      "epoch": 2.8309799234213244,
      "grad_norm": 0.6174436807632446,
      "learning_rate": 1.5633670155765557e-06,
      "loss": 0.9869,
      "step": 69870
    },
    {
      "epoch": 2.831385101598428,
      "grad_norm": 0.6466653943061829,
      "learning_rate": 1.5558973245242582e-06,
      "loss": 1.026,
      "step": 69880
    },
    {
      "epoch": 2.8317902797755314,
      "grad_norm": 0.5942612290382385,
      "learning_rate": 1.548445381225494e-06,
      "loss": 1.0705,
      "step": 69890
    },
    {
      "epoch": 2.8321954579526345,
      "grad_norm": 0.6910516023635864,
      "learning_rate": 1.5410111870236997e-06,
      "loss": 1.031,
      "step": 69900
    },
    {
      "epoch": 2.832600636129738,
      "grad_norm": 0.6820365786552429,
      "learning_rate": 1.5335947432591702e-06,
      "loss": 1.0408,
      "step": 69910
    },
    {
      "epoch": 2.8330058143068415,
      "grad_norm": 0.6434486508369446,
      "learning_rate": 1.5261960512689243e-06,
      "loss": 1.0109,
      "step": 69920
    },
    {
      "epoch": 2.8334109924839446,
      "grad_norm": 0.6299521327018738,
      "learning_rate": 1.5188151123868399e-06,
      "loss": 1.0596,
      "step": 69930
    },
    {
      "epoch": 2.833816170661048,
      "grad_norm": 0.8220200538635254,
      "learning_rate": 1.5114519279435634e-06,
      "loss": 1.0616,
      "step": 69940
    },
    {
      "epoch": 2.8342213488381516,
      "grad_norm": 0.6953267455101013,
      "learning_rate": 1.5041064992665554e-06,
      "loss": 1.0129,
      "step": 69950
    },
    {
      "epoch": 2.834626527015255,
      "grad_norm": 0.6897409558296204,
      "learning_rate": 1.4967788276800455e-06,
      "loss": 1.0726,
      "step": 69960
    },
    {
      "epoch": 2.8350317051923586,
      "grad_norm": 0.6137917041778564,
      "learning_rate": 1.4894689145051099e-06,
      "loss": 1.0415,
      "step": 69970
    },
    {
      "epoch": 2.8354368833694616,
      "grad_norm": 0.6681788563728333,
      "learning_rate": 1.4821767610595837e-06,
      "loss": 1.0315,
      "step": 69980
    },
    {
      "epoch": 2.835842061546565,
      "grad_norm": 0.638327419757843,
      "learning_rate": 1.4749023686581376e-06,
      "loss": 1.0829,
      "step": 69990
    },
    {
      "epoch": 2.8362472397236687,
      "grad_norm": 0.6293131113052368,
      "learning_rate": 1.4676457386121889e-06,
      "loss": 1.0513,
      "step": 70000
    },
    {
      "epoch": 2.8366524179007717,
      "grad_norm": 0.680732011795044,
      "learning_rate": 1.4604068722299912e-06,
      "loss": 1.014,
      "step": 70010
    },
    {
      "epoch": 2.837057596077875,
      "grad_norm": 0.7861828804016113,
      "learning_rate": 1.4531857708165897e-06,
      "loss": 1.0084,
      "step": 70020
    },
    {
      "epoch": 2.8374627742549787,
      "grad_norm": 0.726841926574707,
      "learning_rate": 1.4459824356738206e-06,
      "loss": 1.0656,
      "step": 70030
    },
    {
      "epoch": 2.837867952432082,
      "grad_norm": 0.6173111200332642,
      "learning_rate": 1.4387968681003228e-06,
      "loss": 1.0438,
      "step": 70040
    },
    {
      "epoch": 2.8382731306091853,
      "grad_norm": 0.6611813306808472,
      "learning_rate": 1.4316290693915156e-06,
      "loss": 1.0633,
      "step": 70050
    },
    {
      "epoch": 2.838678308786289,
      "grad_norm": 0.6615204215049744,
      "learning_rate": 1.4244790408396547e-06,
      "loss": 1.0419,
      "step": 70060
    },
    {
      "epoch": 2.8390834869633923,
      "grad_norm": 0.6163116693496704,
      "learning_rate": 1.4173467837337418e-06,
      "loss": 1.0773,
      "step": 70070
    },
    {
      "epoch": 2.839488665140496,
      "grad_norm": 0.6138267517089844,
      "learning_rate": 1.410232299359604e-06,
      "loss": 1.0223,
      "step": 70080
    },
    {
      "epoch": 2.839893843317599,
      "grad_norm": 0.6881386637687683,
      "learning_rate": 1.4031355889998598e-06,
      "loss": 1.0223,
      "step": 70090
    },
    {
      "epoch": 2.8402990214947024,
      "grad_norm": 0.6153237223625183,
      "learning_rate": 1.3960566539339303e-06,
      "loss": 1.05,
      "step": 70100
    },
    {
      "epoch": 2.840704199671806,
      "grad_norm": 0.662213921546936,
      "learning_rate": 1.3889954954380058e-06,
      "loss": 1.0493,
      "step": 70110
    },
    {
      "epoch": 2.841109377848909,
      "grad_norm": 0.6008002758026123,
      "learning_rate": 1.3819521147851123e-06,
      "loss": 1.0286,
      "step": 70120
    },
    {
      "epoch": 2.8415145560260124,
      "grad_norm": 0.6658783555030823,
      "learning_rate": 1.3749265132450228e-06,
      "loss": 1.0353,
      "step": 70130
    },
    {
      "epoch": 2.841919734203116,
      "grad_norm": 0.6919381022453308,
      "learning_rate": 1.3679186920843468e-06,
      "loss": 1.0534,
      "step": 70140
    },
    {
      "epoch": 2.842324912380219,
      "grad_norm": 0.6143175363540649,
      "learning_rate": 1.3609286525664733e-06,
      "loss": 1.0165,
      "step": 70150
    },
    {
      "epoch": 2.8427300905573225,
      "grad_norm": 0.6724287867546082,
      "learning_rate": 1.3539563959515945e-06,
      "loss": 1.0609,
      "step": 70160
    },
    {
      "epoch": 2.843135268734426,
      "grad_norm": 0.623758852481842,
      "learning_rate": 1.3470019234966492e-06,
      "loss": 1.0368,
      "step": 70170
    },
    {
      "epoch": 2.843540446911529,
      "grad_norm": 0.6428565979003906,
      "learning_rate": 1.3400652364554456e-06,
      "loss": 1.0533,
      "step": 70180
    },
    {
      "epoch": 2.8439456250886326,
      "grad_norm": 0.6462804079055786,
      "learning_rate": 1.333146336078528e-06,
      "loss": 1.0435,
      "step": 70190
    },
    {
      "epoch": 2.844350803265736,
      "grad_norm": 0.6184031367301941,
      "learning_rate": 1.326245223613265e-06,
      "loss": 1.0416,
      "step": 70200
    },
    {
      "epoch": 2.8447559814428396,
      "grad_norm": 0.6298539042472839,
      "learning_rate": 1.3193619003037948e-06,
      "loss": 1.0506,
      "step": 70210
    },
    {
      "epoch": 2.845161159619943,
      "grad_norm": 0.6802024245262146,
      "learning_rate": 1.3124963673910696e-06,
      "loss": 1.0553,
      "step": 70220
    },
    {
      "epoch": 2.845566337797046,
      "grad_norm": 0.6317134499549866,
      "learning_rate": 1.3056486261128098e-06,
      "loss": 1.0726,
      "step": 70230
    },
    {
      "epoch": 2.8459715159741497,
      "grad_norm": 0.6493830680847168,
      "learning_rate": 1.2988186777035615e-06,
      "loss": 1.0313,
      "step": 70240
    },
    {
      "epoch": 2.846376694151253,
      "grad_norm": 0.6227492094039917,
      "learning_rate": 1.29200652339464e-06,
      "loss": 1.0329,
      "step": 70250
    },
    {
      "epoch": 2.8467818723283562,
      "grad_norm": 0.7201076745986938,
      "learning_rate": 1.2852121644141623e-06,
      "loss": 1.0571,
      "step": 70260
    },
    {
      "epoch": 2.8471870505054597,
      "grad_norm": 0.6568790078163147,
      "learning_rate": 1.278435601987016e-06,
      "loss": 1.0215,
      "step": 70270
    },
    {
      "epoch": 2.8475922286825632,
      "grad_norm": 0.6390981674194336,
      "learning_rate": 1.271676837334923e-06,
      "loss": 1.0482,
      "step": 70280
    },
    {
      "epoch": 2.8479974068596663,
      "grad_norm": 0.6653500199317932,
      "learning_rate": 1.2649358716763426e-06,
      "loss": 1.0775,
      "step": 70290
    },
    {
      "epoch": 2.84840258503677,
      "grad_norm": 0.6673510670661926,
      "learning_rate": 1.258212706226558e-06,
      "loss": 1.0346,
      "step": 70300
    },
    {
      "epoch": 2.8488077632138733,
      "grad_norm": 0.6337644457817078,
      "learning_rate": 1.2515073421976663e-06,
      "loss": 1.0389,
      "step": 70310
    },
    {
      "epoch": 2.849212941390977,
      "grad_norm": 0.8247761130332947,
      "learning_rate": 1.2448197807985006e-06,
      "loss": 1.0662,
      "step": 70320
    },
    {
      "epoch": 2.8496181195680803,
      "grad_norm": 0.6746142506599426,
      "learning_rate": 1.2381500232347187e-06,
      "loss": 1.0448,
      "step": 70330
    },
    {
      "epoch": 2.8500232977451834,
      "grad_norm": 0.6366877555847168,
      "learning_rate": 1.2314980707087698e-06,
      "loss": 1.0813,
      "step": 70340
    },
    {
      "epoch": 2.850428475922287,
      "grad_norm": 0.6310604214668274,
      "learning_rate": 1.22486392441985e-06,
      "loss": 1.0624,
      "step": 70350
    },
    {
      "epoch": 2.8508336540993904,
      "grad_norm": 0.6859033703804016,
      "learning_rate": 1.2182475855640362e-06,
      "loss": 1.0631,
      "step": 70360
    },
    {
      "epoch": 2.8512388322764934,
      "grad_norm": 0.6570857763290405,
      "learning_rate": 1.2116490553340964e-06,
      "loss": 1.0191,
      "step": 70370
    },
    {
      "epoch": 2.851644010453597,
      "grad_norm": 0.6805106997489929,
      "learning_rate": 1.2050683349196567e-06,
      "loss": 1.0465,
      "step": 70380
    },
    {
      "epoch": 2.8520491886307004,
      "grad_norm": 0.6491690278053284,
      "learning_rate": 1.1985054255070794e-06,
      "loss": 1.0616,
      "step": 70390
    },
    {
      "epoch": 2.8524543668078035,
      "grad_norm": 0.6420149803161621,
      "learning_rate": 1.191960328279551e-06,
      "loss": 1.0222,
      "step": 70400
    },
    {
      "epoch": 2.852859544984907,
      "grad_norm": 0.5462638139724731,
      "learning_rate": 1.1854330444170504e-06,
      "loss": 1.0388,
      "step": 70410
    },
    {
      "epoch": 2.8532647231620105,
      "grad_norm": 0.6627232432365417,
      "learning_rate": 1.1789235750963357e-06,
      "loss": 1.0147,
      "step": 70420
    },
    {
      "epoch": 2.853669901339114,
      "grad_norm": 0.6354261636734009,
      "learning_rate": 1.1724319214909353e-06,
      "loss": 1.0541,
      "step": 70430
    },
    {
      "epoch": 2.8540750795162175,
      "grad_norm": 0.7410251498222351,
      "learning_rate": 1.1659580847711794e-06,
      "loss": 1.0612,
      "step": 70440
    },
    {
      "epoch": 2.8544802576933206,
      "grad_norm": 0.6772418022155762,
      "learning_rate": 1.159502066104201e-06,
      "loss": 1.0133,
      "step": 70450
    },
    {
      "epoch": 2.854885435870424,
      "grad_norm": 0.605535626411438,
      "learning_rate": 1.153063866653903e-06,
      "loss": 1.0458,
      "step": 70460
    },
    {
      "epoch": 2.8552906140475276,
      "grad_norm": 0.633358895778656,
      "learning_rate": 1.1466434875809785e-06,
      "loss": 1.0354,
      "step": 70470
    },
    {
      "epoch": 2.8556957922246307,
      "grad_norm": 0.6721415519714355,
      "learning_rate": 1.1402409300429017e-06,
      "loss": 1.0376,
      "step": 70480
    },
    {
      "epoch": 2.856100970401734,
      "grad_norm": 0.7222537994384766,
      "learning_rate": 1.1338561951939609e-06,
      "loss": 1.0631,
      "step": 70490
    },
    {
      "epoch": 2.8565061485788377,
      "grad_norm": 0.6791864037513733,
      "learning_rate": 1.1274892841852013e-06,
      "loss": 1.0069,
      "step": 70500
    },
    {
      "epoch": 2.8569113267559407,
      "grad_norm": 0.650875985622406,
      "learning_rate": 1.121140198164472e-06,
      "loss": 1.0197,
      "step": 70510
    },
    {
      "epoch": 2.8573165049330442,
      "grad_norm": 0.6472572684288025,
      "learning_rate": 1.1148089382763794e-06,
      "loss": 1.0093,
      "step": 70520
    },
    {
      "epoch": 2.8577216831101477,
      "grad_norm": 0.6164147257804871,
      "learning_rate": 1.1084955056623657e-06,
      "loss": 1.0803,
      "step": 70530
    },
    {
      "epoch": 2.858126861287251,
      "grad_norm": 0.5920906662940979,
      "learning_rate": 1.1021999014606322e-06,
      "loss": 1.0692,
      "step": 70540
    },
    {
      "epoch": 2.8585320394643543,
      "grad_norm": 0.7030291557312012,
      "learning_rate": 1.0959221268061481e-06,
      "loss": 1.0309,
      "step": 70550
    },
    {
      "epoch": 2.858937217641458,
      "grad_norm": 0.7024365663528442,
      "learning_rate": 1.0896621828306974e-06,
      "loss": 1.0586,
      "step": 70560
    },
    {
      "epoch": 2.8593423958185613,
      "grad_norm": 0.6087422370910645,
      "learning_rate": 1.083420070662844e-06,
      "loss": 1.0406,
      "step": 70570
    },
    {
      "epoch": 2.859747573995665,
      "grad_norm": 0.6424879431724548,
      "learning_rate": 1.0771957914279095e-06,
      "loss": 1.0371,
      "step": 70580
    },
    {
      "epoch": 2.860152752172768,
      "grad_norm": 0.7640267014503479,
      "learning_rate": 1.0709893462480637e-06,
      "loss": 1.038,
      "step": 70590
    },
    {
      "epoch": 2.8605579303498714,
      "grad_norm": 0.711650013923645,
      "learning_rate": 1.0648007362421774e-06,
      "loss": 1.0422,
      "step": 70600
    },
    {
      "epoch": 2.860963108526975,
      "grad_norm": 0.762421727180481,
      "learning_rate": 1.05862996252597e-06,
      "loss": 1.0246,
      "step": 70610
    },
    {
      "epoch": 2.861368286704078,
      "grad_norm": 0.5808322429656982,
      "learning_rate": 1.0524770262119289e-06,
      "loss": 1.0477,
      "step": 70620
    },
    {
      "epoch": 2.8617734648811815,
      "grad_norm": 0.6623627543449402,
      "learning_rate": 1.0463419284093112e-06,
      "loss": 1.0284,
      "step": 70630
    },
    {
      "epoch": 2.862178643058285,
      "grad_norm": 0.6560257077217102,
      "learning_rate": 1.0402246702241547e-06,
      "loss": 1.0264,
      "step": 70640
    },
    {
      "epoch": 2.862583821235388,
      "grad_norm": 0.6457691192626953,
      "learning_rate": 1.0341252527593325e-06,
      "loss": 1.0571,
      "step": 70650
    },
    {
      "epoch": 2.8629889994124915,
      "grad_norm": 0.6837692260742188,
      "learning_rate": 1.028043677114421e-06,
      "loss": 1.0501,
      "step": 70660
    },
    {
      "epoch": 2.863394177589595,
      "grad_norm": 0.6846857070922852,
      "learning_rate": 1.0219799443858646e-06,
      "loss": 1.0591,
      "step": 70670
    },
    {
      "epoch": 2.8637993557666985,
      "grad_norm": 0.7183936834335327,
      "learning_rate": 1.0159340556668007e-06,
      "loss": 1.0468,
      "step": 70680
    },
    {
      "epoch": 2.864204533943802,
      "grad_norm": 0.6277278065681458,
      "learning_rate": 1.0099060120472349e-06,
      "loss": 1.0688,
      "step": 70690
    },
    {
      "epoch": 2.864609712120905,
      "grad_norm": 0.684565007686615,
      "learning_rate": 1.0038958146139088e-06,
      "loss": 1.0453,
      "step": 70700
    },
    {
      "epoch": 2.8650148902980086,
      "grad_norm": 0.7267411947250366,
      "learning_rate": 9.979034644503448e-07,
      "loss": 1.0724,
      "step": 70710
    },
    {
      "epoch": 2.865420068475112,
      "grad_norm": 0.6810652017593384,
      "learning_rate": 9.919289626368788e-07,
      "loss": 1.0381,
      "step": 70720
    },
    {
      "epoch": 2.865825246652215,
      "grad_norm": 0.6045084595680237,
      "learning_rate": 9.859723102505825e-07,
      "loss": 1.0512,
      "step": 70730
    },
    {
      "epoch": 2.8662304248293187,
      "grad_norm": 0.6015311479568481,
      "learning_rate": 9.800335083653634e-07,
      "loss": 1.0466,
      "step": 70740
    },
    {
      "epoch": 2.866635603006422,
      "grad_norm": 0.6248055100440979,
      "learning_rate": 9.74112558051865e-07,
      "loss": 1.0188,
      "step": 70750
    },
    {
      "epoch": 2.8670407811835252,
      "grad_norm": 0.6598406434059143,
      "learning_rate": 9.68209460377545e-07,
      "loss": 1.0412,
      "step": 70760
    },
    {
      "epoch": 2.8674459593606287,
      "grad_norm": 0.6459853053092957,
      "learning_rate": 9.623242164066182e-07,
      "loss": 1.0492,
      "step": 70770
    },
    {
      "epoch": 2.8678511375377322,
      "grad_norm": 0.7572448253631592,
      "learning_rate": 9.564568272000917e-07,
      "loss": 1.0362,
      "step": 70780
    },
    {
      "epoch": 2.8682563157148357,
      "grad_norm": 0.6025314331054688,
      "learning_rate": 9.506072938157639e-07,
      "loss": 1.059,
      "step": 70790
    },
    {
      "epoch": 2.8686614938919393,
      "grad_norm": 0.6653679609298706,
      "learning_rate": 9.447756173081912e-07,
      "loss": 1.0385,
      "step": 70800
    },
    {
      "epoch": 2.8690666720690423,
      "grad_norm": 0.63753342628479,
      "learning_rate": 9.389617987287325e-07,
      "loss": 1.0684,
      "step": 70810
    },
    {
      "epoch": 2.869471850246146,
      "grad_norm": 0.6699547171592712,
      "learning_rate": 9.331658391255161e-07,
      "loss": 1.0526,
      "step": 70820
    },
    {
      "epoch": 2.8698770284232493,
      "grad_norm": 0.6132635474205017,
      "learning_rate": 9.273877395434394e-07,
      "loss": 1.0448,
      "step": 70830
    },
    {
      "epoch": 2.8702822066003524,
      "grad_norm": 0.7641729712486267,
      "learning_rate": 9.216275010242248e-07,
      "loss": 1.0537,
      "step": 70840
    },
    {
      "epoch": 2.870687384777456,
      "grad_norm": 0.6680240631103516,
      "learning_rate": 9.158851246062972e-07,
      "loss": 1.0498,
      "step": 70850
    },
    {
      "epoch": 2.8710925629545594,
      "grad_norm": 0.6219913959503174,
      "learning_rate": 9.101606113249395e-07,
      "loss": 1.022,
      "step": 70860
    },
    {
      "epoch": 2.8714977411316625,
      "grad_norm": 0.6238787174224854,
      "learning_rate": 9.044539622121817e-07,
      "loss": 1.047,
      "step": 70870
    },
    {
      "epoch": 2.871902919308766,
      "grad_norm": 0.6256201863288879,
      "learning_rate": 8.987651782968232e-07,
      "loss": 1.0615,
      "step": 70880
    },
    {
      "epoch": 2.8723080974858695,
      "grad_norm": 0.6871959567070007,
      "learning_rate": 8.930942606044435e-07,
      "loss": 1.0576,
      "step": 70890
    },
    {
      "epoch": 2.8727132756629725,
      "grad_norm": 0.6973142623901367,
      "learning_rate": 8.874412101574358e-07,
      "loss": 1.0385,
      "step": 70900
    },
    {
      "epoch": 2.873118453840076,
      "grad_norm": 0.6270265579223633,
      "learning_rate": 8.818060279749185e-07,
      "loss": 1.0111,
      "step": 70910
    },
    {
      "epoch": 2.8735236320171795,
      "grad_norm": 0.6475241184234619,
      "learning_rate": 8.761887150728343e-07,
      "loss": 1.0283,
      "step": 70920
    },
    {
      "epoch": 2.873928810194283,
      "grad_norm": 0.7883360981941223,
      "learning_rate": 8.705892724638732e-07,
      "loss": 1.059,
      "step": 70930
    },
    {
      "epoch": 2.8743339883713865,
      "grad_norm": 0.6858165264129639,
      "learning_rate": 8.650077011575386e-07,
      "loss": 1.0518,
      "step": 70940
    },
    {
      "epoch": 2.8747391665484896,
      "grad_norm": 0.7338923215866089,
      "learning_rate": 8.594440021600703e-07,
      "loss": 1.0719,
      "step": 70950
    },
    {
      "epoch": 2.875144344725593,
      "grad_norm": 0.6952561736106873,
      "learning_rate": 8.538981764745102e-07,
      "loss": 1.0523,
      "step": 70960
    },
    {
      "epoch": 2.8755495229026966,
      "grad_norm": 0.7166057825088501,
      "learning_rate": 8.483702251006809e-07,
      "loss": 1.0761,
      "step": 70970
    },
    {
      "epoch": 2.8759547010797997,
      "grad_norm": 0.7450185418128967,
      "learning_rate": 8.42860149035174e-07,
      "loss": 1.0446,
      "step": 70980
    },
    {
      "epoch": 2.876359879256903,
      "grad_norm": 0.6297682523727417,
      "learning_rate": 8.373679492713504e-07,
      "loss": 1.0449,
      "step": 70990
    },
    {
      "epoch": 2.8767650574340067,
      "grad_norm": 0.6692603826522827,
      "learning_rate": 8.318936267993627e-07,
      "loss": 1.0441,
      "step": 71000
    },
    {
      "epoch": 2.8771702356111097,
      "grad_norm": 0.7217695116996765,
      "learning_rate": 8.264371826061435e-07,
      "loss": 1.0489,
      "step": 71010
    },
    {
      "epoch": 2.8775754137882132,
      "grad_norm": 0.8120151162147522,
      "learning_rate": 8.209986176753948e-07,
      "loss": 1.0601,
      "step": 71020
    },
    {
      "epoch": 2.8779805919653167,
      "grad_norm": 0.5891145467758179,
      "learning_rate": 8.155779329875767e-07,
      "loss": 1.0897,
      "step": 71030
    },
    {
      "epoch": 2.8783857701424203,
      "grad_norm": 0.6195822954177856,
      "learning_rate": 8.101751295199633e-07,
      "loss": 1.0264,
      "step": 71040
    },
    {
      "epoch": 2.8787909483195238,
      "grad_norm": 0.6963731646537781,
      "learning_rate": 8.047902082465753e-07,
      "loss": 1.0043,
      "step": 71050
    },
    {
      "epoch": 2.879196126496627,
      "grad_norm": 0.6473860144615173,
      "learning_rate": 7.994231701382249e-07,
      "loss": 1.0702,
      "step": 71060
    },
    {
      "epoch": 2.8796013046737303,
      "grad_norm": 0.681564211845398,
      "learning_rate": 7.940740161625049e-07,
      "loss": 1.0611,
      "step": 71070
    },
    {
      "epoch": 2.880006482850834,
      "grad_norm": 0.6193928718566895,
      "learning_rate": 7.887427472837439e-07,
      "loss": 1.052,
      "step": 71080
    },
    {
      "epoch": 2.880411661027937,
      "grad_norm": 0.6552684903144836,
      "learning_rate": 7.834293644631064e-07,
      "loss": 1.0369,
      "step": 71090
    },
    {
      "epoch": 2.8808168392050404,
      "grad_norm": 0.7668026089668274,
      "learning_rate": 7.781338686584927e-07,
      "loss": 1.0505,
      "step": 71100
    },
    {
      "epoch": 2.881222017382144,
      "grad_norm": 0.6372596025466919,
      "learning_rate": 7.728562608245837e-07,
      "loss": 1.0195,
      "step": 71110
    },
    {
      "epoch": 2.881627195559247,
      "grad_norm": 0.5738928318023682,
      "learning_rate": 7.675965419128517e-07,
      "loss": 1.0048,
      "step": 71120
    },
    {
      "epoch": 2.8820323737363505,
      "grad_norm": 0.7094082236289978,
      "learning_rate": 7.623547128715158e-07,
      "loss": 1.0939,
      "step": 71130
    },
    {
      "epoch": 2.882437551913454,
      "grad_norm": 0.6967258453369141,
      "learning_rate": 7.57130774645598e-07,
      "loss": 1.0769,
      "step": 71140
    },
    {
      "epoch": 2.8828427300905575,
      "grad_norm": 0.6133907437324524,
      "learning_rate": 7.519247281768894e-07,
      "loss": 1.0575,
      "step": 71150
    },
    {
      "epoch": 2.883247908267661,
      "grad_norm": 0.5723984241485596,
      "learning_rate": 7.467365744039279e-07,
      "loss": 1.019,
      "step": 71160
    },
    {
      "epoch": 2.883653086444764,
      "grad_norm": 0.6442399024963379,
      "learning_rate": 7.415663142620654e-07,
      "loss": 1.0414,
      "step": 71170
    },
    {
      "epoch": 2.8840582646218675,
      "grad_norm": 0.6673558354377747,
      "learning_rate": 7.364139486834009e-07,
      "loss": 1.0261,
      "step": 71180
    },
    {
      "epoch": 2.884463442798971,
      "grad_norm": 0.7459567785263062,
      "learning_rate": 7.312794785968247e-07,
      "loss": 1.0612,
      "step": 71190
    },
    {
      "epoch": 2.884868620976074,
      "grad_norm": 0.6670302152633667,
      "learning_rate": 7.261629049279739e-07,
      "loss": 1.0661,
      "step": 71200
    },
    {
      "epoch": 2.8852737991531776,
      "grad_norm": 0.704867959022522,
      "learning_rate": 7.210642285992996e-07,
      "loss": 1.04,
      "step": 71210
    },
    {
      "epoch": 2.885678977330281,
      "grad_norm": 0.6345574855804443,
      "learning_rate": 7.159834505300001e-07,
      "loss": 1.1064,
      "step": 71220
    },
    {
      "epoch": 2.886084155507384,
      "grad_norm": 0.665203869342804,
      "learning_rate": 7.109205716360312e-07,
      "loss": 1.0777,
      "step": 71230
    },
    {
      "epoch": 2.8864893336844877,
      "grad_norm": 0.6648913621902466,
      "learning_rate": 7.05875592830163e-07,
      "loss": 1.0419,
      "step": 71240
    },
    {
      "epoch": 2.886894511861591,
      "grad_norm": 0.6542073488235474,
      "learning_rate": 7.008485150219124e-07,
      "loss": 1.0735,
      "step": 71250
    },
    {
      "epoch": 2.8872996900386942,
      "grad_norm": 0.6322083473205566,
      "learning_rate": 6.958393391175655e-07,
      "loss": 1.0701,
      "step": 71260
    },
    {
      "epoch": 2.8877048682157977,
      "grad_norm": 1.7169793844223022,
      "learning_rate": 6.908480660201999e-07,
      "loss": 1.013,
      "step": 71270
    },
    {
      "epoch": 2.8881100463929013,
      "grad_norm": 0.8696016073226929,
      "learning_rate": 6.858746966296403e-07,
      "loss": 1.0435,
      "step": 71280
    },
    {
      "epoch": 2.8885152245700048,
      "grad_norm": 0.7509192824363708,
      "learning_rate": 6.80919231842525e-07,
      "loss": 1.0703,
      "step": 71290
    },
    {
      "epoch": 2.8889204027471083,
      "grad_norm": 0.6981762647628784,
      "learning_rate": 6.759816725522062e-07,
      "loss": 1.0487,
      "step": 71300
    },
    {
      "epoch": 2.8893255809242113,
      "grad_norm": 0.733170747756958,
      "learning_rate": 6.710620196488604e-07,
      "loss": 1.0768,
      "step": 71310
    },
    {
      "epoch": 2.889730759101315,
      "grad_norm": 0.6088043451309204,
      "learning_rate": 6.661602740194006e-07,
      "loss": 1.0649,
      "step": 71320
    },
    {
      "epoch": 2.8901359372784183,
      "grad_norm": 0.7993614673614502,
      "learning_rate": 6.612764365475421e-07,
      "loss": 0.9931,
      "step": 71330
    },
    {
      "epoch": 2.8905411154555214,
      "grad_norm": 0.6676160097122192,
      "learning_rate": 6.564105081137473e-07,
      "loss": 1.0612,
      "step": 71340
    },
    {
      "epoch": 2.890946293632625,
      "grad_norm": 0.6159685850143433,
      "learning_rate": 6.515624895952588e-07,
      "loss": 1.0581,
      "step": 71350
    },
    {
      "epoch": 2.8913514718097284,
      "grad_norm": 0.7033995389938354,
      "learning_rate": 6.467323818660887e-07,
      "loss": 1.0496,
      "step": 71360
    },
    {
      "epoch": 2.8917566499868315,
      "grad_norm": 0.6692166924476624,
      "learning_rate": 6.419201857970292e-07,
      "loss": 1.0276,
      "step": 71370
    },
    {
      "epoch": 2.892161828163935,
      "grad_norm": 0.6612096428871155,
      "learning_rate": 6.371259022556198e-07,
      "loss": 1.0147,
      "step": 71380
    },
    {
      "epoch": 2.8925670063410385,
      "grad_norm": 0.6447829008102417,
      "learning_rate": 6.323495321061912e-07,
      "loss": 1.0478,
      "step": 71390
    },
    {
      "epoch": 2.892972184518142,
      "grad_norm": 0.6576183438301086,
      "learning_rate": 6.275910762098547e-07,
      "loss": 1.0487,
      "step": 71400
    },
    {
      "epoch": 2.8933773626952455,
      "grad_norm": 0.6641037464141846,
      "learning_rate": 6.228505354244573e-07,
      "loss": 1.0479,
      "step": 71410
    },
    {
      "epoch": 2.8937825408723485,
      "grad_norm": 0.6610855460166931,
      "learning_rate": 6.181279106046489e-07,
      "loss": 1.0435,
      "step": 71420
    },
    {
      "epoch": 2.894187719049452,
      "grad_norm": 0.7343400716781616,
      "learning_rate": 6.13423202601826e-07,
      "loss": 1.0324,
      "step": 71430
    },
    {
      "epoch": 2.8945928972265556,
      "grad_norm": 0.710221529006958,
      "learning_rate": 6.087364122641659e-07,
      "loss": 1.0558,
      "step": 71440
    },
    {
      "epoch": 2.8949980754036586,
      "grad_norm": 0.6180974245071411,
      "learning_rate": 6.040675404366258e-07,
      "loss": 1.0317,
      "step": 71450
    },
    {
      "epoch": 2.895403253580762,
      "grad_norm": 0.67976313829422,
      "learning_rate": 5.994165879609215e-07,
      "loss": 1.102,
      "step": 71460
    },
    {
      "epoch": 2.8958084317578656,
      "grad_norm": 0.7003075480461121,
      "learning_rate": 5.947835556755265e-07,
      "loss": 1.0556,
      "step": 71470
    },
    {
      "epoch": 2.8962136099349687,
      "grad_norm": 0.6333001255989075,
      "learning_rate": 5.901684444156951e-07,
      "loss": 1.0444,
      "step": 71480
    },
    {
      "epoch": 2.896618788112072,
      "grad_norm": 0.6229148507118225,
      "learning_rate": 5.855712550134729e-07,
      "loss": 1.0362,
      "step": 71490
    },
    {
      "epoch": 2.8970239662891757,
      "grad_norm": 0.653725802898407,
      "learning_rate": 5.8099198829763e-07,
      "loss": 1.01,
      "step": 71500
    },
    {
      "epoch": 2.897429144466279,
      "grad_norm": 0.652742326259613,
      "learning_rate": 5.764306450937284e-07,
      "loss": 1.033,
      "step": 71510
    },
    {
      "epoch": 2.8978343226433827,
      "grad_norm": 0.5803306102752686,
      "learning_rate": 5.718872262241215e-07,
      "loss": 1.0522,
      "step": 71520
    },
    {
      "epoch": 2.8982395008204858,
      "grad_norm": 0.5935584306716919,
      "learning_rate": 5.673617325078873e-07,
      "loss": 1.0547,
      "step": 71530
    },
    {
      "epoch": 2.8986446789975893,
      "grad_norm": 0.6543251276016235,
      "learning_rate": 5.628541647608953e-07,
      "loss": 1.0686,
      "step": 71540
    },
    {
      "epoch": 2.8990498571746928,
      "grad_norm": 0.796481728553772,
      "learning_rate": 5.583645237957847e-07,
      "loss": 1.0539,
      "step": 71550
    },
    {
      "epoch": 2.899455035351796,
      "grad_norm": 0.6762794852256775,
      "learning_rate": 5.538928104219632e-07,
      "loss": 1.037,
      "step": 71560
    },
    {
      "epoch": 2.8998602135288993,
      "grad_norm": 0.6832278966903687,
      "learning_rate": 5.494390254456083e-07,
      "loss": 1.0659,
      "step": 71570
    },
    {
      "epoch": 2.900265391706003,
      "grad_norm": 0.6569262742996216,
      "learning_rate": 5.450031696696445e-07,
      "loss": 1.0357,
      "step": 71580
    },
    {
      "epoch": 2.900670569883106,
      "grad_norm": 0.6184719800949097,
      "learning_rate": 5.405852438937764e-07,
      "loss": 1.0692,
      "step": 71590
    },
    {
      "epoch": 2.9010757480602094,
      "grad_norm": 0.7283926606178284,
      "learning_rate": 5.361852489145003e-07,
      "loss": 1.0325,
      "step": 71600
    },
    {
      "epoch": 2.901480926237313,
      "grad_norm": 0.7023428678512573,
      "learning_rate": 5.318031855250371e-07,
      "loss": 1.0584,
      "step": 71610
    },
    {
      "epoch": 2.901886104414416,
      "grad_norm": 0.7924466729164124,
      "learning_rate": 5.274390545154218e-07,
      "loss": 1.0564,
      "step": 71620
    },
    {
      "epoch": 2.9022912825915195,
      "grad_norm": 0.7225878834724426,
      "learning_rate": 5.230928566724025e-07,
      "loss": 1.0566,
      "step": 71630
    },
    {
      "epoch": 2.902696460768623,
      "grad_norm": 0.6963199973106384,
      "learning_rate": 5.187645927795304e-07,
      "loss": 1.0715,
      "step": 71640
    },
    {
      "epoch": 2.9031016389457265,
      "grad_norm": 0.6520086526870728,
      "learning_rate": 5.144542636171257e-07,
      "loss": 1.0582,
      "step": 71650
    },
    {
      "epoch": 2.90350681712283,
      "grad_norm": 0.6288042068481445,
      "learning_rate": 5.101618699622668e-07,
      "loss": 1.0594,
      "step": 71660
    },
    {
      "epoch": 2.903911995299933,
      "grad_norm": 0.6563532948493958,
      "learning_rate": 5.058874125887903e-07,
      "loss": 1.0377,
      "step": 71670
    },
    {
      "epoch": 2.9043171734770366,
      "grad_norm": 0.7031261324882507,
      "learning_rate": 5.016308922673018e-07,
      "loss": 1.0261,
      "step": 71680
    },
    {
      "epoch": 2.90472235165414,
      "grad_norm": 0.6940221190452576,
      "learning_rate": 4.973923097651878e-07,
      "loss": 1.0391,
      "step": 71690
    },
    {
      "epoch": 2.905127529831243,
      "grad_norm": 0.7024141550064087,
      "learning_rate": 4.931716658465813e-07,
      "loss": 1.0484,
      "step": 71700
    },
    {
      "epoch": 2.9055327080083466,
      "grad_norm": 0.6410354375839233,
      "learning_rate": 4.889689612723958e-07,
      "loss": 1.0114,
      "step": 71710
    },
    {
      "epoch": 2.90593788618545,
      "grad_norm": 0.6572251319885254,
      "learning_rate": 4.84784196800303e-07,
      "loss": 1.0154,
      "step": 71720
    },
    {
      "epoch": 2.906343064362553,
      "grad_norm": 0.7083673477172852,
      "learning_rate": 4.80617373184744e-07,
      "loss": 1.0583,
      "step": 71730
    },
    {
      "epoch": 2.9067482425396567,
      "grad_norm": 0.6293151378631592,
      "learning_rate": 4.7646849117692905e-07,
      "loss": 1.0869,
      "step": 71740
    },
    {
      "epoch": 2.90715342071676,
      "grad_norm": 0.6040356159210205,
      "learning_rate": 4.723375515248263e-07,
      "loss": 1.046,
      "step": 71750
    },
    {
      "epoch": 2.9075585988938637,
      "grad_norm": 0.6594181656837463,
      "learning_rate": 4.682245549731734e-07,
      "loss": 1.0424,
      "step": 71760
    },
    {
      "epoch": 2.907963777070967,
      "grad_norm": 0.7294296622276306,
      "learning_rate": 4.6412950226346616e-07,
      "loss": 1.0359,
      "step": 71770
    },
    {
      "epoch": 2.9083689552480703,
      "grad_norm": 0.6311412453651428,
      "learning_rate": 4.600523941339807e-07,
      "loss": 1.0355,
      "step": 71780
    },
    {
      "epoch": 2.9087741334251738,
      "grad_norm": 0.7162185311317444,
      "learning_rate": 4.5599323131974013e-07,
      "loss": 1.0656,
      "step": 71790
    },
    {
      "epoch": 2.9091793116022773,
      "grad_norm": 0.8093481659889221,
      "learning_rate": 4.5195201455253687e-07,
      "loss": 1.017,
      "step": 71800
    },
    {
      "epoch": 2.9095844897793803,
      "grad_norm": 0.64060378074646,
      "learning_rate": 4.479287445609437e-07,
      "loss": 1.0581,
      "step": 71810
    },
    {
      "epoch": 2.909989667956484,
      "grad_norm": 0.6751753687858582,
      "learning_rate": 4.4392342207026926e-07,
      "loss": 1.0,
      "step": 71820
    },
    {
      "epoch": 2.9103948461335873,
      "grad_norm": 0.6940590143203735,
      "learning_rate": 4.399360478026249e-07,
      "loss": 1.0426,
      "step": 71830
    },
    {
      "epoch": 2.9108000243106904,
      "grad_norm": 0.6505831480026245,
      "learning_rate": 4.359666224768466e-07,
      "loss": 1.0456,
      "step": 71840
    },
    {
      "epoch": 2.911205202487794,
      "grad_norm": 0.6408365368843079,
      "learning_rate": 4.3201514680857315e-07,
      "loss": 1.0643,
      "step": 71850
    },
    {
      "epoch": 2.9116103806648974,
      "grad_norm": 0.7547171115875244,
      "learning_rate": 4.2808162151016795e-07,
      "loss": 1.0681,
      "step": 71860
    },
    {
      "epoch": 2.912015558842001,
      "grad_norm": 0.6383624076843262,
      "learning_rate": 4.241660472907749e-07,
      "loss": 1.0525,
      "step": 71870
    },
    {
      "epoch": 2.9124207370191044,
      "grad_norm": 0.6588336825370789,
      "learning_rate": 4.2026842485630716e-07,
      "loss": 1.026,
      "step": 71880
    },
    {
      "epoch": 2.9128259151962075,
      "grad_norm": 0.7006398439407349,
      "learning_rate": 4.163887549094581e-07,
      "loss": 1.0927,
      "step": 71890
    },
    {
      "epoch": 2.913231093373311,
      "grad_norm": 0.7451176643371582,
      "learning_rate": 4.125270381496349e-07,
      "loss": 1.0281,
      "step": 71900
    },
    {
      "epoch": 2.9136362715504145,
      "grad_norm": 0.6723106503486633,
      "learning_rate": 4.086832752730585e-07,
      "loss": 0.9868,
      "step": 71910
    },
    {
      "epoch": 2.9140414497275176,
      "grad_norm": 0.6432937979698181,
      "learning_rate": 4.048574669726857e-07,
      "loss": 1.0264,
      "step": 71920
    },
    {
      "epoch": 2.914446627904621,
      "grad_norm": 0.7774739265441895,
      "learning_rate": 4.010496139382425e-07,
      "loss": 1.0189,
      "step": 71930
    },
    {
      "epoch": 2.9148518060817246,
      "grad_norm": 0.6344271302223206,
      "learning_rate": 3.972597168562131e-07,
      "loss": 1.0376,
      "step": 71940
    },
    {
      "epoch": 2.9152569842588276,
      "grad_norm": 0.6489766836166382,
      "learning_rate": 3.93487776409851e-07,
      "loss": 1.013,
      "step": 71950
    },
    {
      "epoch": 2.915662162435931,
      "grad_norm": 0.6628417372703552,
      "learning_rate": 3.897337932791789e-07,
      "loss": 1.0297,
      "step": 71960
    },
    {
      "epoch": 2.9160673406130346,
      "grad_norm": 0.6792346835136414,
      "learning_rate": 3.8599776814097767e-07,
      "loss": 1.0287,
      "step": 71970
    },
    {
      "epoch": 2.9164725187901377,
      "grad_norm": 0.6913134455680847,
      "learning_rate": 3.8227970166877516e-07,
      "loss": 1.0425,
      "step": 71980
    },
    {
      "epoch": 2.916877696967241,
      "grad_norm": 0.6313996911048889,
      "learning_rate": 3.785795945328796e-07,
      "loss": 1.0447,
      "step": 71990
    },
    {
      "epoch": 2.9172828751443447,
      "grad_norm": 0.7390498518943787,
      "learning_rate": 3.7489744740034637e-07,
      "loss": 1.0508,
      "step": 72000
    },
    {
      "epoch": 2.917688053321448,
      "grad_norm": 0.6620385050773621,
      "learning_rate": 3.712332609350222e-07,
      "loss": 1.0204,
      "step": 72010
    },
    {
      "epoch": 2.9180932314985517,
      "grad_norm": 0.6584206223487854,
      "learning_rate": 3.675870357974787e-07,
      "loss": 1.0536,
      "step": 72020
    },
    {
      "epoch": 2.9184984096756548,
      "grad_norm": 0.6329182386398315,
      "learning_rate": 3.6395877264507883e-07,
      "loss": 1.0251,
      "step": 72030
    },
    {
      "epoch": 2.9189035878527583,
      "grad_norm": 0.6251294612884521,
      "learning_rate": 3.603484721319217e-07,
      "loss": 1.0551,
      "step": 72040
    },
    {
      "epoch": 2.919308766029862,
      "grad_norm": 0.7064741849899292,
      "learning_rate": 3.567561349088866e-07,
      "loss": 1.0314,
      "step": 72050
    },
    {
      "epoch": 2.919713944206965,
      "grad_norm": 0.6258947253227234,
      "learning_rate": 3.5318176162362216e-07,
      "loss": 1.0217,
      "step": 72060
    },
    {
      "epoch": 2.9201191223840683,
      "grad_norm": 0.7107666730880737,
      "learning_rate": 3.496253529205018e-07,
      "loss": 1.0433,
      "step": 72070
    },
    {
      "epoch": 2.920524300561172,
      "grad_norm": 0.6372382044792175,
      "learning_rate": 3.4608690944071263e-07,
      "loss": 0.9996,
      "step": 72080
    },
    {
      "epoch": 2.920929478738275,
      "grad_norm": 0.627644956111908,
      "learning_rate": 3.425664318221444e-07,
      "loss": 1.055,
      "step": 72090
    },
    {
      "epoch": 2.9213346569153784,
      "grad_norm": 0.7004165649414062,
      "learning_rate": 3.3906392069950056e-07,
      "loss": 1.0631,
      "step": 72100
    },
    {
      "epoch": 2.921739835092482,
      "grad_norm": 0.6051058769226074,
      "learning_rate": 3.3557937670420927e-07,
      "loss": 1.0088,
      "step": 72110
    },
    {
      "epoch": 2.9221450132695854,
      "grad_norm": 0.6499029994010925,
      "learning_rate": 3.321128004644791e-07,
      "loss": 1.0377,
      "step": 72120
    },
    {
      "epoch": 2.922550191446689,
      "grad_norm": 0.6353285908699036,
      "learning_rate": 3.286641926052658e-07,
      "loss": 1.0639,
      "step": 72130
    },
    {
      "epoch": 2.922955369623792,
      "grad_norm": 0.6738817691802979,
      "learning_rate": 3.252335537483164e-07,
      "loss": 1.0171,
      "step": 72140
    },
    {
      "epoch": 2.9233605478008955,
      "grad_norm": 0.6833760142326355,
      "learning_rate": 3.218208845120807e-07,
      "loss": 1.0562,
      "step": 72150
    },
    {
      "epoch": 2.923765725977999,
      "grad_norm": 0.6081545948982239,
      "learning_rate": 3.18426185511822e-07,
      "loss": 1.0444,
      "step": 72160
    },
    {
      "epoch": 2.924170904155102,
      "grad_norm": 0.6868718266487122,
      "learning_rate": 3.150494573595508e-07,
      "loss": 1.0386,
      "step": 72170
    },
    {
      "epoch": 2.9245760823322056,
      "grad_norm": 0.6079546213150024,
      "learning_rate": 3.116907006640246e-07,
      "loss": 1.0356,
      "step": 72180
    },
    {
      "epoch": 2.924981260509309,
      "grad_norm": 0.762519896030426,
      "learning_rate": 3.083499160307701e-07,
      "loss": 1.0391,
      "step": 72190
    },
    {
      "epoch": 2.925386438686412,
      "grad_norm": 0.7331106066703796,
      "learning_rate": 3.050271040620722e-07,
      "loss": 1.0748,
      "step": 72200
    },
    {
      "epoch": 2.9257916168635156,
      "grad_norm": 0.6493947505950928,
      "learning_rate": 3.0172226535697403e-07,
      "loss": 1.0686,
      "step": 72210
    },
    {
      "epoch": 2.926196795040619,
      "grad_norm": 0.7256808280944824,
      "learning_rate": 2.984354005112766e-07,
      "loss": 1.0481,
      "step": 72220
    },
    {
      "epoch": 2.9266019732177226,
      "grad_norm": 0.6886846423149109,
      "learning_rate": 2.9516651011756156e-07,
      "loss": 1.0292,
      "step": 72230
    },
    {
      "epoch": 2.927007151394826,
      "grad_norm": 0.6505215764045715,
      "learning_rate": 2.919155947651353e-07,
      "loss": 1.0464,
      "step": 72240
    },
    {
      "epoch": 2.927412329571929,
      "grad_norm": 0.6335743069648743,
      "learning_rate": 2.886826550400845e-07,
      "loss": 1.0218,
      "step": 72250
    },
    {
      "epoch": 2.9278175077490327,
      "grad_norm": 0.7032020092010498,
      "learning_rate": 2.854676915252541e-07,
      "loss": 1.032,
      "step": 72260
    },
    {
      "epoch": 2.928222685926136,
      "grad_norm": 0.6730289459228516,
      "learning_rate": 2.82270704800236e-07,
      "loss": 1.0225,
      "step": 72270
    },
    {
      "epoch": 2.9286278641032393,
      "grad_norm": 0.5900511145591736,
      "learning_rate": 2.7909169544141354e-07,
      "loss": 1.0363,
      "step": 72280
    },
    {
      "epoch": 2.929033042280343,
      "grad_norm": 0.5965068340301514,
      "learning_rate": 2.7593066402189504e-07,
      "loss": 1.0486,
      "step": 72290
    },
    {
      "epoch": 2.9294382204574463,
      "grad_norm": 0.6276459097862244,
      "learning_rate": 2.727876111115579e-07,
      "loss": 1.0824,
      "step": 72300
    },
    {
      "epoch": 2.9298433986345493,
      "grad_norm": 0.7543630003929138,
      "learning_rate": 2.6966253727703783e-07,
      "loss": 1.0888,
      "step": 72310
    },
    {
      "epoch": 2.930248576811653,
      "grad_norm": 0.6491178274154663,
      "learning_rate": 2.6655544308172853e-07,
      "loss": 1.0683,
      "step": 72320
    },
    {
      "epoch": 2.9306537549887564,
      "grad_norm": 0.653498113155365,
      "learning_rate": 2.6346632908579304e-07,
      "loss": 1.0763,
      "step": 72330
    },
    {
      "epoch": 2.9310589331658594,
      "grad_norm": 0.6984241008758545,
      "learning_rate": 2.6039519584614146e-07,
      "loss": 1.0385,
      "step": 72340
    },
    {
      "epoch": 2.931464111342963,
      "grad_norm": 0.7401130795478821,
      "learning_rate": 2.57342043916442e-07,
      "loss": 1.0723,
      "step": 72350
    },
    {
      "epoch": 2.9318692895200664,
      "grad_norm": 0.735802173614502,
      "learning_rate": 2.5430687384713215e-07,
      "loss": 1.0104,
      "step": 72360
    },
    {
      "epoch": 2.93227446769717,
      "grad_norm": 0.6855007410049438,
      "learning_rate": 2.512896861853964e-07,
      "loss": 1.0017,
      "step": 72370
    },
    {
      "epoch": 2.9326796458742734,
      "grad_norm": 0.6669368147850037,
      "learning_rate": 2.4829048147517744e-07,
      "loss": 1.035,
      "step": 72380
    },
    {
      "epoch": 2.9330848240513765,
      "grad_norm": 0.677689254283905,
      "learning_rate": 2.45309260257176e-07,
      "loss": 1.0728,
      "step": 72390
    },
    {
      "epoch": 2.93349000222848,
      "grad_norm": 0.7323001027107239,
      "learning_rate": 2.4234602306887344e-07,
      "loss": 1.0655,
      "step": 72400
    },
    {
      "epoch": 2.9338951804055835,
      "grad_norm": 0.6418764591217041,
      "learning_rate": 2.3940077044447564e-07,
      "loss": 1.0439,
      "step": 72410
    },
    {
      "epoch": 2.9343003585826866,
      "grad_norm": 0.6955676674842834,
      "learning_rate": 2.3647350291495784e-07,
      "loss": 1.0608,
      "step": 72420
    },
    {
      "epoch": 2.93470553675979,
      "grad_norm": 0.7624679207801819,
      "learning_rate": 2.3356422100805353e-07,
      "loss": 1.0683,
      "step": 72430
    },
    {
      "epoch": 2.9351107149368936,
      "grad_norm": 0.7057994604110718,
      "learning_rate": 2.3067292524826534e-07,
      "loss": 1.0298,
      "step": 72440
    },
    {
      "epoch": 2.9355158931139966,
      "grad_norm": 0.634841799736023,
      "learning_rate": 2.2779961615684297e-07,
      "loss": 1.051,
      "step": 72450
    },
    {
      "epoch": 2.9359210712911,
      "grad_norm": 0.6719446182250977,
      "learning_rate": 2.249442942517832e-07,
      "loss": 0.9998,
      "step": 72460
    },
    {
      "epoch": 2.9363262494682036,
      "grad_norm": 0.648200273513794,
      "learning_rate": 2.2210696004786314e-07,
      "loss": 1.059,
      "step": 72470
    },
    {
      "epoch": 2.936731427645307,
      "grad_norm": 0.7404764890670776,
      "learning_rate": 2.192876140565958e-07,
      "loss": 1.0454,
      "step": 72480
    },
    {
      "epoch": 2.9371366058224107,
      "grad_norm": 0.6750277876853943,
      "learning_rate": 2.1648625678627466e-07,
      "loss": 1.0067,
      "step": 72490
    },
    {
      "epoch": 2.9375417839995137,
      "grad_norm": 0.6566174626350403,
      "learning_rate": 2.137028887419068e-07,
      "loss": 1.0358,
      "step": 72500
    },
    {
      "epoch": 2.937946962176617,
      "grad_norm": 0.6380519866943359,
      "learning_rate": 2.1093751042531306e-07,
      "loss": 1.0195,
      "step": 72510
    },
    {
      "epoch": 2.9383521403537207,
      "grad_norm": 0.6481935381889343,
      "learning_rate": 2.0819012233502798e-07,
      "loss": 1.0284,
      "step": 72520
    },
    {
      "epoch": 2.938757318530824,
      "grad_norm": 0.7286896109580994,
      "learning_rate": 2.054607249663665e-07,
      "loss": 1.0925,
      "step": 72530
    },
    {
      "epoch": 2.9391624967079273,
      "grad_norm": 0.6591909527778625,
      "learning_rate": 2.0274931881137937e-07,
      "loss": 1.0351,
      "step": 72540
    },
    {
      "epoch": 2.939567674885031,
      "grad_norm": 0.6772924065589905,
      "learning_rate": 2.0005590435889786e-07,
      "loss": 1.0598,
      "step": 72550
    },
    {
      "epoch": 2.939972853062134,
      "grad_norm": 0.7096996307373047,
      "learning_rate": 1.9738048209450022e-07,
      "loss": 1.0405,
      "step": 72560
    },
    {
      "epoch": 2.9403780312392374,
      "grad_norm": 0.6833885908126831,
      "learning_rate": 1.947230525005006e-07,
      "loss": 1.0415,
      "step": 72570
    },
    {
      "epoch": 2.940783209416341,
      "grad_norm": 0.6825345754623413,
      "learning_rate": 1.9208361605599356e-07,
      "loss": 1.0345,
      "step": 72580
    },
    {
      "epoch": 2.9411883875934444,
      "grad_norm": 0.6228518486022949,
      "learning_rate": 1.894621732368429e-07,
      "loss": 1.0652,
      "step": 72590
    },
    {
      "epoch": 2.941593565770548,
      "grad_norm": 0.6373928785324097,
      "learning_rate": 1.8685872451562614e-07,
      "loss": 1.0387,
      "step": 72600
    },
    {
      "epoch": 2.941998743947651,
      "grad_norm": 0.8017003536224365,
      "learning_rate": 1.8427327036170117e-07,
      "loss": 1.0332,
      "step": 72610
    },
    {
      "epoch": 2.9424039221247544,
      "grad_norm": 0.6275177001953125,
      "learning_rate": 1.8170581124118403e-07,
      "loss": 1.0542,
      "step": 72620
    },
    {
      "epoch": 2.942809100301858,
      "grad_norm": 0.7442981004714966,
      "learning_rate": 1.7915634761695999e-07,
      "loss": 1.0313,
      "step": 72630
    },
    {
      "epoch": 2.943214278478961,
      "grad_norm": 0.705350935459137,
      "learning_rate": 1.7662487994862808e-07,
      "loss": 1.0814,
      "step": 72640
    },
    {
      "epoch": 2.9436194566560645,
      "grad_norm": 0.6568024158477783,
      "learning_rate": 1.741114086925788e-07,
      "loss": 1.0523,
      "step": 72650
    },
    {
      "epoch": 2.944024634833168,
      "grad_norm": 0.6562081575393677,
      "learning_rate": 1.7161593430193857e-07,
      "loss": 1.0241,
      "step": 72660
    },
    {
      "epoch": 2.944429813010271,
      "grad_norm": 0.7403770089149475,
      "learning_rate": 1.6913845722661414e-07,
      "loss": 1.0297,
      "step": 72670
    },
    {
      "epoch": 2.9448349911873746,
      "grad_norm": 0.6801671385765076,
      "learning_rate": 1.6667897791323718e-07,
      "loss": 1.0456,
      "step": 72680
    },
    {
      "epoch": 2.945240169364478,
      "grad_norm": 0.6003435850143433,
      "learning_rate": 1.642374968052196e-07,
      "loss": 0.9914,
      "step": 72690
    },
    {
      "epoch": 2.945645347541581,
      "grad_norm": 0.6444256901741028,
      "learning_rate": 1.6181401434270937e-07,
      "loss": 1.0065,
      "step": 72700
    },
    {
      "epoch": 2.9460505257186846,
      "grad_norm": 0.6249054670333862,
      "learning_rate": 1.5940853096262365e-07,
      "loss": 1.0052,
      "step": 72710
    },
    {
      "epoch": 2.946455703895788,
      "grad_norm": 0.6216754913330078,
      "learning_rate": 1.5702104709862664e-07,
      "loss": 1.0386,
      "step": 72720
    },
    {
      "epoch": 2.9468608820728917,
      "grad_norm": 0.6321436166763306,
      "learning_rate": 1.546515631811296e-07,
      "loss": 1.0398,
      "step": 72730
    },
    {
      "epoch": 2.947266060249995,
      "grad_norm": 0.6578000783920288,
      "learning_rate": 1.5230007963733528e-07,
      "loss": 1.0296,
      "step": 72740
    },
    {
      "epoch": 2.947671238427098,
      "grad_norm": 0.6688061356544495,
      "learning_rate": 1.49966596891149e-07,
      "loss": 1.0338,
      "step": 72750
    },
    {
      "epoch": 2.9480764166042017,
      "grad_norm": 0.6924017667770386,
      "learning_rate": 1.476511153632676e-07,
      "loss": 1.0444,
      "step": 72760
    },
    {
      "epoch": 2.9484815947813052,
      "grad_norm": 0.6300328373908997,
      "learning_rate": 1.4535363547113488e-07,
      "loss": 1.0544,
      "step": 72770
    },
    {
      "epoch": 2.9488867729584083,
      "grad_norm": 0.6478044986724854,
      "learning_rate": 1.4307415762893074e-07,
      "loss": 1.045,
      "step": 72780
    },
    {
      "epoch": 2.949291951135512,
      "grad_norm": 0.6222946047782898,
      "learning_rate": 1.4081268224762634e-07,
      "loss": 1.0033,
      "step": 72790
    },
    {
      "epoch": 2.9496971293126153,
      "grad_norm": 0.6720356345176697,
      "learning_rate": 1.3856920973492892e-07,
      "loss": 1.0503,
      "step": 72800
    },
    {
      "epoch": 2.9501023074897184,
      "grad_norm": 0.699970543384552,
      "learning_rate": 1.3634374049527054e-07,
      "loss": 1.049,
      "step": 72810
    },
    {
      "epoch": 2.950507485666822,
      "grad_norm": 0.6254333853721619,
      "learning_rate": 1.3413627492988578e-07,
      "loss": 1.0226,
      "step": 72820
    },
    {
      "epoch": 2.9509126638439254,
      "grad_norm": 0.6497865319252014,
      "learning_rate": 1.319468134367341e-07,
      "loss": 1.0081,
      "step": 72830
    },
    {
      "epoch": 2.951317842021029,
      "grad_norm": 0.6295515894889832,
      "learning_rate": 1.2977535641054417e-07,
      "loss": 1.0611,
      "step": 72840
    },
    {
      "epoch": 2.9517230201981324,
      "grad_norm": 0.6180931925773621,
      "learning_rate": 1.2762190424278063e-07,
      "loss": 1.0109,
      "step": 72850
    },
    {
      "epoch": 2.9521281983752354,
      "grad_norm": 0.7631228566169739,
      "learning_rate": 1.2548645732167742e-07,
      "loss": 1.0627,
      "step": 72860
    },
    {
      "epoch": 2.952533376552339,
      "grad_norm": 0.6450006365776062,
      "learning_rate": 1.233690160322265e-07,
      "loss": 1.0433,
      "step": 72870
    },
    {
      "epoch": 2.9529385547294424,
      "grad_norm": 0.6220306754112244,
      "learning_rate": 1.212695807561559e-07,
      "loss": 1.0499,
      "step": 72880
    },
    {
      "epoch": 2.9533437329065455,
      "grad_norm": 0.6318346858024597,
      "learning_rate": 1.1918815187196286e-07,
      "loss": 1.045,
      "step": 72890
    },
    {
      "epoch": 2.953748911083649,
      "grad_norm": 0.6775373220443726,
      "learning_rate": 1.1712472975489164e-07,
      "loss": 1.0655,
      "step": 72900
    },
    {
      "epoch": 2.9541540892607525,
      "grad_norm": 0.6835343837738037,
      "learning_rate": 1.1507931477694467e-07,
      "loss": 1.0656,
      "step": 72910
    },
    {
      "epoch": 2.9545592674378556,
      "grad_norm": 0.6507272124290466,
      "learning_rate": 1.1305190730686032e-07,
      "loss": 1.0754,
      "step": 72920
    },
    {
      "epoch": 2.954964445614959,
      "grad_norm": 0.6005121469497681,
      "learning_rate": 1.1104250771015734e-07,
      "loss": 1.0487,
      "step": 72930
    },
    {
      "epoch": 2.9553696237920626,
      "grad_norm": 0.6442080140113831,
      "learning_rate": 1.0905111634910148e-07,
      "loss": 1.042,
      "step": 72940
    },
    {
      "epoch": 2.955774801969166,
      "grad_norm": 0.6313483715057373,
      "learning_rate": 1.0707773358268336e-07,
      "loss": 1.0283,
      "step": 72950
    },
    {
      "epoch": 2.9561799801462696,
      "grad_norm": 0.6387807130813599,
      "learning_rate": 1.0512235976669616e-07,
      "loss": 1.0611,
      "step": 72960
    },
    {
      "epoch": 2.9565851583233727,
      "grad_norm": 0.6312589645385742,
      "learning_rate": 1.0318499525363568e-07,
      "loss": 1.0144,
      "step": 72970
    },
    {
      "epoch": 2.956990336500476,
      "grad_norm": 0.6310266852378845,
      "learning_rate": 1.0126564039278918e-07,
      "loss": 1.0159,
      "step": 72980
    },
    {
      "epoch": 2.9573955146775797,
      "grad_norm": 0.6653308868408203,
      "learning_rate": 9.936429553017989e-08,
      "loss": 1.0583,
      "step": 72990
    },
    {
      "epoch": 2.9578006928546827,
      "grad_norm": 0.6655440330505371,
      "learning_rate": 9.748096100858917e-08,
      "loss": 1.0642,
      "step": 73000
    },
    {
      "epoch": 2.9582058710317862,
      "grad_norm": 0.6626750826835632,
      "learning_rate": 9.56156371675454e-08,
      "loss": 1.0398,
      "step": 73010
    },
    {
      "epoch": 2.9586110492088897,
      "grad_norm": 0.6215863227844238,
      "learning_rate": 9.376832434333515e-08,
      "loss": 1.0552,
      "step": 73020
    },
    {
      "epoch": 2.959016227385993,
      "grad_norm": 0.6991133093833923,
      "learning_rate": 9.193902286900313e-08,
      "loss": 1.0785,
      "step": 73030
    },
    {
      "epoch": 2.9594214055630963,
      "grad_norm": 0.7273557782173157,
      "learning_rate": 9.012773307432998e-08,
      "loss": 1.0621,
      "step": 73040
    },
    {
      "epoch": 2.9598265837402,
      "grad_norm": 0.6827940940856934,
      "learning_rate": 8.83344552858545e-08,
      "loss": 1.0547,
      "step": 73050
    },
    {
      "epoch": 2.960231761917303,
      "grad_norm": 0.6541701555252075,
      "learning_rate": 8.655918982689581e-08,
      "loss": 1.0817,
      "step": 73060
    },
    {
      "epoch": 2.9606369400944064,
      "grad_norm": 0.6168338060379028,
      "learning_rate": 8.480193701748684e-08,
      "loss": 1.0516,
      "step": 73070
    },
    {
      "epoch": 2.96104211827151,
      "grad_norm": 0.715668261051178,
      "learning_rate": 8.306269717442971e-08,
      "loss": 1.0288,
      "step": 73080
    },
    {
      "epoch": 2.9614472964486134,
      "grad_norm": 0.674734890460968,
      "learning_rate": 8.134147061128472e-08,
      "loss": 1.0592,
      "step": 73090
    },
    {
      "epoch": 2.961852474625717,
      "grad_norm": 0.6356984972953796,
      "learning_rate": 7.963825763835919e-08,
      "loss": 1.042,
      "step": 73100
    },
    {
      "epoch": 2.96225765280282,
      "grad_norm": 0.6838613152503967,
      "learning_rate": 7.795305856270751e-08,
      "loss": 1.0588,
      "step": 73110
    },
    {
      "epoch": 2.9626628309799234,
      "grad_norm": 0.677923321723938,
      "learning_rate": 7.628587368814222e-08,
      "loss": 1.02,
      "step": 73120
    },
    {
      "epoch": 2.963068009157027,
      "grad_norm": 0.6633855700492859,
      "learning_rate": 7.463670331523398e-08,
      "loss": 1.057,
      "step": 73130
    },
    {
      "epoch": 2.96347318733413,
      "grad_norm": 0.7159264087677002,
      "learning_rate": 7.300554774128943e-08,
      "loss": 1.0923,
      "step": 73140
    },
    {
      "epoch": 2.9638783655112335,
      "grad_norm": 0.6231776475906372,
      "learning_rate": 7.139240726038443e-08,
      "loss": 1.0695,
      "step": 73150
    },
    {
      "epoch": 2.964283543688337,
      "grad_norm": 0.6638532280921936,
      "learning_rate": 6.979728216334192e-08,
      "loss": 1.0578,
      "step": 73160
    },
    {
      "epoch": 2.96468872186544,
      "grad_norm": 0.6981136202812195,
      "learning_rate": 6.822017273773184e-08,
      "loss": 1.0691,
      "step": 73170
    },
    {
      "epoch": 2.9650939000425436,
      "grad_norm": 0.7614142298698425,
      "learning_rate": 6.666107926788235e-08,
      "loss": 1.0363,
      "step": 73180
    },
    {
      "epoch": 2.965499078219647,
      "grad_norm": 0.6296616792678833,
      "learning_rate": 6.512000203485747e-08,
      "loss": 1.0565,
      "step": 73190
    },
    {
      "epoch": 2.9659042563967506,
      "grad_norm": 0.6688551306724548,
      "learning_rate": 6.359694131650162e-08,
      "loss": 1.0408,
      "step": 73200
    },
    {
      "epoch": 2.966309434573854,
      "grad_norm": 0.6536861658096313,
      "learning_rate": 6.209189738739519e-08,
      "loss": 1.0436,
      "step": 73210
    },
    {
      "epoch": 2.966714612750957,
      "grad_norm": 0.6677948832511902,
      "learning_rate": 6.060487051887664e-08,
      "loss": 1.0833,
      "step": 73220
    },
    {
      "epoch": 2.9671197909280607,
      "grad_norm": 0.652291476726532,
      "learning_rate": 5.913586097900936e-08,
      "loss": 1.0795,
      "step": 73230
    },
    {
      "epoch": 2.967524969105164,
      "grad_norm": 0.7357792854309082,
      "learning_rate": 5.7684869032648135e-08,
      "loss": 1.0838,
      "step": 73240
    },
    {
      "epoch": 2.9679301472822672,
      "grad_norm": 0.5922243595123291,
      "learning_rate": 5.625189494137262e-08,
      "loss": 1.0242,
      "step": 73250
    },
    {
      "epoch": 2.9683353254593707,
      "grad_norm": 0.6342863440513611,
      "learning_rate": 5.48369389635206e-08,
      "loss": 1.0345,
      "step": 73260
    },
    {
      "epoch": 2.9687405036364742,
      "grad_norm": 0.6364824175834656,
      "learning_rate": 5.344000135419913e-08,
      "loss": 1.0859,
      "step": 73270
    },
    {
      "epoch": 2.9691456818135773,
      "grad_norm": 0.7041483521461487,
      "learning_rate": 5.20610823652401e-08,
      "loss": 1.0689,
      "step": 73280
    },
    {
      "epoch": 2.969550859990681,
      "grad_norm": 0.7007980942726135,
      "learning_rate": 5.070018224524464e-08,
      "loss": 1.0578,
      "step": 73290
    },
    {
      "epoch": 2.9699560381677843,
      "grad_norm": 0.7764629125595093,
      "learning_rate": 4.935730123954985e-08,
      "loss": 1.0615,
      "step": 73300
    },
    {
      "epoch": 2.970361216344888,
      "grad_norm": 0.6605483889579773,
      "learning_rate": 4.8032439590262043e-08,
      "loss": 1.1045,
      "step": 73310
    },
    {
      "epoch": 2.9707663945219913,
      "grad_norm": 0.5976810455322266,
      "learning_rate": 4.672559753623462e-08,
      "loss": 1.0119,
      "step": 73320
    },
    {
      "epoch": 2.9711715726990944,
      "grad_norm": 0.6108452081680298,
      "learning_rate": 4.54367753130569e-08,
      "loss": 1.0487,
      "step": 73330
    },
    {
      "epoch": 2.971576750876198,
      "grad_norm": 0.6067768931388855,
      "learning_rate": 4.416597315307636e-08,
      "loss": 1.0303,
      "step": 73340
    },
    {
      "epoch": 2.9719819290533014,
      "grad_norm": 0.5971623063087463,
      "learning_rate": 4.2913191285420814e-08,
      "loss": 1.0588,
      "step": 73350
    },
    {
      "epoch": 2.9723871072304044,
      "grad_norm": 0.6584298610687256,
      "learning_rate": 4.167842993592075e-08,
      "loss": 1.0457,
      "step": 73360
    },
    {
      "epoch": 2.972792285407508,
      "grad_norm": 0.6887508034706116,
      "learning_rate": 4.046168932719807e-08,
      "loss": 1.0048,
      "step": 73370
    },
    {
      "epoch": 2.9731974635846115,
      "grad_norm": 0.6151015162467957,
      "learning_rate": 3.926296967859955e-08,
      "loss": 1.0631,
      "step": 73380
    },
    {
      "epoch": 2.9736026417617145,
      "grad_norm": 0.6573004722595215,
      "learning_rate": 3.808227120623009e-08,
      "loss": 1.0416,
      "step": 73390
    },
    {
      "epoch": 2.974007819938818,
      "grad_norm": 0.7294960021972656,
      "learning_rate": 3.6919594122963864e-08,
      "loss": 1.0707,
      "step": 73400
    },
    {
      "epoch": 2.9744129981159215,
      "grad_norm": 0.6661362648010254,
      "learning_rate": 3.577493863841097e-08,
      "loss": 1.0505,
      "step": 73410
    },
    {
      "epoch": 2.9748181762930246,
      "grad_norm": 0.6096512079238892,
      "learning_rate": 3.464830495891747e-08,
      "loss": 1.0353,
      "step": 73420
    },
    {
      "epoch": 2.975223354470128,
      "grad_norm": 0.6777715086936951,
      "learning_rate": 3.353969328759865e-08,
      "loss": 1.064,
      "step": 73430
    },
    {
      "epoch": 2.9756285326472316,
      "grad_norm": 0.6257754564285278,
      "learning_rate": 3.2449103824327976e-08,
      "loss": 1.0453,
      "step": 73440
    },
    {
      "epoch": 2.976033710824335,
      "grad_norm": 0.5788432359695435,
      "learning_rate": 3.1376536765714834e-08,
      "loss": 1.0587,
      "step": 73450
    },
    {
      "epoch": 2.9764388890014386,
      "grad_norm": 0.6788859367370605,
      "learning_rate": 3.032199230512678e-08,
      "loss": 1.0339,
      "step": 73460
    },
    {
      "epoch": 2.9768440671785417,
      "grad_norm": 0.6696068048477173,
      "learning_rate": 2.928547063267839e-08,
      "loss": 1.0463,
      "step": 73470
    },
    {
      "epoch": 2.977249245355645,
      "grad_norm": 0.6375132203102112,
      "learning_rate": 2.8266971935231313e-08,
      "loss": 1.0585,
      "step": 73480
    },
    {
      "epoch": 2.9776544235327487,
      "grad_norm": 0.6440410614013672,
      "learning_rate": 2.7266496396405328e-08,
      "loss": 1.077,
      "step": 73490
    },
    {
      "epoch": 2.9780596017098517,
      "grad_norm": 0.7085183262825012,
      "learning_rate": 2.628404419656727e-08,
      "loss": 1.0471,
      "step": 73500
    },
    {
      "epoch": 2.9784647798869552,
      "grad_norm": 0.6320995688438416,
      "learning_rate": 2.531961551284212e-08,
      "loss": 1.0462,
      "step": 73510
    },
    {
      "epoch": 2.9788699580640587,
      "grad_norm": 0.6423352360725403,
      "learning_rate": 2.437321051910191e-08,
      "loss": 1.0256,
      "step": 73520
    },
    {
      "epoch": 2.979275136241162,
      "grad_norm": 0.6499541997909546,
      "learning_rate": 2.3444829385954603e-08,
      "loss": 1.0803,
      "step": 73530
    },
    {
      "epoch": 2.9796803144182653,
      "grad_norm": 0.7023487091064453,
      "learning_rate": 2.2534472280788534e-08,
      "loss": 1.1021,
      "step": 73540
    },
    {
      "epoch": 2.980085492595369,
      "grad_norm": 0.6989961862564087,
      "learning_rate": 2.164213936770576e-08,
      "loss": 1.0374,
      "step": 73550
    },
    {
      "epoch": 2.9804906707724723,
      "grad_norm": 0.6772953867912292,
      "learning_rate": 2.0767830807588707e-08,
      "loss": 1.0384,
      "step": 73560
    },
    {
      "epoch": 2.980895848949576,
      "grad_norm": 0.6550063490867615,
      "learning_rate": 1.9911546758055733e-08,
      "loss": 1.0405,
      "step": 73570
    },
    {
      "epoch": 2.981301027126679,
      "grad_norm": 0.6759507060050964,
      "learning_rate": 1.9073287373494454e-08,
      "loss": 1.0682,
      "step": 73580
    },
    {
      "epoch": 2.9817062053037824,
      "grad_norm": 0.6400299072265625,
      "learning_rate": 1.825305280500622e-08,
      "loss": 1.0375,
      "step": 73590
    },
    {
      "epoch": 2.982111383480886,
      "grad_norm": 0.6851537227630615,
      "learning_rate": 1.7450843200472745e-08,
      "loss": 1.0528,
      "step": 73600
    },
    {
      "epoch": 2.982516561657989,
      "grad_norm": 0.6653602123260498,
      "learning_rate": 1.6666658704533878e-08,
      "loss": 1.0407,
      "step": 73610
    },
    {
      "epoch": 2.9829217398350925,
      "grad_norm": 0.6592442989349365,
      "learning_rate": 1.5900499458543215e-08,
      "loss": 1.0978,
      "step": 73620
    },
    {
      "epoch": 2.983326918012196,
      "grad_norm": 0.7049837708473206,
      "learning_rate": 1.5152365600634712e-08,
      "loss": 1.0193,
      "step": 73630
    },
    {
      "epoch": 2.983732096189299,
      "grad_norm": 0.6383169889450073,
      "learning_rate": 1.4422257265689353e-08,
      "loss": 1.0679,
      "step": 73640
    },
    {
      "epoch": 2.9841372743664025,
      "grad_norm": 0.7441301941871643,
      "learning_rate": 1.3710174585312984e-08,
      "loss": 1.0658,
      "step": 73650
    },
    {
      "epoch": 2.984542452543506,
      "grad_norm": 0.6271211504936218,
      "learning_rate": 1.3016117687913998e-08,
      "loss": 1.0747,
      "step": 73660
    },
    {
      "epoch": 2.9849476307206095,
      "grad_norm": 0.6747803688049316,
      "learning_rate": 1.2340086698592323e-08,
      "loss": 1.0328,
      "step": 73670
    },
    {
      "epoch": 2.985352808897713,
      "grad_norm": 0.6250914931297302,
      "learning_rate": 1.1682081739228246e-08,
      "loss": 1.0791,
      "step": 73680
    },
    {
      "epoch": 2.985757987074816,
      "grad_norm": 0.5877902507781982,
      "learning_rate": 1.1042102928460196e-08,
      "loss": 1.057,
      "step": 73690
    },
    {
      "epoch": 2.9861631652519196,
      "grad_norm": 0.6637125611305237,
      "learning_rate": 1.042015038165145e-08,
      "loss": 1.0794,
      "step": 73700
    },
    {
      "epoch": 2.986568343429023,
      "grad_norm": 0.6127917170524597,
      "learning_rate": 9.816224210934532e-09,
      "loss": 1.0233,
      "step": 73710
    },
    {
      "epoch": 2.986973521606126,
      "grad_norm": 0.6136419177055359,
      "learning_rate": 9.230324525200118e-09,
      "loss": 1.0217,
      "step": 73720
    },
    {
      "epoch": 2.9873786997832297,
      "grad_norm": 0.7262508273124695,
      "learning_rate": 8.66245143005262e-09,
      "loss": 1.0322,
      "step": 73730
    },
    {
      "epoch": 2.987783877960333,
      "grad_norm": 0.6274868845939636,
      "learning_rate": 8.112605027887909e-09,
      "loss": 1.0341,
      "step": 73740
    },
    {
      "epoch": 2.9881890561374362,
      "grad_norm": 0.6452353000640869,
      "learning_rate": 7.580785417815594e-09,
      "loss": 1.0443,
      "step": 73750
    },
    {
      "epoch": 2.9885942343145397,
      "grad_norm": 0.6604489684104919,
      "learning_rate": 7.066992695736741e-09,
      "loss": 1.0368,
      "step": 73760
    },
    {
      "epoch": 2.9889994124916432,
      "grad_norm": 0.6108314990997314,
      "learning_rate": 6.571226954266152e-09,
      "loss": 1.0676,
      "step": 73770
    },
    {
      "epoch": 2.9894045906687463,
      "grad_norm": 0.6605530381202698,
      "learning_rate": 6.0934882827767825e-09,
      "loss": 1.0198,
      "step": 73780
    },
    {
      "epoch": 2.98980976884585,
      "grad_norm": 0.6443606019020081,
      "learning_rate": 5.6337767673997344e-09,
      "loss": 1.0669,
      "step": 73790
    },
    {
      "epoch": 2.9902149470229533,
      "grad_norm": 0.6662459969520569,
      "learning_rate": 5.192092491024258e-09,
      "loss": 1.0444,
      "step": 73800
    },
    {
      "epoch": 2.990620125200057,
      "grad_norm": 0.7076582908630371,
      "learning_rate": 4.768435533264448e-09,
      "loss": 1.0744,
      "step": 73810
    },
    {
      "epoch": 2.9910253033771603,
      "grad_norm": 0.6577154994010925,
      "learning_rate": 4.362805970503647e-09,
      "loss": 1.0333,
      "step": 73820
    },
    {
      "epoch": 2.9914304815542634,
      "grad_norm": 0.706880509853363,
      "learning_rate": 3.975203875861144e-09,
      "loss": 1.0579,
      "step": 73830
    },
    {
      "epoch": 2.991835659731367,
      "grad_norm": 0.6207107305526733,
      "learning_rate": 3.605629319236581e-09,
      "loss": 1.0636,
      "step": 73840
    },
    {
      "epoch": 2.9922408379084704,
      "grad_norm": 0.6080130934715271,
      "learning_rate": 3.2540823672322363e-09,
      "loss": 1.071,
      "step": 73850
    },
    {
      "epoch": 2.9926460160855735,
      "grad_norm": 0.6367288827896118,
      "learning_rate": 2.9205630832529475e-09,
      "loss": 1.0348,
      "step": 73860
    },
    {
      "epoch": 2.993051194262677,
      "grad_norm": 0.6080479025840759,
      "learning_rate": 2.6050715274061886e-09,
      "loss": 1.0559,
      "step": 73870
    },
    {
      "epoch": 2.9934563724397805,
      "grad_norm": 0.6881083250045776,
      "learning_rate": 2.307607756568686e-09,
      "loss": 1.0594,
      "step": 73880
    },
    {
      "epoch": 2.9938615506168835,
      "grad_norm": 0.6784109473228455,
      "learning_rate": 2.0281718243753134e-09,
      "loss": 1.0256,
      "step": 73890
    },
    {
      "epoch": 2.994266728793987,
      "grad_norm": 0.5977571606636047,
      "learning_rate": 1.7667637812079917e-09,
      "loss": 1.0445,
      "step": 73900
    },
    {
      "epoch": 2.9946719069710905,
      "grad_norm": 0.7185128927230835,
      "learning_rate": 1.5233836741845863e-09,
      "loss": 1.0543,
      "step": 73910
    },
    {
      "epoch": 2.995077085148194,
      "grad_norm": 0.704338014125824,
      "learning_rate": 1.2980315471811112e-09,
      "loss": 1.0614,
      "step": 73920
    },
    {
      "epoch": 2.9954822633252975,
      "grad_norm": 0.655094563961029,
      "learning_rate": 1.0907074408428309e-09,
      "loss": 1.0733,
      "step": 73930
    },
    {
      "epoch": 2.9958874415024006,
      "grad_norm": 0.6484269499778748,
      "learning_rate": 9.014113925176481e-10,
      "loss": 1.0404,
      "step": 73940
    },
    {
      "epoch": 2.996292619679504,
      "grad_norm": 0.7279788851737976,
      "learning_rate": 7.301434363560234e-10,
      "loss": 1.0105,
      "step": 73950
    },
    {
      "epoch": 2.9966977978566076,
      "grad_norm": 0.6520195007324219,
      "learning_rate": 5.769036032332586e-10,
      "loss": 1.0499,
      "step": 73960
    },
    {
      "epoch": 2.9971029760337107,
      "grad_norm": 0.7948886752128601,
      "learning_rate": 4.416919207606007e-10,
      "loss": 1.0311,
      "step": 73970
    },
    {
      "epoch": 2.997508154210814,
      "grad_norm": 0.6282539367675781,
      "learning_rate": 3.2450841332964945e-10,
      "loss": 1.0354,
      "step": 73980
    },
    {
      "epoch": 2.9979133323879177,
      "grad_norm": 0.6510153412818909,
      "learning_rate": 2.2535310204574445e-10,
      "loss": 1.0299,
      "step": 73990
    },
    {
      "epoch": 2.9983185105650207,
      "grad_norm": 0.682633638381958,
      "learning_rate": 1.4422600481678316e-10,
      "loss": 1.0504,
      "step": 74000
    },
    {
      "epoch": 2.9987236887421242,
      "grad_norm": 0.656071662902832,
      "learning_rate": 8.112713624219836e-11,
      "loss": 1.0779,
      "step": 74010
    },
    {
      "epoch": 2.9991288669192278,
      "grad_norm": 0.6221069097518921,
      "learning_rate": 3.605650770177604e-11,
      "loss": 1.0519,
      "step": 74020
    },
    {
      "epoch": 2.9995340450963313,
      "grad_norm": 0.6510832905769348,
      "learning_rate": 9.014127333450972e-12,
      "loss": 1.0411,
      "step": 74030
    },
    {
      "epoch": 2.9999392232734348,
      "grad_norm": 0.7113467454910278,
      "learning_rate": 0.0,
      "loss": 1.0497,
      "step": 74040
    }
  ],
  "logging_steps": 10,
  "max_steps": 74040,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4159951690874937e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}