{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 148.87126296522268,
  "eval_steps": 500,
  "global_step": 30500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.09762050030506407,
      "grad_norm": 0.8231198191642761,
      "learning_rate": 4e-05,
      "loss": 2.624,
      "step": 20
    },
    {
      "epoch": 0.19524100061012814,
      "grad_norm": 1.040634274482727,
      "learning_rate": 8e-05,
      "loss": 2.5599,
      "step": 40
    },
    {
      "epoch": 0.2928615009151922,
      "grad_norm": 1.0133676528930664,
      "learning_rate": 0.00012,
      "loss": 2.1985,
      "step": 60
    },
    {
      "epoch": 0.3904820012202563,
      "grad_norm": 1.0250593423843384,
      "learning_rate": 0.00016,
      "loss": 1.9481,
      "step": 80
    },
    {
      "epoch": 0.4881025015253203,
      "grad_norm": 0.9968363046646118,
      "learning_rate": 0.0002,
      "loss": 1.7738,
      "step": 100
    },
    {
      "epoch": 0.5857230018303844,
      "grad_norm": 1.4279309511184692,
      "learning_rate": 0.0001998688524590164,
      "loss": 1.7329,
      "step": 120
    },
    {
      "epoch": 0.6833435021354485,
      "grad_norm": 1.3037158250808716,
      "learning_rate": 0.0001997377049180328,
      "loss": 1.6784,
      "step": 140
    },
    {
      "epoch": 0.7809640024405126,
      "grad_norm": 1.404618263244629,
      "learning_rate": 0.00019960655737704918,
      "loss": 1.6264,
      "step": 160
    },
    {
      "epoch": 0.8785845027455765,
      "grad_norm": 1.058852195739746,
      "learning_rate": 0.0001994754098360656,
      "loss": 1.6077,
      "step": 180
    },
    {
      "epoch": 0.9762050030506406,
      "grad_norm": 1.8023818731307983,
      "learning_rate": 0.00019934426229508198,
      "loss": 1.5252,
      "step": 200
    },
    {
      "epoch": 1.0738255033557047,
      "grad_norm": 1.2388545274734497,
      "learning_rate": 0.00019921311475409837,
      "loss": 1.5621,
      "step": 220
    },
    {
      "epoch": 1.1714460036607688,
      "grad_norm": 1.5438932180404663,
      "learning_rate": 0.00019908196721311476,
      "loss": 1.4877,
      "step": 240
    },
    {
      "epoch": 1.2690665039658329,
      "grad_norm": 1.590529203414917,
      "learning_rate": 0.00019895081967213115,
      "loss": 1.5568,
      "step": 260
    },
    {
      "epoch": 1.366687004270897,
      "grad_norm": 1.2864124774932861,
      "learning_rate": 0.00019881967213114757,
      "loss": 1.5082,
      "step": 280
    },
    {
      "epoch": 1.4643075045759608,
      "grad_norm": 1.4031420946121216,
      "learning_rate": 0.00019868852459016393,
      "loss": 1.467,
      "step": 300
    },
    {
      "epoch": 1.561928004881025,
      "grad_norm": 1.5916552543640137,
      "learning_rate": 0.00019855737704918035,
      "loss": 1.4055,
      "step": 320
    },
    {
      "epoch": 1.659548505186089,
      "grad_norm": 1.5989781618118286,
      "learning_rate": 0.00019842622950819674,
      "loss": 1.358,
      "step": 340
    },
    {
      "epoch": 1.757169005491153,
      "grad_norm": 1.7474373579025269,
      "learning_rate": 0.00019829508196721313,
      "loss": 1.3905,
      "step": 360
    },
    {
      "epoch": 1.8547895057962172,
      "grad_norm": 1.8275643587112427,
      "learning_rate": 0.00019816393442622951,
      "loss": 1.4449,
      "step": 380
    },
    {
      "epoch": 1.9524100061012812,
      "grad_norm": 1.7919280529022217,
      "learning_rate": 0.0001980327868852459,
      "loss": 1.4057,
      "step": 400
    },
    {
      "epoch": 2.0500305064063453,
      "grad_norm": 2.0150179862976074,
      "learning_rate": 0.00019790163934426232,
      "loss": 1.3659,
      "step": 420
    },
    {
      "epoch": 2.1476510067114094,
      "grad_norm": 2.0520079135894775,
      "learning_rate": 0.00019777049180327868,
      "loss": 1.3735,
      "step": 440
    },
    {
      "epoch": 2.2452715070164735,
      "grad_norm": 2.1597766876220703,
      "learning_rate": 0.0001976393442622951,
      "loss": 1.318,
      "step": 460
    },
    {
      "epoch": 2.3428920073215376,
      "grad_norm": 1.8374860286712646,
      "learning_rate": 0.0001975081967213115,
      "loss": 1.3469,
      "step": 480
    },
    {
      "epoch": 2.4405125076266017,
      "grad_norm": 2.2065134048461914,
      "learning_rate": 0.00019737704918032788,
      "loss": 1.3822,
      "step": 500
    },
    {
      "epoch": 2.5381330079316657,
      "grad_norm": 2.5370113849639893,
      "learning_rate": 0.00019724590163934427,
      "loss": 1.2652,
      "step": 520
    },
    {
      "epoch": 2.63575350823673,
      "grad_norm": 2.0484936237335205,
      "learning_rate": 0.00019711475409836066,
      "loss": 1.2971,
      "step": 540
    },
    {
      "epoch": 2.733374008541794,
      "grad_norm": 1.699357509613037,
      "learning_rate": 0.00019698360655737707,
      "loss": 1.3195,
      "step": 560
    },
    {
      "epoch": 2.830994508846858,
      "grad_norm": 2.4636785984039307,
      "learning_rate": 0.00019685245901639344,
      "loss": 1.2474,
      "step": 580
    },
    {
      "epoch": 2.9286150091519216,
      "grad_norm": 3.6711232662200928,
      "learning_rate": 0.00019672131147540985,
      "loss": 1.3091,
      "step": 600
    },
    {
      "epoch": 3.026235509456986,
      "grad_norm": 1.9547876119613647,
      "learning_rate": 0.00019659016393442624,
      "loss": 1.2966,
      "step": 620
    },
    {
      "epoch": 3.1238560097620502,
      "grad_norm": 2.135209560394287,
      "learning_rate": 0.00019645901639344263,
      "loss": 1.2201,
      "step": 640
    },
    {
      "epoch": 3.221476510067114,
      "grad_norm": 2.2378487586975098,
      "learning_rate": 0.00019632786885245902,
      "loss": 1.2226,
      "step": 660
    },
    {
      "epoch": 3.319097010372178,
      "grad_norm": 2.5407207012176514,
      "learning_rate": 0.0001961967213114754,
      "loss": 1.224,
      "step": 680
    },
    {
      "epoch": 3.416717510677242,
      "grad_norm": 2.5137102603912354,
      "learning_rate": 0.00019606557377049183,
      "loss": 1.271,
      "step": 700
    },
    {
      "epoch": 3.514338010982306,
      "grad_norm": 2.5121419429779053,
      "learning_rate": 0.00019593442622950822,
      "loss": 1.2469,
      "step": 720
    },
    {
      "epoch": 3.61195851128737,
      "grad_norm": 2.5250003337860107,
      "learning_rate": 0.0001958032786885246,
      "loss": 1.1799,
      "step": 740
    },
    {
      "epoch": 3.7095790115924343,
      "grad_norm": 1.8390239477157593,
      "learning_rate": 0.000195672131147541,
      "loss": 1.2629,
      "step": 760
    },
    {
      "epoch": 3.8071995118974984,
      "grad_norm": 2.065147638320923,
      "learning_rate": 0.00019554098360655738,
      "loss": 1.2158,
      "step": 780
    },
    {
      "epoch": 3.9048200122025625,
      "grad_norm": 2.7138381004333496,
      "learning_rate": 0.00019540983606557377,
      "loss": 1.1778,
      "step": 800
    },
    {
      "epoch": 4.002440512507627,
      "grad_norm": 3.124082565307617,
      "learning_rate": 0.00019527868852459016,
      "loss": 1.2258,
      "step": 820
    },
    {
      "epoch": 4.100061012812691,
      "grad_norm": 2.8344528675079346,
      "learning_rate": 0.00019514754098360658,
      "loss": 1.1718,
      "step": 840
    },
    {
      "epoch": 4.197681513117755,
      "grad_norm": 2.2852656841278076,
      "learning_rate": 0.00019501639344262297,
      "loss": 1.132,
      "step": 860
    },
    {
      "epoch": 4.295302013422819,
      "grad_norm": 2.662973642349243,
      "learning_rate": 0.00019488524590163936,
      "loss": 1.158,
      "step": 880
    },
    {
      "epoch": 4.392922513727883,
      "grad_norm": 2.1522879600524902,
      "learning_rate": 0.00019475409836065575,
      "loss": 1.1263,
      "step": 900
    },
    {
      "epoch": 4.490543014032947,
      "grad_norm": 2.541273832321167,
      "learning_rate": 0.00019462295081967214,
      "loss": 1.1807,
      "step": 920
    },
    {
      "epoch": 4.588163514338011,
      "grad_norm": 2.0520172119140625,
      "learning_rate": 0.00019449180327868855,
      "loss": 1.1164,
      "step": 940
    },
    {
      "epoch": 4.685784014643075,
      "grad_norm": 2.2306621074676514,
      "learning_rate": 0.00019436065573770491,
      "loss": 1.117,
      "step": 960
    },
    {
      "epoch": 4.783404514948139,
      "grad_norm": 3.0042741298675537,
      "learning_rate": 0.00019422950819672133,
      "loss": 1.2039,
      "step": 980
    },
    {
      "epoch": 4.881025015253203,
      "grad_norm": 2.1427409648895264,
      "learning_rate": 0.00019409836065573772,
      "loss": 1.1225,
      "step": 1000
    },
    {
      "epoch": 4.978645515558267,
      "grad_norm": 3.128009557723999,
      "learning_rate": 0.0001939672131147541,
      "loss": 1.1619,
      "step": 1020
    },
    {
      "epoch": 5.0762660158633315,
      "grad_norm": 3.7390689849853516,
      "learning_rate": 0.0001938360655737705,
      "loss": 1.1612,
      "step": 1040
    },
    {
      "epoch": 5.173886516168396,
      "grad_norm": 2.554919958114624,
      "learning_rate": 0.0001937049180327869,
      "loss": 1.1003,
      "step": 1060
    },
    {
      "epoch": 5.27150701647346,
      "grad_norm": 3.6076526641845703,
      "learning_rate": 0.0001935737704918033,
      "loss": 1.105,
      "step": 1080
    },
    {
      "epoch": 5.369127516778524,
      "grad_norm": 3.665255069732666,
      "learning_rate": 0.00019344262295081967,
      "loss": 1.039,
      "step": 1100
    },
    {
      "epoch": 5.466748017083588,
      "grad_norm": 2.7409827709198,
      "learning_rate": 0.00019331147540983608,
      "loss": 1.0942,
      "step": 1120
    },
    {
      "epoch": 5.564368517388652,
      "grad_norm": 2.087810754776001,
      "learning_rate": 0.00019318032786885247,
      "loss": 1.0797,
      "step": 1140
    },
    {
      "epoch": 5.661989017693716,
      "grad_norm": 2.9384937286376953,
      "learning_rate": 0.00019304918032786886,
      "loss": 1.1056,
      "step": 1160
    },
    {
      "epoch": 5.75960951799878,
      "grad_norm": 2.7793357372283936,
      "learning_rate": 0.00019291803278688525,
      "loss": 1.1048,
      "step": 1180
    },
    {
      "epoch": 5.857230018303844,
      "grad_norm": 2.1878364086151123,
      "learning_rate": 0.00019278688524590164,
      "loss": 1.1033,
      "step": 1200
    },
    {
      "epoch": 5.954850518608908,
      "grad_norm": 2.3006527423858643,
      "learning_rate": 0.00019265573770491806,
      "loss": 1.0708,
      "step": 1220
    },
    {
      "epoch": 6.052471018913972,
      "grad_norm": 2.899059295654297,
      "learning_rate": 0.00019252459016393442,
      "loss": 1.0686,
      "step": 1240
    },
    {
      "epoch": 6.150091519219036,
      "grad_norm": 2.8099234104156494,
      "learning_rate": 0.00019239344262295084,
      "loss": 0.9753,
      "step": 1260
    },
    {
      "epoch": 6.2477120195241005,
      "grad_norm": 3.555896282196045,
      "learning_rate": 0.00019226229508196723,
      "loss": 1.0251,
      "step": 1280
    },
    {
      "epoch": 6.345332519829164,
      "grad_norm": 2.860172748565674,
      "learning_rate": 0.00019213114754098362,
      "loss": 1.0532,
      "step": 1300
    },
    {
      "epoch": 6.442953020134228,
      "grad_norm": 3.038743734359741,
      "learning_rate": 0.000192,
      "loss": 1.0339,
      "step": 1320
    },
    {
      "epoch": 6.540573520439292,
      "grad_norm": 3.338714838027954,
      "learning_rate": 0.0001918688524590164,
      "loss": 1.0632,
      "step": 1340
    },
    {
      "epoch": 6.638194020744356,
      "grad_norm": 2.932199478149414,
      "learning_rate": 0.0001917377049180328,
      "loss": 1.1157,
      "step": 1360
    },
    {
      "epoch": 6.73581452104942,
      "grad_norm": 3.783275604248047,
      "learning_rate": 0.00019160655737704917,
      "loss": 1.0041,
      "step": 1380
    },
    {
      "epoch": 6.833435021354484,
      "grad_norm": 3.5673303604125977,
      "learning_rate": 0.0001914754098360656,
      "loss": 1.013,
      "step": 1400
    },
    {
      "epoch": 6.931055521659548,
      "grad_norm": 2.917269468307495,
      "learning_rate": 0.00019134426229508198,
      "loss": 1.1148,
      "step": 1420
    },
    {
      "epoch": 7.028676021964612,
      "grad_norm": 3.269557237625122,
      "learning_rate": 0.00019121311475409837,
      "loss": 0.9701,
      "step": 1440
    },
    {
      "epoch": 7.126296522269676,
      "grad_norm": 3.337118625640869,
      "learning_rate": 0.00019108196721311476,
      "loss": 0.9921,
      "step": 1460
    },
    {
      "epoch": 7.22391702257474,
      "grad_norm": 3.016709089279175,
      "learning_rate": 0.00019095081967213115,
      "loss": 0.9514,
      "step": 1480
    },
    {
      "epoch": 7.3215375228798045,
      "grad_norm": 2.8816025257110596,
      "learning_rate": 0.00019081967213114756,
      "loss": 0.9545,
      "step": 1500
    },
    {
      "epoch": 7.419158023184869,
      "grad_norm": 2.9574875831604004,
      "learning_rate": 0.00019068852459016395,
      "loss": 1.0192,
      "step": 1520
    },
    {
      "epoch": 7.516778523489933,
      "grad_norm": 2.7323741912841797,
      "learning_rate": 0.00019055737704918034,
      "loss": 0.9998,
      "step": 1540
    },
    {
      "epoch": 7.614399023794997,
      "grad_norm": 3.1870975494384766,
      "learning_rate": 0.00019042622950819673,
      "loss": 0.983,
      "step": 1560
    },
    {
      "epoch": 7.712019524100061,
      "grad_norm": 2.5255308151245117,
      "learning_rate": 0.00019029508196721312,
      "loss": 1.0236,
      "step": 1580
    },
    {
      "epoch": 7.809640024405125,
      "grad_norm": 3.190338134765625,
      "learning_rate": 0.0001901639344262295,
      "loss": 1.0263,
      "step": 1600
    },
    {
      "epoch": 7.907260524710189,
      "grad_norm": 2.7343392372131348,
      "learning_rate": 0.0001900327868852459,
      "loss": 0.981,
      "step": 1620
    },
    {
      "epoch": 8.004881025015253,
      "grad_norm": 2.4128050804138184,
      "learning_rate": 0.00018990163934426232,
      "loss": 1.0471,
      "step": 1640
    },
    {
      "epoch": 8.102501525320317,
      "grad_norm": 3.3373966217041016,
      "learning_rate": 0.0001897704918032787,
      "loss": 0.9562,
      "step": 1660
    },
    {
      "epoch": 8.200122025625381,
      "grad_norm": 4.268326282501221,
      "learning_rate": 0.0001896393442622951,
      "loss": 0.9365,
      "step": 1680
    },
    {
      "epoch": 8.297742525930445,
      "grad_norm": 4.0396728515625,
      "learning_rate": 0.00018950819672131148,
      "loss": 0.9469,
      "step": 1700
    },
    {
      "epoch": 8.39536302623551,
      "grad_norm": 2.5835509300231934,
      "learning_rate": 0.00018937704918032787,
      "loss": 0.8998,
      "step": 1720
    },
    {
      "epoch": 8.492983526540574,
      "grad_norm": 4.017479419708252,
      "learning_rate": 0.0001892459016393443,
      "loss": 0.9843,
      "step": 1740
    },
    {
      "epoch": 8.590604026845638,
      "grad_norm": 2.7477169036865234,
      "learning_rate": 0.00018911475409836065,
      "loss": 0.9164,
      "step": 1760
    },
    {
      "epoch": 8.688224527150702,
      "grad_norm": 3.388787031173706,
      "learning_rate": 0.00018898360655737707,
      "loss": 0.9626,
      "step": 1780
    },
    {
      "epoch": 8.785845027455766,
      "grad_norm": 2.938905954360962,
      "learning_rate": 0.00018885245901639346,
      "loss": 0.9662,
      "step": 1800
    },
    {
      "epoch": 8.88346552776083,
      "grad_norm": 2.854912519454956,
      "learning_rate": 0.00018872131147540985,
      "loss": 1.0023,
      "step": 1820
    },
    {
      "epoch": 8.981086028065894,
      "grad_norm": 2.120589017868042,
      "learning_rate": 0.00018859016393442624,
      "loss": 0.9616,
      "step": 1840
    },
    {
      "epoch": 9.078706528370958,
      "grad_norm": 2.8356106281280518,
      "learning_rate": 0.00018845901639344263,
      "loss": 0.9589,
      "step": 1860
    },
    {
      "epoch": 9.176327028676022,
      "grad_norm": 2.9466371536254883,
      "learning_rate": 0.00018832786885245904,
      "loss": 0.9196,
      "step": 1880
    },
    {
      "epoch": 9.273947528981086,
      "grad_norm": 3.5278165340423584,
      "learning_rate": 0.0001881967213114754,
      "loss": 0.8946,
      "step": 1900
    },
    {
      "epoch": 9.37156802928615,
      "grad_norm": 3.5878446102142334,
      "learning_rate": 0.00018806557377049182,
      "loss": 0.8855,
      "step": 1920
    },
    {
      "epoch": 9.469188529591214,
      "grad_norm": 2.8943283557891846,
      "learning_rate": 0.0001879344262295082,
      "loss": 0.9094,
      "step": 1940
    },
    {
      "epoch": 9.566809029896278,
      "grad_norm": 2.825993061065674,
      "learning_rate": 0.0001878032786885246,
      "loss": 0.9591,
      "step": 1960
    },
    {
      "epoch": 9.664429530201343,
      "grad_norm": 3.7361643314361572,
      "learning_rate": 0.000187672131147541,
      "loss": 0.9368,
      "step": 1980
    },
    {
      "epoch": 9.762050030506407,
      "grad_norm": 3.8567728996276855,
      "learning_rate": 0.00018754098360655738,
      "loss": 0.8843,
      "step": 2000
    },
    {
      "epoch": 9.85967053081147,
      "grad_norm": 3.694343090057373,
      "learning_rate": 0.0001874098360655738,
      "loss": 0.9235,
      "step": 2020
    },
    {
      "epoch": 9.957291031116535,
      "grad_norm": 4.091517448425293,
      "learning_rate": 0.00018727868852459016,
      "loss": 0.9418,
      "step": 2040
    },
    {
      "epoch": 10.054911531421599,
      "grad_norm": 3.1028010845184326,
      "learning_rate": 0.00018714754098360657,
      "loss": 0.8961,
      "step": 2060
    },
    {
      "epoch": 10.152532031726663,
      "grad_norm": 2.9279282093048096,
      "learning_rate": 0.00018701639344262296,
      "loss": 0.8457,
      "step": 2080
    },
    {
      "epoch": 10.250152532031727,
      "grad_norm": 2.750516176223755,
      "learning_rate": 0.00018688524590163935,
      "loss": 0.8941,
      "step": 2100
    },
    {
      "epoch": 10.347773032336791,
      "grad_norm": 3.211580276489258,
      "learning_rate": 0.00018675409836065574,
      "loss": 0.8799,
      "step": 2120
    },
    {
      "epoch": 10.445393532641855,
      "grad_norm": 3.7746126651763916,
      "learning_rate": 0.00018662295081967213,
      "loss": 0.8288,
      "step": 2140
    },
    {
      "epoch": 10.54301403294692,
      "grad_norm": 2.8005576133728027,
      "learning_rate": 0.00018649180327868855,
      "loss": 0.9162,
      "step": 2160
    },
    {
      "epoch": 10.640634533251983,
      "grad_norm": 2.9262828826904297,
      "learning_rate": 0.00018636065573770494,
      "loss": 0.83,
      "step": 2180
    },
    {
      "epoch": 10.738255033557047,
      "grad_norm": 4.560007095336914,
      "learning_rate": 0.00018622950819672133,
      "loss": 0.956,
      "step": 2200
    },
    {
      "epoch": 10.835875533862112,
      "grad_norm": 2.39521861076355,
      "learning_rate": 0.00018609836065573772,
      "loss": 0.9038,
      "step": 2220
    },
    {
      "epoch": 10.933496034167176,
      "grad_norm": 2.711963653564453,
      "learning_rate": 0.0001859672131147541,
      "loss": 0.9322,
      "step": 2240
    },
    {
      "epoch": 11.03111653447224,
      "grad_norm": 2.9960291385650635,
      "learning_rate": 0.0001858360655737705,
      "loss": 0.8956,
      "step": 2260
    },
    {
      "epoch": 11.128737034777304,
      "grad_norm": 3.7117059230804443,
      "learning_rate": 0.00018570491803278688,
      "loss": 0.8253,
      "step": 2280
    },
    {
      "epoch": 11.226357535082368,
      "grad_norm": 2.5004775524139404,
      "learning_rate": 0.0001855737704918033,
      "loss": 0.8075,
      "step": 2300
    },
    {
      "epoch": 11.323978035387432,
      "grad_norm": 2.690229654312134,
      "learning_rate": 0.0001854426229508197,
      "loss": 0.8528,
      "step": 2320
    },
    {
      "epoch": 11.421598535692496,
      "grad_norm": 2.5582730770111084,
      "learning_rate": 0.00018531147540983608,
      "loss": 0.9235,
      "step": 2340
    },
    {
      "epoch": 11.51921903599756,
      "grad_norm": 3.388638973236084,
      "learning_rate": 0.00018518032786885247,
      "loss": 0.8562,
      "step": 2360
    },
    {
      "epoch": 11.616839536302624,
      "grad_norm": 2.6668691635131836,
      "learning_rate": 0.00018504918032786886,
      "loss": 0.8447,
      "step": 2380
    },
    {
      "epoch": 11.714460036607688,
      "grad_norm": 2.5166635513305664,
      "learning_rate": 0.00018491803278688527,
      "loss": 0.8748,
      "step": 2400
    },
    {
      "epoch": 11.812080536912752,
      "grad_norm": 2.641568183898926,
      "learning_rate": 0.00018478688524590164,
      "loss": 0.8401,
      "step": 2420
    },
    {
      "epoch": 11.909701037217816,
      "grad_norm": 4.179554462432861,
      "learning_rate": 0.00018465573770491805,
      "loss": 0.8231,
      "step": 2440
    },
    {
      "epoch": 12.00732153752288,
      "grad_norm": 3.014526844024658,
      "learning_rate": 0.00018452459016393444,
      "loss": 0.8792,
      "step": 2460
    },
    {
      "epoch": 12.104942037827945,
      "grad_norm": 4.275758743286133,
      "learning_rate": 0.0001843934426229508,
      "loss": 0.8161,
      "step": 2480
    },
    {
      "epoch": 12.202562538133009,
      "grad_norm": 2.949131965637207,
      "learning_rate": 0.00018426229508196722,
      "loss": 0.8304,
      "step": 2500
    },
    {
      "epoch": 12.300183038438073,
      "grad_norm": 3.018010377883911,
      "learning_rate": 0.0001841311475409836,
      "loss": 0.8373,
      "step": 2520
    },
    {
      "epoch": 12.397803538743137,
      "grad_norm": 4.015832424163818,
      "learning_rate": 0.00018400000000000003,
      "loss": 0.779,
      "step": 2540
    },
    {
      "epoch": 12.495424039048201,
      "grad_norm": 3.1773481369018555,
      "learning_rate": 0.0001838688524590164,
      "loss": 0.8428,
      "step": 2560
    },
    {
      "epoch": 12.593044539353265,
      "grad_norm": 3.50673508644104,
      "learning_rate": 0.0001837377049180328,
      "loss": 0.8148,
      "step": 2580
    },
    {
      "epoch": 12.690665039658327,
      "grad_norm": 2.848912239074707,
      "learning_rate": 0.0001836065573770492,
      "loss": 0.8427,
      "step": 2600
    },
    {
      "epoch": 12.788285539963393,
      "grad_norm": 3.488842725753784,
      "learning_rate": 0.00018347540983606558,
      "loss": 0.8374,
      "step": 2620
    },
    {
      "epoch": 12.885906040268456,
      "grad_norm": 3.831639528274536,
      "learning_rate": 0.00018334426229508197,
      "loss": 0.7835,
      "step": 2640
    },
    {
      "epoch": 12.98352654057352,
      "grad_norm": 3.8596065044403076,
      "learning_rate": 0.00018321311475409836,
      "loss": 0.8719,
      "step": 2660
    },
    {
      "epoch": 13.081147040878584,
      "grad_norm": 4.338437557220459,
      "learning_rate": 0.00018308196721311478,
      "loss": 0.7768,
      "step": 2680
    },
    {
      "epoch": 13.178767541183648,
      "grad_norm": 4.147676944732666,
      "learning_rate": 0.00018295081967213114,
      "loss": 0.7686,
      "step": 2700
    },
    {
      "epoch": 13.276388041488712,
      "grad_norm": 3.430076837539673,
      "learning_rate": 0.00018281967213114756,
      "loss": 0.7069,
      "step": 2720
    },
    {
      "epoch": 13.374008541793776,
      "grad_norm": 3.6902527809143066,
      "learning_rate": 0.00018268852459016395,
      "loss": 0.8241,
      "step": 2740
    },
    {
      "epoch": 13.47162904209884,
      "grad_norm": 3.362191915512085,
      "learning_rate": 0.00018255737704918034,
      "loss": 0.8096,
      "step": 2760
    },
    {
      "epoch": 13.569249542403904,
      "grad_norm": 3.1833972930908203,
      "learning_rate": 0.00018242622950819673,
      "loss": 0.8258,
      "step": 2780
    },
    {
      "epoch": 13.666870042708968,
      "grad_norm": 3.001095771789551,
      "learning_rate": 0.00018229508196721312,
      "loss": 0.7793,
      "step": 2800
    },
    {
      "epoch": 13.764490543014032,
      "grad_norm": 4.468006134033203,
      "learning_rate": 0.00018216393442622953,
      "loss": 0.7837,
      "step": 2820
    },
    {
      "epoch": 13.862111043319096,
      "grad_norm": 3.7999494075775146,
      "learning_rate": 0.00018203278688524592,
      "loss": 0.8158,
      "step": 2840
    },
    {
      "epoch": 13.95973154362416,
      "grad_norm": 2.7308099269866943,
      "learning_rate": 0.0001819016393442623,
      "loss": 0.8657,
      "step": 2860
    },
    {
      "epoch": 14.057352043929225,
      "grad_norm": 3.631223678588867,
      "learning_rate": 0.0001817704918032787,
      "loss": 0.8219,
      "step": 2880
    },
    {
      "epoch": 14.154972544234289,
      "grad_norm": 3.7246315479278564,
      "learning_rate": 0.0001816393442622951,
      "loss": 0.738,
      "step": 2900
    },
    {
      "epoch": 14.252593044539353,
      "grad_norm": 3.8941688537597656,
      "learning_rate": 0.00018150819672131148,
      "loss": 0.7882,
      "step": 2920
    },
    {
      "epoch": 14.350213544844417,
      "grad_norm": 4.167781352996826,
      "learning_rate": 0.00018137704918032787,
      "loss": 0.7396,
      "step": 2940
    },
    {
      "epoch": 14.44783404514948,
      "grad_norm": 3.553039312362671,
      "learning_rate": 0.00018124590163934429,
      "loss": 0.7614,
      "step": 2960
    },
    {
      "epoch": 14.545454545454545,
      "grad_norm": 3.1469032764434814,
      "learning_rate": 0.00018111475409836067,
      "loss": 0.7651,
      "step": 2980
    },
    {
      "epoch": 14.643075045759609,
      "grad_norm": 5.265510559082031,
      "learning_rate": 0.00018098360655737704,
      "loss": 0.8169,
      "step": 3000
    },
    {
      "epoch": 14.740695546064673,
      "grad_norm": 2.910022020339966,
      "learning_rate": 0.00018085245901639345,
      "loss": 0.7746,
      "step": 3020
    },
    {
      "epoch": 14.838316046369737,
      "grad_norm": 2.8065860271453857,
      "learning_rate": 0.00018072131147540984,
      "loss": 0.8028,
      "step": 3040
    },
    {
      "epoch": 14.935936546674801,
      "grad_norm": 2.44608736038208,
      "learning_rate": 0.00018059016393442626,
      "loss": 0.7915,
      "step": 3060
    },
    {
      "epoch": 15.033557046979865,
      "grad_norm": 2.230102300643921,
      "learning_rate": 0.00018045901639344262,
      "loss": 0.7283,
      "step": 3080
    },
    {
      "epoch": 15.13117754728493,
      "grad_norm": 3.1589298248291016,
      "learning_rate": 0.00018032786885245904,
      "loss": 0.7117,
      "step": 3100
    },
    {
      "epoch": 15.228798047589994,
      "grad_norm": 3.1552984714508057,
      "learning_rate": 0.00018019672131147543,
      "loss": 0.7047,
      "step": 3120
    },
    {
      "epoch": 15.326418547895058,
      "grad_norm": 3.4300222396850586,
      "learning_rate": 0.0001800655737704918,
      "loss": 0.7328,
      "step": 3140
    },
    {
      "epoch": 15.424039048200122,
      "grad_norm": 2.5847089290618896,
      "learning_rate": 0.0001799344262295082,
      "loss": 0.7774,
      "step": 3160
    },
    {
      "epoch": 15.521659548505186,
      "grad_norm": 2.6597721576690674,
      "learning_rate": 0.0001798032786885246,
      "loss": 0.7427,
      "step": 3180
    },
    {
      "epoch": 15.61928004881025,
      "grad_norm": 3.029202699661255,
      "learning_rate": 0.000179672131147541,
      "loss": 0.7037,
      "step": 3200
    },
    {
      "epoch": 15.716900549115314,
      "grad_norm": 4.356511116027832,
      "learning_rate": 0.00017954098360655737,
      "loss": 0.7797,
      "step": 3220
    },
    {
      "epoch": 15.814521049420378,
      "grad_norm": 3.456430435180664,
      "learning_rate": 0.0001794098360655738,
      "loss": 0.7839,
      "step": 3240
    },
    {
      "epoch": 15.912141549725442,
      "grad_norm": 2.608675718307495,
      "learning_rate": 0.00017927868852459018,
      "loss": 0.7987,
      "step": 3260
    },
    {
      "epoch": 16.009762050030506,
      "grad_norm": 3.393676519393921,
      "learning_rate": 0.00017914754098360657,
      "loss": 0.7627,
      "step": 3280
    },
    {
      "epoch": 16.107382550335572,
      "grad_norm": 3.7134811878204346,
      "learning_rate": 0.00017901639344262296,
      "loss": 0.6441,
      "step": 3300
    },
    {
      "epoch": 16.205003050640634,
      "grad_norm": 4.507261276245117,
      "learning_rate": 0.00017888524590163935,
      "loss": 0.6972,
      "step": 3320
    },
    {
      "epoch": 16.3026235509457,
      "grad_norm": 3.150489091873169,
      "learning_rate": 0.00017875409836065576,
      "loss": 0.7079,
      "step": 3340
    },
    {
      "epoch": 16.400244051250763,
      "grad_norm": 3.351140022277832,
      "learning_rate": 0.00017862295081967213,
      "loss": 0.7319,
      "step": 3360
    },
    {
      "epoch": 16.49786455155583,
      "grad_norm": 3.533992290496826,
      "learning_rate": 0.00017849180327868852,
      "loss": 0.7095,
      "step": 3380
    },
    {
      "epoch": 16.59548505186089,
      "grad_norm": 3.086233139038086,
      "learning_rate": 0.00017836065573770493,
      "loss": 0.7653,
      "step": 3400
    },
    {
      "epoch": 16.693105552165953,
      "grad_norm": 2.891784191131592,
      "learning_rate": 0.00017822950819672132,
      "loss": 0.7096,
      "step": 3420
    },
    {
      "epoch": 16.79072605247102,
      "grad_norm": 3.5396888256073,
      "learning_rate": 0.0001780983606557377,
      "loss": 0.7385,
      "step": 3440
    },
    {
      "epoch": 16.888346552776085,
      "grad_norm": 3.5215485095977783,
      "learning_rate": 0.0001779672131147541,
      "loss": 0.791,
      "step": 3460
    },
    {
      "epoch": 16.985967053081147,
      "grad_norm": 3.2734858989715576,
      "learning_rate": 0.00017783606557377052,
      "loss": 0.7904,
      "step": 3480
    },
    {
      "epoch": 17.08358755338621,
      "grad_norm": 3.1748645305633545,
      "learning_rate": 0.0001777049180327869,
      "loss": 0.6706,
      "step": 3500
    },
    {
      "epoch": 17.181208053691275,
      "grad_norm": 2.6187920570373535,
      "learning_rate": 0.00017757377049180327,
      "loss": 0.7038,
      "step": 3520
    },
    {
      "epoch": 17.278828553996338,
      "grad_norm": 2.9449117183685303,
      "learning_rate": 0.00017744262295081969,
      "loss": 0.7214,
      "step": 3540
    },
    {
      "epoch": 17.376449054301403,
      "grad_norm": 2.324122667312622,
      "learning_rate": 0.00017731147540983607,
      "loss": 0.7377,
      "step": 3560
    },
    {
      "epoch": 17.474069554606466,
      "grad_norm": 3.7478408813476562,
      "learning_rate": 0.00017718032786885246,
      "loss": 0.6557,
      "step": 3580
    },
    {
      "epoch": 17.57169005491153,
      "grad_norm": 2.6289916038513184,
      "learning_rate": 0.00017704918032786885,
      "loss": 0.7385,
      "step": 3600
    },
    {
      "epoch": 17.669310555216594,
      "grad_norm": 3.6009535789489746,
      "learning_rate": 0.00017691803278688527,
      "loss": 0.7495,
      "step": 3620
    },
    {
      "epoch": 17.76693105552166,
      "grad_norm": 4.0030035972595215,
      "learning_rate": 0.00017678688524590166,
      "loss": 0.6941,
      "step": 3640
    },
    {
      "epoch": 17.864551555826722,
      "grad_norm": 2.4140219688415527,
      "learning_rate": 0.00017665573770491802,
      "loss": 0.6882,
      "step": 3660
    },
    {
      "epoch": 17.962172056131788,
      "grad_norm": 3.9716620445251465,
      "learning_rate": 0.00017652459016393444,
      "loss": 0.7051,
      "step": 3680
    },
    {
      "epoch": 18.05979255643685,
      "grad_norm": 3.0810763835906982,
      "learning_rate": 0.00017639344262295083,
      "loss": 0.7016,
      "step": 3700
    },
    {
      "epoch": 18.157413056741916,
      "grad_norm": 3.1639134883880615,
      "learning_rate": 0.00017626229508196724,
      "loss": 0.699,
      "step": 3720
    },
    {
      "epoch": 18.25503355704698,
      "grad_norm": 3.6788012981414795,
      "learning_rate": 0.0001761311475409836,
      "loss": 0.6673,
      "step": 3740
    },
    {
      "epoch": 18.352654057352044,
      "grad_norm": 3.2544620037078857,
      "learning_rate": 0.00017600000000000002,
      "loss": 0.7313,
      "step": 3760
    },
    {
      "epoch": 18.450274557657107,
      "grad_norm": 3.900455951690674,
      "learning_rate": 0.0001758688524590164,
      "loss": 0.7271,
      "step": 3780
    },
    {
      "epoch": 18.547895057962172,
      "grad_norm": 3.3089077472686768,
      "learning_rate": 0.00017573770491803277,
      "loss": 0.6452,
      "step": 3800
    },
    {
      "epoch": 18.645515558267235,
      "grad_norm": 3.921875,
      "learning_rate": 0.0001756065573770492,
      "loss": 0.6498,
      "step": 3820
    },
    {
      "epoch": 18.7431360585723,
      "grad_norm": 3.5202882289886475,
      "learning_rate": 0.00017547540983606558,
      "loss": 0.685,
      "step": 3840
    },
    {
      "epoch": 18.840756558877363,
      "grad_norm": 4.708593368530273,
      "learning_rate": 0.000175344262295082,
      "loss": 0.6824,
      "step": 3860
    },
    {
      "epoch": 18.93837705918243,
      "grad_norm": 3.8522069454193115,
      "learning_rate": 0.00017521311475409836,
      "loss": 0.6827,
      "step": 3880
    },
    {
      "epoch": 19.03599755948749,
      "grad_norm": 3.08516788482666,
      "learning_rate": 0.00017508196721311475,
      "loss": 0.6848,
      "step": 3900
    },
    {
      "epoch": 19.133618059792557,
      "grad_norm": 6.1903815269470215,
      "learning_rate": 0.00017495081967213116,
      "loss": 0.5924,
      "step": 3920
    },
    {
      "epoch": 19.23123856009762,
      "grad_norm": 4.409329891204834,
      "learning_rate": 0.00017481967213114753,
      "loss": 0.6484,
      "step": 3940
    },
    {
      "epoch": 19.328859060402685,
      "grad_norm": 3.1954550743103027,
      "learning_rate": 0.00017468852459016394,
      "loss": 0.6759,
      "step": 3960
    },
    {
      "epoch": 19.426479560707747,
      "grad_norm": 4.925632953643799,
      "learning_rate": 0.00017455737704918033,
      "loss": 0.6447,
      "step": 3980
    },
    {
      "epoch": 19.524100061012813,
      "grad_norm": 3.4807968139648438,
      "learning_rate": 0.00017442622950819675,
      "loss": 0.6905,
      "step": 4000
    },
    {
      "epoch": 19.621720561317876,
      "grad_norm": 3.031001567840576,
      "learning_rate": 0.0001742950819672131,
      "loss": 0.68,
      "step": 4020
    },
    {
      "epoch": 19.71934106162294,
      "grad_norm": 2.873664140701294,
      "learning_rate": 0.0001741639344262295,
      "loss": 0.6986,
      "step": 4040
    },
    {
      "epoch": 19.816961561928004,
      "grad_norm": 3.4913737773895264,
      "learning_rate": 0.00017403278688524592,
      "loss": 0.7319,
      "step": 4060
    },
    {
      "epoch": 19.91458206223307,
      "grad_norm": 4.207777976989746,
      "learning_rate": 0.0001739016393442623,
      "loss": 0.6479,
      "step": 4080
    },
    {
      "epoch": 20.012202562538132,
      "grad_norm": 2.822380781173706,
      "learning_rate": 0.0001737704918032787,
      "loss": 0.683,
      "step": 4100
    },
    {
      "epoch": 20.109823062843198,
      "grad_norm": 3.4512064456939697,
      "learning_rate": 0.00017363934426229509,
      "loss": 0.6329,
      "step": 4120
    },
    {
      "epoch": 20.20744356314826,
      "grad_norm": 4.552077293395996,
      "learning_rate": 0.0001735081967213115,
      "loss": 0.6622,
      "step": 4140
    },
    {
      "epoch": 20.305064063453326,
      "grad_norm": 3.7285406589508057,
      "learning_rate": 0.00017337704918032786,
      "loss": 0.6287,
      "step": 4160
    },
    {
      "epoch": 20.40268456375839,
      "grad_norm": 3.0254034996032715,
      "learning_rate": 0.00017324590163934425,
      "loss": 0.6696,
      "step": 4180
    },
    {
      "epoch": 20.500305064063454,
      "grad_norm": 3.474151372909546,
      "learning_rate": 0.00017311475409836067,
      "loss": 0.623,
      "step": 4200
    },
    {
      "epoch": 20.597925564368516,
      "grad_norm": 3.686966896057129,
      "learning_rate": 0.00017298360655737706,
      "loss": 0.7025,
      "step": 4220
    },
    {
      "epoch": 20.695546064673582,
      "grad_norm": 4.044944763183594,
      "learning_rate": 0.00017285245901639345,
      "loss": 0.6287,
      "step": 4240
    },
    {
      "epoch": 20.793166564978645,
      "grad_norm": 4.093704700469971,
      "learning_rate": 0.00017272131147540984,
      "loss": 0.6523,
      "step": 4260
    },
    {
      "epoch": 20.89078706528371,
      "grad_norm": 3.9666953086853027,
      "learning_rate": 0.00017259016393442625,
      "loss": 0.6513,
      "step": 4280
    },
    {
      "epoch": 20.988407565588773,
      "grad_norm": 3.888939142227173,
      "learning_rate": 0.00017245901639344264,
      "loss": 0.682,
      "step": 4300
    },
    {
      "epoch": 21.08602806589384,
      "grad_norm": 3.227346420288086,
      "learning_rate": 0.000172327868852459,
      "loss": 0.6342,
      "step": 4320
    },
    {
      "epoch": 21.1836485661989,
      "grad_norm": 2.9979476928710938,
      "learning_rate": 0.00017219672131147542,
      "loss": 0.6028,
      "step": 4340
    },
    {
      "epoch": 21.281269066503967,
      "grad_norm": 3.6194779872894287,
      "learning_rate": 0.0001720655737704918,
      "loss": 0.6294,
      "step": 4360
    },
    {
      "epoch": 21.37888956680903,
      "grad_norm": 4.482040882110596,
      "learning_rate": 0.0001719344262295082,
      "loss": 0.6283,
      "step": 4380
    },
    {
      "epoch": 21.476510067114095,
      "grad_norm": 3.0028133392333984,
      "learning_rate": 0.0001718032786885246,
      "loss": 0.6292,
      "step": 4400
    },
    {
      "epoch": 21.574130567419157,
      "grad_norm": 3.0038769245147705,
      "learning_rate": 0.00017167213114754098,
      "loss": 0.6349,
      "step": 4420
    },
    {
      "epoch": 21.671751067724223,
      "grad_norm": 2.854794502258301,
      "learning_rate": 0.0001715409836065574,
      "loss": 0.632,
      "step": 4440
    },
    {
      "epoch": 21.769371568029285,
      "grad_norm": 4.145231246948242,
      "learning_rate": 0.00017140983606557376,
      "loss": 0.6518,
      "step": 4460
    },
    {
      "epoch": 21.86699206833435,
      "grad_norm": 2.7917041778564453,
      "learning_rate": 0.00017127868852459018,
      "loss": 0.6595,
      "step": 4480
    },
    {
      "epoch": 21.964612568639414,
      "grad_norm": 2.858912706375122,
      "learning_rate": 0.00017114754098360656,
      "loss": 0.6507,
      "step": 4500
    },
    {
      "epoch": 22.06223306894448,
      "grad_norm": 4.860498428344727,
      "learning_rate": 0.00017101639344262298,
      "loss": 0.6025,
      "step": 4520
    },
    {
      "epoch": 22.15985356924954,
      "grad_norm": 3.04990291595459,
      "learning_rate": 0.00017088524590163934,
      "loss": 0.5384,
      "step": 4540
    },
    {
      "epoch": 22.257474069554608,
      "grad_norm": 3.768587827682495,
      "learning_rate": 0.00017075409836065573,
      "loss": 0.5799,
      "step": 4560
    },
    {
      "epoch": 22.35509456985967,
      "grad_norm": 3.870783805847168,
      "learning_rate": 0.00017062295081967215,
      "loss": 0.6453,
      "step": 4580
    },
    {
      "epoch": 22.452715070164736,
      "grad_norm": 4.49683952331543,
      "learning_rate": 0.0001704918032786885,
      "loss": 0.6728,
      "step": 4600
    },
    {
      "epoch": 22.550335570469798,
      "grad_norm": 3.2914700508117676,
      "learning_rate": 0.00017036065573770493,
      "loss": 0.6644,
      "step": 4620
    },
    {
      "epoch": 22.647956070774864,
      "grad_norm": 2.4408113956451416,
      "learning_rate": 0.00017022950819672132,
      "loss": 0.6034,
      "step": 4640
    },
    {
      "epoch": 22.745576571079926,
      "grad_norm": 4.22420072555542,
      "learning_rate": 0.00017009836065573773,
      "loss": 0.6464,
      "step": 4660
    },
    {
      "epoch": 22.843197071384992,
      "grad_norm": 3.422456979751587,
      "learning_rate": 0.0001699672131147541,
      "loss": 0.6017,
      "step": 4680
    },
    {
      "epoch": 22.940817571690054,
      "grad_norm": 2.6850616931915283,
      "learning_rate": 0.00016983606557377049,
      "loss": 0.6059,
      "step": 4700
    },
    {
      "epoch": 23.03843807199512,
      "grad_norm": 4.270930290222168,
      "learning_rate": 0.0001697049180327869,
      "loss": 0.6273,
      "step": 4720
    },
    {
      "epoch": 23.136058572300183,
      "grad_norm": 3.8276960849761963,
      "learning_rate": 0.0001695737704918033,
      "loss": 0.5779,
      "step": 4740
    },
    {
      "epoch": 23.23367907260525,
      "grad_norm": 4.058398723602295,
      "learning_rate": 0.00016944262295081968,
      "loss": 0.593,
      "step": 4760
    },
    {
      "epoch": 23.33129957291031,
      "grad_norm": 4.362659454345703,
      "learning_rate": 0.00016931147540983607,
      "loss": 0.5927,
      "step": 4780
    },
    {
      "epoch": 23.428920073215377,
      "grad_norm": 3.7719762325286865,
      "learning_rate": 0.00016918032786885249,
      "loss": 0.6487,
      "step": 4800
    },
    {
      "epoch": 23.52654057352044,
      "grad_norm": 2.9975759983062744,
      "learning_rate": 0.00016904918032786885,
      "loss": 0.614,
      "step": 4820
    },
    {
      "epoch": 23.624161073825505,
      "grad_norm": 3.0442936420440674,
      "learning_rate": 0.00016891803278688524,
      "loss": 0.5966,
      "step": 4840
    },
    {
      "epoch": 23.721781574130567,
      "grad_norm": 2.8749730587005615,
      "learning_rate": 0.00016878688524590165,
      "loss": 0.6166,
      "step": 4860
    },
    {
      "epoch": 23.819402074435633,
      "grad_norm": 3.6400644779205322,
      "learning_rate": 0.00016865573770491804,
      "loss": 0.6329,
      "step": 4880
    },
    {
      "epoch": 23.917022574740695,
      "grad_norm": 3.3705861568450928,
      "learning_rate": 0.00016852459016393443,
      "loss": 0.5753,
      "step": 4900
    },
    {
      "epoch": 24.01464307504576,
      "grad_norm": 2.9685709476470947,
      "learning_rate": 0.00016839344262295082,
      "loss": 0.6302,
      "step": 4920
    },
    {
      "epoch": 24.112263575350823,
      "grad_norm": 3.0799593925476074,
      "learning_rate": 0.0001682622950819672,
      "loss": 0.5936,
      "step": 4940
    },
    {
      "epoch": 24.20988407565589,
      "grad_norm": 2.598597764968872,
      "learning_rate": 0.00016813114754098363,
      "loss": 0.5836,
      "step": 4960
    },
    {
      "epoch": 24.30750457596095,
      "grad_norm": 3.719918966293335,
      "learning_rate": 0.000168,
      "loss": 0.5754,
      "step": 4980
    },
    {
      "epoch": 24.405125076266017,
      "grad_norm": 3.592268466949463,
      "learning_rate": 0.0001678688524590164,
      "loss": 0.5683,
      "step": 5000
    },
    {
      "epoch": 24.50274557657108,
      "grad_norm": 2.984259605407715,
      "learning_rate": 0.0001677377049180328,
      "loss": 0.5795,
      "step": 5020
    },
    {
      "epoch": 24.600366076876146,
      "grad_norm": 2.833623170852661,
      "learning_rate": 0.00016760655737704919,
      "loss": 0.618,
      "step": 5040
    },
    {
      "epoch": 24.697986577181208,
      "grad_norm": 5.3122172355651855,
      "learning_rate": 0.00016747540983606558,
      "loss": 0.5863,
      "step": 5060
    },
    {
      "epoch": 24.795607077486274,
      "grad_norm": 2.9093785285949707,
      "learning_rate": 0.00016734426229508196,
      "loss": 0.5863,
      "step": 5080
    },
    {
      "epoch": 24.893227577791336,
      "grad_norm": 2.8090052604675293,
      "learning_rate": 0.00016721311475409838,
      "loss": 0.6045,
      "step": 5100
    },
    {
      "epoch": 24.990848078096402,
      "grad_norm": 2.9333415031433105,
      "learning_rate": 0.00016708196721311474,
      "loss": 0.612,
      "step": 5120
    },
    {
      "epoch": 25.088468578401464,
      "grad_norm": 2.933706760406494,
      "learning_rate": 0.00016695081967213116,
      "loss": 0.5534,
      "step": 5140
    },
    {
      "epoch": 25.18608907870653,
      "grad_norm": 4.508391380310059,
      "learning_rate": 0.00016681967213114755,
      "loss": 0.5384,
      "step": 5160
    },
    {
      "epoch": 25.283709579011592,
      "grad_norm": 4.106027603149414,
      "learning_rate": 0.00016668852459016397,
      "loss": 0.5464,
      "step": 5180
    },
    {
      "epoch": 25.381330079316655,
      "grad_norm": 4.005650997161865,
      "learning_rate": 0.00016655737704918033,
      "loss": 0.5805,
      "step": 5200
    },
    {
      "epoch": 25.47895057962172,
      "grad_norm": 3.259774923324585,
      "learning_rate": 0.00016642622950819672,
      "loss": 0.5966,
      "step": 5220
    },
    {
      "epoch": 25.576571079926783,
      "grad_norm": 3.695301055908203,
      "learning_rate": 0.00016629508196721313,
      "loss": 0.5344,
      "step": 5240
    },
    {
      "epoch": 25.67419158023185,
      "grad_norm": 2.9240682125091553,
      "learning_rate": 0.0001661639344262295,
      "loss": 0.6109,
      "step": 5260
    },
    {
      "epoch": 25.77181208053691,
      "grad_norm": 5.290099620819092,
      "learning_rate": 0.0001660327868852459,
      "loss": 0.5831,
      "step": 5280
    },
    {
      "epoch": 25.869432580841977,
      "grad_norm": 3.4903934001922607,
      "learning_rate": 0.0001659016393442623,
      "loss": 0.6106,
      "step": 5300
    },
    {
      "epoch": 25.96705308114704,
      "grad_norm": 4.101973533630371,
      "learning_rate": 0.0001657704918032787,
      "loss": 0.6319,
      "step": 5320
    },
    {
      "epoch": 26.064673581452105,
      "grad_norm": 3.657115936279297,
      "learning_rate": 0.00016563934426229508,
      "loss": 0.5218,
      "step": 5340
    },
    {
      "epoch": 26.162294081757167,
      "grad_norm": 3.4918181896209717,
      "learning_rate": 0.00016550819672131147,
      "loss": 0.5848,
      "step": 5360
    },
    {
      "epoch": 26.259914582062233,
      "grad_norm": 3.117476224899292,
      "learning_rate": 0.00016537704918032789,
      "loss": 0.5579,
      "step": 5380
    },
    {
      "epoch": 26.357535082367296,
      "grad_norm": 4.165419578552246,
      "learning_rate": 0.00016524590163934428,
      "loss": 0.535,
      "step": 5400
    },
    {
      "epoch": 26.45515558267236,
      "grad_norm": 4.682600021362305,
      "learning_rate": 0.00016511475409836067,
      "loss": 0.6043,
      "step": 5420
    },
    {
      "epoch": 26.552776082977424,
      "grad_norm": 3.7358362674713135,
      "learning_rate": 0.00016498360655737705,
      "loss": 0.5605,
      "step": 5440
    },
    {
      "epoch": 26.65039658328249,
      "grad_norm": 3.1440377235412598,
      "learning_rate": 0.00016485245901639344,
      "loss": 0.5344,
      "step": 5460
    },
    {
      "epoch": 26.748017083587552,
      "grad_norm": 4.453829288482666,
      "learning_rate": 0.00016472131147540983,
      "loss": 0.564,
      "step": 5480
    },
    {
      "epoch": 26.845637583892618,
      "grad_norm": 3.959137439727783,
      "learning_rate": 0.00016459016393442622,
      "loss": 0.5986,
      "step": 5500
    },
    {
      "epoch": 26.94325808419768,
      "grad_norm": 3.9074790477752686,
      "learning_rate": 0.00016445901639344264,
      "loss": 0.5789,
      "step": 5520
    },
    {
      "epoch": 27.040878584502746,
      "grad_norm": 2.9816102981567383,
      "learning_rate": 0.00016432786885245903,
      "loss": 0.5651,
      "step": 5540
    },
    {
      "epoch": 27.13849908480781,
      "grad_norm": 2.41890549659729,
      "learning_rate": 0.00016419672131147542,
      "loss": 0.5405,
      "step": 5560
    },
    {
      "epoch": 27.236119585112874,
      "grad_norm": 3.1744871139526367,
      "learning_rate": 0.0001640655737704918,
      "loss": 0.5646,
      "step": 5580
    },
    {
      "epoch": 27.333740085417936,
      "grad_norm": 3.6110246181488037,
      "learning_rate": 0.0001639344262295082,
      "loss": 0.5532,
      "step": 5600
    },
    {
      "epoch": 27.431360585723002,
      "grad_norm": 4.252525806427002,
      "learning_rate": 0.0001638032786885246,
      "loss": 0.5243,
      "step": 5620
    },
    {
      "epoch": 27.528981086028065,
      "grad_norm": 3.710491895675659,
      "learning_rate": 0.00016367213114754098,
      "loss": 0.57,
      "step": 5640
    },
    {
      "epoch": 27.62660158633313,
      "grad_norm": 3.5450077056884766,
      "learning_rate": 0.0001635409836065574,
      "loss": 0.5401,
      "step": 5660
    },
    {
      "epoch": 27.724222086638193,
      "grad_norm": 3.5488388538360596,
      "learning_rate": 0.00016340983606557378,
      "loss": 0.5345,
      "step": 5680
    },
    {
      "epoch": 27.82184258694326,
      "grad_norm": 4.381565570831299,
      "learning_rate": 0.00016327868852459017,
      "loss": 0.5362,
      "step": 5700
    },
    {
      "epoch": 27.91946308724832,
      "grad_norm": 4.077484130859375,
      "learning_rate": 0.00016314754098360656,
      "loss": 0.6039,
      "step": 5720
    },
    {
      "epoch": 28.017083587553387,
      "grad_norm": 3.217458486557007,
      "learning_rate": 0.00016301639344262295,
      "loss": 0.6027,
      "step": 5740
    },
    {
      "epoch": 28.11470408785845,
      "grad_norm": 3.741176128387451,
      "learning_rate": 0.00016288524590163937,
      "loss": 0.4955,
      "step": 5760
    },
    {
      "epoch": 28.212324588163515,
      "grad_norm": 4.43143892288208,
      "learning_rate": 0.00016275409836065573,
      "loss": 0.5326,
      "step": 5780
    },
    {
      "epoch": 28.309945088468577,
      "grad_norm": 3.040767192840576,
      "learning_rate": 0.00016262295081967214,
      "loss": 0.5207,
      "step": 5800
    },
    {
      "epoch": 28.407565588773643,
      "grad_norm": 3.47582745552063,
      "learning_rate": 0.00016249180327868853,
      "loss": 0.5524,
      "step": 5820
    },
    {
      "epoch": 28.505186089078705,
      "grad_norm": 4.133968830108643,
      "learning_rate": 0.00016236065573770492,
      "loss": 0.5553,
      "step": 5840
    },
    {
      "epoch": 28.60280658938377,
      "grad_norm": 3.2506773471832275,
      "learning_rate": 0.0001622295081967213,
      "loss": 0.5287,
      "step": 5860
    },
    {
      "epoch": 28.700427089688834,
      "grad_norm": 4.527762413024902,
      "learning_rate": 0.0001620983606557377,
      "loss": 0.507,
      "step": 5880
    },
    {
      "epoch": 28.7980475899939,
      "grad_norm": 4.723377704620361,
      "learning_rate": 0.00016196721311475412,
      "loss": 0.6089,
      "step": 5900
    },
    {
      "epoch": 28.89566809029896,
      "grad_norm": 3.6172468662261963,
      "learning_rate": 0.00016183606557377048,
      "loss": 0.542,
      "step": 5920
    },
    {
      "epoch": 28.993288590604028,
      "grad_norm": 3.735133647918701,
      "learning_rate": 0.0001617049180327869,
      "loss": 0.5656,
      "step": 5940
    },
    {
      "epoch": 29.09090909090909,
      "grad_norm": 5.235744476318359,
      "learning_rate": 0.0001615737704918033,
      "loss": 0.5355,
      "step": 5960
    },
    {
      "epoch": 29.188529591214156,
      "grad_norm": 4.541784763336182,
      "learning_rate": 0.00016144262295081968,
      "loss": 0.5122,
      "step": 5980
    },
    {
      "epoch": 29.286150091519218,
      "grad_norm": 3.0607311725616455,
      "learning_rate": 0.00016131147540983607,
      "loss": 0.5047,
      "step": 6000
    },
    {
      "epoch": 29.383770591824284,
      "grad_norm": 4.826447486877441,
      "learning_rate": 0.00016118032786885245,
      "loss": 0.5077,
      "step": 6020
    },
    {
      "epoch": 29.481391092129346,
      "grad_norm": 3.6418137550354004,
      "learning_rate": 0.00016104918032786887,
      "loss": 0.5401,
      "step": 6040
    },
    {
      "epoch": 29.579011592434412,
      "grad_norm": 6.086111545562744,
      "learning_rate": 0.00016091803278688526,
      "loss": 0.5311,
      "step": 6060
    },
    {
      "epoch": 29.676632092739474,
      "grad_norm": 2.81947922706604,
      "learning_rate": 0.00016078688524590165,
      "loss": 0.523,
      "step": 6080
    },
    {
      "epoch": 29.77425259304454,
      "grad_norm": 4.2681355476379395,
      "learning_rate": 0.00016065573770491804,
      "loss": 0.5439,
      "step": 6100
    },
    {
      "epoch": 29.871873093349603,
      "grad_norm": 3.6135213375091553,
      "learning_rate": 0.00016052459016393443,
      "loss": 0.5177,
      "step": 6120
    },
    {
      "epoch": 29.96949359365467,
      "grad_norm": 3.3115713596343994,
      "learning_rate": 0.00016039344262295082,
      "loss": 0.581,
      "step": 6140
    },
    {
      "epoch": 30.06711409395973,
      "grad_norm": 3.5007998943328857,
      "learning_rate": 0.0001602622950819672,
      "loss": 0.5092,
      "step": 6160
    },
    {
      "epoch": 30.164734594264797,
      "grad_norm": 2.9520623683929443,
      "learning_rate": 0.00016013114754098362,
      "loss": 0.4636,
      "step": 6180
    },
    {
      "epoch": 30.26235509456986,
      "grad_norm": 3.135349750518799,
      "learning_rate": 0.00016,
      "loss": 0.5391,
      "step": 6200
    },
    {
      "epoch": 30.359975594874925,
      "grad_norm": 4.124795913696289,
      "learning_rate": 0.0001598688524590164,
      "loss": 0.5185,
      "step": 6220
    },
    {
      "epoch": 30.457596095179987,
      "grad_norm": 3.6246888637542725,
      "learning_rate": 0.0001597377049180328,
      "loss": 0.532,
      "step": 6240
    },
    {
      "epoch": 30.555216595485053,
      "grad_norm": 3.1558213233947754,
      "learning_rate": 0.00015960655737704918,
      "loss": 0.5261,
      "step": 6260
    },
    {
      "epoch": 30.652837095790115,
      "grad_norm": 3.2610013484954834,
      "learning_rate": 0.0001594754098360656,
      "loss": 0.5104,
      "step": 6280
    },
    {
      "epoch": 30.75045759609518,
      "grad_norm": 3.8648669719696045,
      "learning_rate": 0.00015934426229508196,
      "loss": 0.538,
      "step": 6300
    },
    {
      "epoch": 30.848078096400243,
      "grad_norm": 3.0972094535827637,
      "learning_rate": 0.00015921311475409838,
      "loss": 0.5362,
      "step": 6320
    },
    {
      "epoch": 30.94569859670531,
      "grad_norm": 4.238556385040283,
      "learning_rate": 0.00015908196721311477,
      "loss": 0.54,
      "step": 6340
    },
    {
      "epoch": 31.04331909701037,
      "grad_norm": 4.820079326629639,
      "learning_rate": 0.00015895081967213116,
      "loss": 0.4893,
      "step": 6360
    },
    {
      "epoch": 31.140939597315437,
      "grad_norm": 3.2895243167877197,
      "learning_rate": 0.00015881967213114754,
      "loss": 0.4502,
      "step": 6380
    },
    {
      "epoch": 31.2385600976205,
      "grad_norm": 3.5625295639038086,
      "learning_rate": 0.00015868852459016393,
      "loss": 0.5033,
      "step": 6400
    },
    {
      "epoch": 31.336180597925566,
      "grad_norm": 3.8144872188568115,
      "learning_rate": 0.00015855737704918035,
      "loss": 0.515,
      "step": 6420
    },
    {
      "epoch": 31.433801098230628,
      "grad_norm": 4.3353271484375,
      "learning_rate": 0.0001584262295081967,
      "loss": 0.5356,
      "step": 6440
    },
    {
      "epoch": 31.531421598535694,
      "grad_norm": 4.205932140350342,
      "learning_rate": 0.00015829508196721313,
      "loss": 0.5295,
      "step": 6460
    },
    {
      "epoch": 31.629042098840756,
      "grad_norm": 4.171141147613525,
      "learning_rate": 0.00015816393442622952,
      "loss": 0.4502,
      "step": 6480
    },
    {
      "epoch": 31.726662599145822,
      "grad_norm": 3.7091734409332275,
      "learning_rate": 0.0001580327868852459,
      "loss": 0.4885,
      "step": 6500
    },
    {
      "epoch": 31.824283099450884,
      "grad_norm": 3.9756462574005127,
      "learning_rate": 0.0001579016393442623,
      "loss": 0.546,
      "step": 6520
    },
    {
      "epoch": 31.92190359975595,
      "grad_norm": 4.506149768829346,
      "learning_rate": 0.0001577704918032787,
      "loss": 0.513,
      "step": 6540
    },
    {
      "epoch": 32.01952410006101,
      "grad_norm": 3.1480486392974854,
      "learning_rate": 0.0001576393442622951,
      "loss": 0.5799,
      "step": 6560
    },
    {
      "epoch": 32.117144600366075,
      "grad_norm": 2.476172685623169,
      "learning_rate": 0.00015750819672131147,
      "loss": 0.455,
      "step": 6580
    },
    {
      "epoch": 32.214765100671144,
      "grad_norm": 3.0843114852905273,
      "learning_rate": 0.00015737704918032788,
      "loss": 0.4701,
      "step": 6600
    },
    {
      "epoch": 32.31238560097621,
      "grad_norm": 3.4512338638305664,
      "learning_rate": 0.00015724590163934427,
      "loss": 0.4908,
      "step": 6620
    },
    {
      "epoch": 32.41000610128127,
      "grad_norm": 3.0433006286621094,
      "learning_rate": 0.00015711475409836066,
      "loss": 0.4857,
      "step": 6640
    },
    {
      "epoch": 32.50762660158633,
      "grad_norm": 3.1231653690338135,
      "learning_rate": 0.00015698360655737705,
      "loss": 0.5053,
      "step": 6660
    },
    {
      "epoch": 32.6052471018914,
      "grad_norm": 3.121457576751709,
      "learning_rate": 0.00015685245901639344,
      "loss": 0.4962,
      "step": 6680
    },
    {
      "epoch": 32.70286760219646,
      "grad_norm": 3.6397838592529297,
      "learning_rate": 0.00015672131147540986,
      "loss": 0.5083,
      "step": 6700
    },
    {
      "epoch": 32.800488102501525,
      "grad_norm": 4.634298801422119,
      "learning_rate": 0.00015659016393442622,
      "loss": 0.5268,
      "step": 6720
    },
    {
      "epoch": 32.89810860280659,
      "grad_norm": 3.6741957664489746,
      "learning_rate": 0.00015645901639344263,
      "loss": 0.5518,
      "step": 6740
    },
    {
      "epoch": 32.99572910311166,
      "grad_norm": 3.7481977939605713,
      "learning_rate": 0.00015632786885245902,
      "loss": 0.5242,
      "step": 6760
    },
    {
      "epoch": 33.09334960341672,
      "grad_norm": 3.2336630821228027,
      "learning_rate": 0.0001561967213114754,
      "loss": 0.4854,
      "step": 6780
    },
    {
      "epoch": 33.19097010372178,
      "grad_norm": 3.153439521789551,
      "learning_rate": 0.0001560655737704918,
      "loss": 0.4512,
      "step": 6800
    },
    {
      "epoch": 33.288590604026844,
      "grad_norm": 4.410555839538574,
      "learning_rate": 0.0001559344262295082,
      "loss": 0.4135,
      "step": 6820
    },
    {
      "epoch": 33.38621110433191,
      "grad_norm": 4.749181747436523,
      "learning_rate": 0.0001558032786885246,
      "loss": 0.5023,
      "step": 6840
    },
    {
      "epoch": 33.483831604636975,
      "grad_norm": 5.4415459632873535,
      "learning_rate": 0.000155672131147541,
      "loss": 0.4793,
      "step": 6860
    },
    {
      "epoch": 33.58145210494204,
      "grad_norm": 4.493134498596191,
      "learning_rate": 0.0001555409836065574,
      "loss": 0.4827,
      "step": 6880
    },
    {
      "epoch": 33.6790726052471,
      "grad_norm": 5.345827579498291,
      "learning_rate": 0.00015540983606557378,
      "loss": 0.4878,
      "step": 6900
    },
    {
      "epoch": 33.77669310555217,
      "grad_norm": 5.483647346496582,
      "learning_rate": 0.00015527868852459017,
      "loss": 0.4976,
      "step": 6920
    },
    {
      "epoch": 33.87431360585723,
      "grad_norm": 3.2896759510040283,
      "learning_rate": 0.00015514754098360656,
      "loss": 0.5163,
      "step": 6940
    },
    {
      "epoch": 33.971934106162294,
      "grad_norm": 3.426711082458496,
      "learning_rate": 0.00015501639344262294,
      "loss": 0.6068,
      "step": 6960
    },
    {
      "epoch": 34.06955460646736,
      "grad_norm": 5.182300567626953,
      "learning_rate": 0.00015488524590163936,
      "loss": 0.4459,
      "step": 6980
    },
    {
      "epoch": 34.16717510677242,
      "grad_norm": 3.9542787075042725,
      "learning_rate": 0.00015475409836065575,
      "loss": 0.4688,
      "step": 7000
    },
    {
      "epoch": 34.26479560707749,
      "grad_norm": 3.966188669204712,
      "learning_rate": 0.00015462295081967214,
      "loss": 0.4913,
      "step": 7020
    },
    {
      "epoch": 34.36241610738255,
      "grad_norm": 3.175008535385132,
      "learning_rate": 0.00015449180327868853,
      "loss": 0.4564,
      "step": 7040
    },
    {
      "epoch": 34.46003660768761,
      "grad_norm": 5.1779584884643555,
      "learning_rate": 0.00015436065573770492,
      "loss": 0.4807,
      "step": 7060
    },
    {
      "epoch": 34.557657107992675,
      "grad_norm": 3.4940545558929443,
      "learning_rate": 0.00015422950819672133,
      "loss": 0.4948,
      "step": 7080
    },
    {
      "epoch": 34.655277608297745,
      "grad_norm": 4.1444196701049805,
      "learning_rate": 0.0001540983606557377,
      "loss": 0.4756,
      "step": 7100
    },
    {
      "epoch": 34.75289810860281,
      "grad_norm": 5.219560146331787,
      "learning_rate": 0.00015396721311475411,
      "loss": 0.5207,
      "step": 7120
    },
    {
      "epoch": 34.85051860890787,
      "grad_norm": 3.7527666091918945,
      "learning_rate": 0.0001538360655737705,
      "loss": 0.484,
      "step": 7140
    },
    {
      "epoch": 34.94813910921293,
      "grad_norm": 3.0565104484558105,
      "learning_rate": 0.0001537049180327869,
      "loss": 0.4795,
      "step": 7160
    },
    {
      "epoch": 35.045759609518,
      "grad_norm": 2.8091022968292236,
      "learning_rate": 0.00015357377049180328,
      "loss": 0.4657,
      "step": 7180
    },
    {
      "epoch": 35.14338010982306,
      "grad_norm": 2.8542001247406006,
      "learning_rate": 0.00015344262295081967,
      "loss": 0.4532,
      "step": 7200
    },
    {
      "epoch": 35.241000610128125,
      "grad_norm": 3.4882893562316895,
      "learning_rate": 0.0001533114754098361,
      "loss": 0.4629,
      "step": 7220
    },
    {
      "epoch": 35.33862111043319,
      "grad_norm": 3.3786075115203857,
      "learning_rate": 0.00015318032786885245,
      "loss": 0.4388,
      "step": 7240
    },
    {
      "epoch": 35.43624161073826,
      "grad_norm": 4.602392673492432,
      "learning_rate": 0.00015304918032786887,
      "loss": 0.4788,
      "step": 7260
    },
    {
      "epoch": 35.53386211104332,
      "grad_norm": 3.4427242279052734,
      "learning_rate": 0.00015291803278688526,
      "loss": 0.474,
      "step": 7280
    },
    {
      "epoch": 35.63148261134838,
      "grad_norm": 3.069155216217041,
      "learning_rate": 0.00015278688524590165,
      "loss": 0.4385,
      "step": 7300
    },
    {
      "epoch": 35.729103111653444,
      "grad_norm": 3.971026659011841,
      "learning_rate": 0.00015265573770491803,
      "loss": 0.5036,
      "step": 7320
    },
    {
      "epoch": 35.82672361195851,
      "grad_norm": 3.3750109672546387,
      "learning_rate": 0.00015252459016393442,
      "loss": 0.4957,
      "step": 7340
    },
    {
      "epoch": 35.924344112263576,
      "grad_norm": 3.382016181945801,
      "learning_rate": 0.00015239344262295084,
      "loss": 0.5176,
      "step": 7360
    },
    {
      "epoch": 36.02196461256864,
      "grad_norm": 2.5936732292175293,
      "learning_rate": 0.0001522622950819672,
      "loss": 0.4807,
      "step": 7380
    },
    {
      "epoch": 36.1195851128737,
      "grad_norm": 3.9948599338531494,
      "learning_rate": 0.00015213114754098362,
      "loss": 0.4285,
      "step": 7400
    },
    {
      "epoch": 36.21720561317877,
      "grad_norm": 3.3358399868011475,
      "learning_rate": 0.000152,
      "loss": 0.5013,
      "step": 7420
    },
    {
      "epoch": 36.31482611348383,
      "grad_norm": 3.7922635078430176,
      "learning_rate": 0.0001518688524590164,
      "loss": 0.4572,
      "step": 7440
    },
    {
      "epoch": 36.412446613788894,
      "grad_norm": 2.7253546714782715,
      "learning_rate": 0.0001517377049180328,
      "loss": 0.4151,
      "step": 7460
    },
    {
      "epoch": 36.51006711409396,
      "grad_norm": 5.344095706939697,
      "learning_rate": 0.00015160655737704918,
      "loss": 0.4718,
      "step": 7480
    },
    {
      "epoch": 36.607687614399026,
      "grad_norm": 3.768033981323242,
      "learning_rate": 0.0001514754098360656,
      "loss": 0.4406,
      "step": 7500
    },
    {
      "epoch": 36.70530811470409,
      "grad_norm": 3.884988784790039,
      "learning_rate": 0.00015134426229508198,
      "loss": 0.4605,
      "step": 7520
    },
    {
      "epoch": 36.80292861500915,
      "grad_norm": 4.368161678314209,
      "learning_rate": 0.00015121311475409837,
      "loss": 0.4609,
      "step": 7540
    },
    {
      "epoch": 36.90054911531421,
      "grad_norm": 5.337547779083252,
      "learning_rate": 0.00015108196721311476,
      "loss": 0.5078,
      "step": 7560
    },
    {
      "epoch": 36.99816961561928,
      "grad_norm": 3.7230191230773926,
      "learning_rate": 0.00015095081967213115,
      "loss": 0.5003,
      "step": 7580
    },
    {
      "epoch": 37.095790115924345,
      "grad_norm": 2.9239673614501953,
      "learning_rate": 0.00015081967213114754,
      "loss": 0.4312,
      "step": 7600
    },
    {
      "epoch": 37.19341061622941,
      "grad_norm": 2.9180755615234375,
      "learning_rate": 0.00015068852459016393,
      "loss": 0.4876,
      "step": 7620
    },
    {
      "epoch": 37.29103111653447,
      "grad_norm": 3.115180253982544,
      "learning_rate": 0.00015055737704918035,
      "loss": 0.4507,
      "step": 7640
    },
    {
      "epoch": 37.38865161683954,
      "grad_norm": 4.005005359649658,
      "learning_rate": 0.00015042622950819673,
      "loss": 0.458,
      "step": 7660
    },
    {
      "epoch": 37.4862721171446,
      "grad_norm": 3.3003063201904297,
      "learning_rate": 0.00015029508196721312,
      "loss": 0.4805,
      "step": 7680
    },
    {
      "epoch": 37.58389261744966,
      "grad_norm": 2.783266544342041,
      "learning_rate": 0.00015016393442622951,
      "loss": 0.4742,
      "step": 7700
    },
    {
      "epoch": 37.681513117754726,
      "grad_norm": 3.3052291870117188,
      "learning_rate": 0.0001500327868852459,
      "loss": 0.4415,
      "step": 7720
    },
    {
      "epoch": 37.779133618059795,
      "grad_norm": 4.352939605712891,
      "learning_rate": 0.00014990163934426232,
      "loss": 0.4489,
      "step": 7740
    },
    {
      "epoch": 37.87675411836486,
      "grad_norm": 2.9084181785583496,
      "learning_rate": 0.00014977049180327868,
      "loss": 0.439,
      "step": 7760
    },
    {
      "epoch": 37.97437461866992,
      "grad_norm": 3.3146231174468994,
      "learning_rate": 0.0001496393442622951,
      "loss": 0.4568,
      "step": 7780
    },
    {
      "epoch": 38.07199511897498,
      "grad_norm": 3.5614709854125977,
      "learning_rate": 0.0001495081967213115,
      "loss": 0.4648,
      "step": 7800
    },
    {
      "epoch": 38.16961561928005,
      "grad_norm": 3.782663345336914,
      "learning_rate": 0.00014937704918032788,
      "loss": 0.4636,
      "step": 7820
    },
    {
      "epoch": 38.267236119585114,
      "grad_norm": 3.2751007080078125,
      "learning_rate": 0.00014924590163934427,
      "loss": 0.4506,
      "step": 7840
    },
    {
      "epoch": 38.364856619890176,
      "grad_norm": 4.198697090148926,
      "learning_rate": 0.00014911475409836066,
      "loss": 0.4271,
      "step": 7860
    },
    {
      "epoch": 38.46247712019524,
      "grad_norm": 3.5415964126586914,
      "learning_rate": 0.00014898360655737707,
      "loss": 0.4112,
      "step": 7880
    },
    {
      "epoch": 38.56009762050031,
      "grad_norm": 3.3707010746002197,
      "learning_rate": 0.00014885245901639343,
      "loss": 0.4788,
      "step": 7900
    },
    {
      "epoch": 38.65771812080537,
      "grad_norm": 3.5206825733184814,
      "learning_rate": 0.00014872131147540985,
      "loss": 0.4456,
      "step": 7920
    },
    {
      "epoch": 38.75533862111043,
      "grad_norm": 3.870487689971924,
      "learning_rate": 0.00014859016393442624,
      "loss": 0.4346,
      "step": 7940
    },
    {
      "epoch": 38.852959121415495,
      "grad_norm": 4.031324863433838,
      "learning_rate": 0.00014845901639344263,
      "loss": 0.4272,
      "step": 7960
    },
    {
      "epoch": 38.950579621720564,
      "grad_norm": 3.552272081375122,
      "learning_rate": 0.00014832786885245902,
      "loss": 0.455,
      "step": 7980
    },
    {
      "epoch": 39.04820012202563,
      "grad_norm": 4.429355621337891,
      "learning_rate": 0.0001481967213114754,
      "loss": 0.474,
      "step": 8000
    },
    {
      "epoch": 39.14582062233069,
      "grad_norm": 2.7380106449127197,
      "learning_rate": 0.00014806557377049182,
      "loss": 0.3892,
      "step": 8020
    },
    {
      "epoch": 39.24344112263575,
      "grad_norm": 3.533512830734253,
      "learning_rate": 0.0001479344262295082,
      "loss": 0.4225,
      "step": 8040
    },
    {
      "epoch": 39.34106162294082,
      "grad_norm": 4.216946125030518,
      "learning_rate": 0.0001478032786885246,
      "loss": 0.4288,
      "step": 8060
    },
    {
      "epoch": 39.43868212324588,
      "grad_norm": 4.304781436920166,
      "learning_rate": 0.000147672131147541,
      "loss": 0.4678,
      "step": 8080
    },
    {
      "epoch": 39.536302623550945,
      "grad_norm": 3.1035361289978027,
      "learning_rate": 0.00014754098360655738,
      "loss": 0.436,
      "step": 8100
    },
    {
      "epoch": 39.63392312385601,
      "grad_norm": 4.312572956085205,
      "learning_rate": 0.00014740983606557377,
      "loss": 0.4418,
      "step": 8120
    },
    {
      "epoch": 39.73154362416108,
      "grad_norm": 4.513503074645996,
      "learning_rate": 0.00014727868852459016,
      "loss": 0.4613,
      "step": 8140
    },
    {
      "epoch": 39.82916412446614,
      "grad_norm": 3.7922120094299316,
      "learning_rate": 0.00014714754098360658,
      "loss": 0.4339,
      "step": 8160
    },
    {
      "epoch": 39.9267846247712,
      "grad_norm": 4.1964616775512695,
      "learning_rate": 0.00014701639344262297,
      "loss": 0.4799,
      "step": 8180
    },
    {
      "epoch": 40.024405125076264,
      "grad_norm": 3.0314066410064697,
      "learning_rate": 0.00014688524590163936,
      "loss": 0.4324,
      "step": 8200
    },
    {
      "epoch": 40.12202562538133,
      "grad_norm": 3.27500319480896,
      "learning_rate": 0.00014675409836065575,
      "loss": 0.4353,
      "step": 8220
    },
    {
      "epoch": 40.219646125686396,
      "grad_norm": 3.132371664047241,
      "learning_rate": 0.00014662295081967214,
      "loss": 0.4221,
      "step": 8240
    },
    {
      "epoch": 40.31726662599146,
      "grad_norm": 4.715281009674072,
      "learning_rate": 0.00014649180327868852,
      "loss": 0.4344,
      "step": 8260
    },
    {
      "epoch": 40.41488712629652,
      "grad_norm": 3.406825065612793,
      "learning_rate": 0.00014636065573770491,
      "loss": 0.4362,
      "step": 8280
    },
    {
      "epoch": 40.51250762660159,
      "grad_norm": 3.668811082839966,
      "learning_rate": 0.00014622950819672133,
      "loss": 0.4217,
      "step": 8300
    },
    {
      "epoch": 40.61012812690665,
      "grad_norm": 4.667581081390381,
      "learning_rate": 0.00014609836065573772,
      "loss": 0.4115,
      "step": 8320
    },
    {
      "epoch": 40.707748627211714,
      "grad_norm": 4.834279537200928,
      "learning_rate": 0.0001459672131147541,
      "loss": 0.4292,
      "step": 8340
    },
    {
      "epoch": 40.80536912751678,
      "grad_norm": 4.528176307678223,
      "learning_rate": 0.0001458360655737705,
      "loss": 0.4521,
      "step": 8360
    },
    {
      "epoch": 40.902989627821846,
      "grad_norm": 4.21350622177124,
      "learning_rate": 0.0001457049180327869,
      "loss": 0.4593,
      "step": 8380
    },
    {
      "epoch": 41.00061012812691,
      "grad_norm": 3.694138526916504,
      "learning_rate": 0.0001455737704918033,
      "loss": 0.434,
      "step": 8400
    },
    {
      "epoch": 41.09823062843197,
      "grad_norm": 3.2935163974761963,
      "learning_rate": 0.00014544262295081967,
      "loss": 0.3825,
      "step": 8420
    },
    {
      "epoch": 41.19585112873703,
      "grad_norm": 4.162698745727539,
      "learning_rate": 0.00014531147540983608,
      "loss": 0.3929,
      "step": 8440
    },
    {
      "epoch": 41.2934716290421,
      "grad_norm": 3.650602340698242,
      "learning_rate": 0.00014518032786885247,
      "loss": 0.4134,
      "step": 8460
    },
    {
      "epoch": 41.391092129347165,
      "grad_norm": 4.435211181640625,
      "learning_rate": 0.00014504918032786886,
      "loss": 0.4438,
      "step": 8480
    },
    {
      "epoch": 41.48871262965223,
      "grad_norm": 3.3488500118255615,
      "learning_rate": 0.00014491803278688525,
      "loss": 0.3997,
      "step": 8500
    },
    {
      "epoch": 41.58633312995729,
      "grad_norm": 3.115065574645996,
      "learning_rate": 0.00014478688524590164,
      "loss": 0.4443,
      "step": 8520
    },
    {
      "epoch": 41.68395363026236,
      "grad_norm": 3.3622965812683105,
      "learning_rate": 0.00014465573770491806,
      "loss": 0.4335,
      "step": 8540
    },
    {
      "epoch": 41.78157413056742,
      "grad_norm": 3.420353889465332,
      "learning_rate": 0.00014452459016393442,
      "loss": 0.4868,
      "step": 8560
    },
    {
      "epoch": 41.87919463087248,
      "grad_norm": 4.102779388427734,
      "learning_rate": 0.00014439344262295084,
      "loss": 0.4201,
      "step": 8580
    },
    {
      "epoch": 41.976815131177545,
      "grad_norm": 3.436736583709717,
      "learning_rate": 0.00014426229508196722,
      "loss": 0.4527,
      "step": 8600
    },
    {
      "epoch": 42.074435631482615,
      "grad_norm": 3.5886449813842773,
      "learning_rate": 0.00014413114754098361,
      "loss": 0.4072,
      "step": 8620
    },
    {
      "epoch": 42.17205613178768,
      "grad_norm": 3.397456645965576,
      "learning_rate": 0.000144,
      "loss": 0.3978,
      "step": 8640
    },
    {
      "epoch": 42.26967663209274,
      "grad_norm": 3.1985559463500977,
      "learning_rate": 0.0001438688524590164,
      "loss": 0.4335,
      "step": 8660
    },
    {
      "epoch": 42.3672971323978,
      "grad_norm": 3.780562400817871,
      "learning_rate": 0.0001437377049180328,
      "loss": 0.3899,
      "step": 8680
    },
    {
      "epoch": 42.464917632702864,
      "grad_norm": 3.2960257530212402,
      "learning_rate": 0.00014360655737704917,
      "loss": 0.4167,
      "step": 8700
    },
    {
      "epoch": 42.56253813300793,
      "grad_norm": 3.9879534244537354,
      "learning_rate": 0.0001434754098360656,
      "loss": 0.3799,
      "step": 8720
    },
    {
      "epoch": 42.660158633312996,
      "grad_norm": 4.146468162536621,
      "learning_rate": 0.00014334426229508198,
      "loss": 0.4341,
      "step": 8740
    },
    {
      "epoch": 42.75777913361806,
      "grad_norm": 3.994354724884033,
      "learning_rate": 0.00014321311475409837,
      "loss": 0.458,
      "step": 8760
    },
    {
      "epoch": 42.85539963392312,
      "grad_norm": 3.043018341064453,
      "learning_rate": 0.00014308196721311476,
      "loss": 0.4277,
      "step": 8780
    },
    {
      "epoch": 42.95302013422819,
      "grad_norm": 3.4882431030273438,
      "learning_rate": 0.00014295081967213115,
      "loss": 0.4491,
      "step": 8800
    },
    {
      "epoch": 43.05064063453325,
      "grad_norm": 3.6433932781219482,
      "learning_rate": 0.00014281967213114756,
      "loss": 0.4009,
      "step": 8820
    },
    {
      "epoch": 43.148261134838314,
      "grad_norm": 3.4401466846466064,
      "learning_rate": 0.00014268852459016395,
      "loss": 0.4191,
      "step": 8840
    },
    {
      "epoch": 43.24588163514338,
      "grad_norm": 3.963707685470581,
      "learning_rate": 0.00014255737704918034,
      "loss": 0.3961,
      "step": 8860
    },
    {
      "epoch": 43.343502135448446,
      "grad_norm": 3.0136444568634033,
      "learning_rate": 0.00014242622950819673,
      "loss": 0.4052,
      "step": 8880
    },
    {
      "epoch": 43.44112263575351,
      "grad_norm": 3.807661533355713,
      "learning_rate": 0.00014229508196721312,
      "loss": 0.3929,
      "step": 8900
    },
    {
      "epoch": 43.53874313605857,
      "grad_norm": 3.722825765609741,
      "learning_rate": 0.0001421639344262295,
      "loss": 0.4183,
      "step": 8920
    },
    {
      "epoch": 43.63636363636363,
      "grad_norm": 3.470569610595703,
      "learning_rate": 0.0001420327868852459,
      "loss": 0.4169,
      "step": 8940
    },
    {
      "epoch": 43.7339841366687,
      "grad_norm": 3.4966650009155273,
      "learning_rate": 0.00014190163934426231,
      "loss": 0.4121,
      "step": 8960
    },
    {
      "epoch": 43.831604636973765,
      "grad_norm": 4.7074174880981445,
      "learning_rate": 0.0001417704918032787,
      "loss": 0.4634,
      "step": 8980
    },
    {
      "epoch": 43.92922513727883,
      "grad_norm": 3.609379291534424,
      "learning_rate": 0.0001416393442622951,
      "loss": 0.4405,
      "step": 9000
    },
    {
      "epoch": 44.02684563758389,
      "grad_norm": 3.337123394012451,
      "learning_rate": 0.00014150819672131148,
      "loss": 0.3883,
      "step": 9020
    },
    {
      "epoch": 44.12446613788896,
      "grad_norm": 3.321878433227539,
      "learning_rate": 0.00014137704918032787,
      "loss": 0.4024,
      "step": 9040
    },
    {
      "epoch": 44.22208663819402,
      "grad_norm": 3.427896738052368,
      "learning_rate": 0.00014124590163934426,
      "loss": 0.4032,
      "step": 9060
    },
    {
      "epoch": 44.31970713849908,
      "grad_norm": 4.160934925079346,
      "learning_rate": 0.00014111475409836065,
      "loss": 0.3718,
      "step": 9080
    },
    {
      "epoch": 44.417327638804146,
      "grad_norm": 3.591139316558838,
      "learning_rate": 0.00014098360655737707,
      "loss": 0.3829,
      "step": 9100
    },
    {
      "epoch": 44.514948139109215,
      "grad_norm": 2.9896583557128906,
      "learning_rate": 0.00014085245901639346,
      "loss": 0.4231,
      "step": 9120
    },
    {
      "epoch": 44.61256863941428,
      "grad_norm": 4.670585632324219,
      "learning_rate": 0.00014072131147540985,
      "loss": 0.4285,
      "step": 9140
    },
    {
      "epoch": 44.71018913971934,
      "grad_norm": 4.087858200073242,
      "learning_rate": 0.00014059016393442624,
      "loss": 0.4061,
      "step": 9160
    },
    {
      "epoch": 44.8078096400244,
      "grad_norm": 4.61581563949585,
      "learning_rate": 0.00014045901639344262,
      "loss": 0.4245,
      "step": 9180
    },
    {
      "epoch": 44.90543014032947,
      "grad_norm": 4.715487957000732,
      "learning_rate": 0.00014032786885245904,
      "loss": 0.4223,
      "step": 9200
    },
    {
      "epoch": 45.003050640634534,
      "grad_norm": 3.8034169673919678,
      "learning_rate": 0.0001401967213114754,
      "loss": 0.407,
      "step": 9220
    },
    {
      "epoch": 45.100671140939596,
      "grad_norm": 2.813342332839966,
      "learning_rate": 0.00014006557377049182,
      "loss": 0.3936,
      "step": 9240
    },
    {
      "epoch": 45.19829164124466,
      "grad_norm": 3.130986452102661,
      "learning_rate": 0.0001399344262295082,
      "loss": 0.3897,
      "step": 9260
    },
    {
      "epoch": 45.29591214154973,
      "grad_norm": 3.6343700885772705,
      "learning_rate": 0.0001398032786885246,
      "loss": 0.3611,
      "step": 9280
    },
    {
      "epoch": 45.39353264185479,
      "grad_norm": 3.8157641887664795,
      "learning_rate": 0.000139672131147541,
      "loss": 0.419,
      "step": 9300
    },
    {
      "epoch": 45.49115314215985,
      "grad_norm": 3.9056179523468018,
      "learning_rate": 0.00013954098360655738,
      "loss": 0.3746,
      "step": 9320
    },
    {
      "epoch": 45.588773642464915,
      "grad_norm": 3.952216625213623,
      "learning_rate": 0.0001394098360655738,
      "loss": 0.3876,
      "step": 9340
    },
    {
      "epoch": 45.686394142769984,
      "grad_norm": 3.924835443496704,
      "learning_rate": 0.00013927868852459016,
      "loss": 0.4199,
      "step": 9360
    },
    {
      "epoch": 45.78401464307505,
      "grad_norm": 4.389228343963623,
      "learning_rate": 0.00013914754098360657,
      "loss": 0.3949,
      "step": 9380
    },
    {
      "epoch": 45.88163514338011,
      "grad_norm": 3.311279296875,
      "learning_rate": 0.00013901639344262296,
      "loss": 0.408,
      "step": 9400
    },
    {
      "epoch": 45.97925564368517,
      "grad_norm": 3.3349342346191406,
      "learning_rate": 0.00013888524590163935,
      "loss": 0.4578,
      "step": 9420
    },
    {
      "epoch": 46.07687614399024,
      "grad_norm": 2.8730790615081787,
      "learning_rate": 0.00013875409836065574,
      "loss": 0.3864,
      "step": 9440
    },
    {
      "epoch": 46.1744966442953,
      "grad_norm": 5.034847259521484,
      "learning_rate": 0.00013862295081967213,
      "loss": 0.3707,
      "step": 9460
    },
    {
      "epoch": 46.272117144600365,
      "grad_norm": 4.460078716278076,
      "learning_rate": 0.00013849180327868855,
      "loss": 0.3823,
      "step": 9480
    },
    {
      "epoch": 46.36973764490543,
      "grad_norm": 3.5436556339263916,
      "learning_rate": 0.0001383606557377049,
      "loss": 0.4063,
      "step": 9500
    },
    {
      "epoch": 46.4673581452105,
      "grad_norm": 3.1509130001068115,
      "learning_rate": 0.00013822950819672133,
      "loss": 0.3805,
      "step": 9520
    },
    {
      "epoch": 46.56497864551556,
      "grad_norm": 3.3589935302734375,
      "learning_rate": 0.00013809836065573771,
      "loss": 0.4332,
      "step": 9540
    },
    {
      "epoch": 46.66259914582062,
      "grad_norm": 3.7478573322296143,
      "learning_rate": 0.0001379672131147541,
      "loss": 0.3721,
      "step": 9560
    },
    {
      "epoch": 46.760219646125684,
      "grad_norm": 3.930676221847534,
      "learning_rate": 0.0001378360655737705,
      "loss": 0.3847,
      "step": 9580
    },
    {
      "epoch": 46.85784014643075,
      "grad_norm": 4.218331336975098,
      "learning_rate": 0.00013770491803278688,
      "loss": 0.4248,
      "step": 9600
    },
    {
      "epoch": 46.955460646735816,
      "grad_norm": 3.993345022201538,
      "learning_rate": 0.0001375737704918033,
      "loss": 0.3976,
      "step": 9620
    },
    {
      "epoch": 47.05308114704088,
      "grad_norm": 3.4169094562530518,
      "learning_rate": 0.0001374426229508197,
      "loss": 0.4126,
      "step": 9640
    },
    {
      "epoch": 47.15070164734594,
      "grad_norm": 3.3602797985076904,
      "learning_rate": 0.00013731147540983608,
      "loss": 0.3675,
      "step": 9660
    },
    {
      "epoch": 47.24832214765101,
      "grad_norm": 3.6925911903381348,
      "learning_rate": 0.00013718032786885247,
      "loss": 0.3732,
      "step": 9680
    },
    {
      "epoch": 47.34594264795607,
      "grad_norm": 3.3294148445129395,
      "learning_rate": 0.00013704918032786886,
      "loss": 0.3576,
      "step": 9700
    },
    {
      "epoch": 47.443563148261134,
      "grad_norm": 4.107402801513672,
      "learning_rate": 0.00013691803278688525,
      "loss": 0.376,
      "step": 9720
    },
    {
      "epoch": 47.5411836485662,
      "grad_norm": 3.5636045932769775,
      "learning_rate": 0.00013678688524590164,
      "loss": 0.3918,
      "step": 9740
    },
    {
      "epoch": 47.638804148871266,
      "grad_norm": 4.105456829071045,
      "learning_rate": 0.00013665573770491805,
      "loss": 0.3864,
      "step": 9760
    },
    {
      "epoch": 47.73642464917633,
      "grad_norm": 3.6053054332733154,
      "learning_rate": 0.00013652459016393444,
      "loss": 0.4009,
      "step": 9780
    },
    {
      "epoch": 47.83404514948139,
      "grad_norm": 4.397184371948242,
      "learning_rate": 0.00013639344262295083,
      "loss": 0.4094,
      "step": 9800
    },
    {
      "epoch": 47.93166564978645,
      "grad_norm": 3.121466875076294,
      "learning_rate": 0.00013626229508196722,
      "loss": 0.4276,
      "step": 9820
    },
    {
      "epoch": 48.02928615009152,
      "grad_norm": 3.9471912384033203,
      "learning_rate": 0.0001361311475409836,
      "loss": 0.3981,
      "step": 9840
    },
    {
      "epoch": 48.126906650396585,
      "grad_norm": 2.690577983856201,
      "learning_rate": 0.00013600000000000003,
      "loss": 0.3281,
      "step": 9860
    },
    {
      "epoch": 48.22452715070165,
      "grad_norm": 2.87288498878479,
      "learning_rate": 0.0001358688524590164,
      "loss": 0.4053,
      "step": 9880
    },
    {
      "epoch": 48.32214765100671,
      "grad_norm": 3.6237525939941406,
      "learning_rate": 0.0001357377049180328,
      "loss": 0.3436,
      "step": 9900
    },
    {
      "epoch": 48.41976815131178,
      "grad_norm": 3.1614110469818115,
      "learning_rate": 0.0001356065573770492,
      "loss": 0.3732,
      "step": 9920
    },
    {
      "epoch": 48.51738865161684,
      "grad_norm": 4.337550640106201,
      "learning_rate": 0.00013547540983606556,
      "loss": 0.3558,
      "step": 9940
    },
    {
      "epoch": 48.6150091519219,
      "grad_norm": 2.904919385910034,
      "learning_rate": 0.00013534426229508197,
      "loss": 0.4165,
      "step": 9960
    },
    {
      "epoch": 48.712629652226966,
      "grad_norm": 3.3883869647979736,
      "learning_rate": 0.00013521311475409836,
      "loss": 0.37,
      "step": 9980
    },
    {
      "epoch": 48.810250152532035,
      "grad_norm": 3.388711452484131,
      "learning_rate": 0.00013508196721311478,
      "loss": 0.4075,
      "step": 10000
    },
    {
      "epoch": 48.9078706528371,
      "grad_norm": 3.972365617752075,
      "learning_rate": 0.00013495081967213114,
      "loss": 0.3944,
      "step": 10020
    },
    {
      "epoch": 49.00549115314216,
      "grad_norm": 2.836026668548584,
      "learning_rate": 0.00013481967213114756,
      "loss": 0.4315,
      "step": 10040
    },
    {
      "epoch": 49.10311165344722,
      "grad_norm": 3.727874755859375,
      "learning_rate": 0.00013468852459016395,
      "loss": 0.2791,
      "step": 10060
    },
    {
      "epoch": 49.20073215375229,
      "grad_norm": 3.036069869995117,
      "learning_rate": 0.00013455737704918034,
      "loss": 0.358,
      "step": 10080
    },
    {
      "epoch": 49.298352654057354,
      "grad_norm": 3.044581651687622,
      "learning_rate": 0.00013442622950819673,
      "loss": 0.423,
      "step": 10100
    },
    {
      "epoch": 49.395973154362416,
      "grad_norm": 3.956517457962036,
      "learning_rate": 0.00013429508196721311,
      "loss": 0.3988,
      "step": 10120
    },
    {
      "epoch": 49.49359365466748,
      "grad_norm": 4.359031677246094,
      "learning_rate": 0.00013416393442622953,
      "loss": 0.352,
      "step": 10140
    },
    {
      "epoch": 49.59121415497255,
      "grad_norm": 4.619460582733154,
      "learning_rate": 0.0001340327868852459,
      "loss": 0.357,
      "step": 10160
    },
    {
      "epoch": 49.68883465527761,
      "grad_norm": 4.051796913146973,
      "learning_rate": 0.0001339016393442623,
      "loss": 0.4048,
      "step": 10180
    },
    {
      "epoch": 49.78645515558267,
      "grad_norm": 3.5017082691192627,
      "learning_rate": 0.0001337704918032787,
      "loss": 0.3937,
      "step": 10200
    },
    {
      "epoch": 49.884075655887735,
      "grad_norm": 4.194153308868408,
      "learning_rate": 0.0001336393442622951,
      "loss": 0.3989,
      "step": 10220
    },
    {
      "epoch": 49.981696156192804,
      "grad_norm": 3.5315494537353516,
      "learning_rate": 0.00013350819672131148,
      "loss": 0.4032,
      "step": 10240
    },
    {
      "epoch": 50.079316656497866,
      "grad_norm": 3.8597731590270996,
      "learning_rate": 0.00013337704918032787,
      "loss": 0.3786,
      "step": 10260
    },
    {
      "epoch": 50.17693715680293,
      "grad_norm": 3.1231532096862793,
      "learning_rate": 0.00013324590163934428,
      "loss": 0.3764,
      "step": 10280
    },
    {
      "epoch": 50.27455765710799,
      "grad_norm": 3.272097110748291,
      "learning_rate": 0.00013311475409836067,
      "loss": 0.3943,
      "step": 10300
    },
    {
      "epoch": 50.37217815741306,
      "grad_norm": 5.884640216827393,
      "learning_rate": 0.00013298360655737706,
      "loss": 0.312,
      "step": 10320
    },
    {
      "epoch": 50.46979865771812,
      "grad_norm": 3.424651622772217,
      "learning_rate": 0.00013285245901639345,
      "loss": 0.3602,
      "step": 10340
    },
    {
      "epoch": 50.567419158023185,
      "grad_norm": 3.703669786453247,
      "learning_rate": 0.00013272131147540984,
      "loss": 0.3992,
      "step": 10360
    },
    {
      "epoch": 50.66503965832825,
      "grad_norm": 3.4607040882110596,
      "learning_rate": 0.00013259016393442623,
      "loss": 0.3773,
      "step": 10380
    },
    {
      "epoch": 50.76266015863331,
      "grad_norm": 4.518139362335205,
      "learning_rate": 0.00013245901639344262,
      "loss": 0.3649,
      "step": 10400
    },
    {
      "epoch": 50.86028065893838,
      "grad_norm": 3.7281651496887207,
      "learning_rate": 0.00013232786885245904,
      "loss": 0.3797,
      "step": 10420
    },
    {
      "epoch": 50.95790115924344,
      "grad_norm": 3.4530558586120605,
      "learning_rate": 0.00013219672131147543,
      "loss": 0.3854,
      "step": 10440
    },
    {
      "epoch": 51.0555216595485,
      "grad_norm": 3.220987558364868,
      "learning_rate": 0.0001320655737704918,
      "loss": 0.3606,
      "step": 10460
    },
    {
      "epoch": 51.153142159853566,
      "grad_norm": 4.0888352394104,
      "learning_rate": 0.0001319344262295082,
      "loss": 0.3342,
      "step": 10480
    },
    {
      "epoch": 51.250762660158635,
      "grad_norm": 3.6605489253997803,
      "learning_rate": 0.0001318032786885246,
      "loss": 0.3785,
      "step": 10500
    },
    {
      "epoch": 51.3483831604637,
      "grad_norm": 3.1391944885253906,
      "learning_rate": 0.000131672131147541,
      "loss": 0.3741,
      "step": 10520
    },
    {
      "epoch": 51.44600366076876,
      "grad_norm": 4.535563945770264,
      "learning_rate": 0.00013154098360655737,
      "loss": 0.3897,
      "step": 10540
    },
    {
      "epoch": 51.54362416107382,
      "grad_norm": 2.85030198097229,
      "learning_rate": 0.0001314098360655738,
      "loss": 0.3335,
      "step": 10560
    },
    {
      "epoch": 51.64124466137889,
      "grad_norm": 3.172010660171509,
      "learning_rate": 0.00013127868852459018,
      "loss": 0.3223,
      "step": 10580
    },
    {
      "epoch": 51.738865161683954,
      "grad_norm": 3.834644079208374,
      "learning_rate": 0.00013114754098360654,
      "loss": 0.3897,
      "step": 10600
    },
    {
      "epoch": 51.836485661989016,
      "grad_norm": 3.7358360290527344,
      "learning_rate": 0.00013101639344262296,
      "loss": 0.3853,
      "step": 10620
    },
    {
      "epoch": 51.93410616229408,
      "grad_norm": 4.027935981750488,
      "learning_rate": 0.00013088524590163935,
      "loss": 0.3775,
      "step": 10640
    },
    {
      "epoch": 52.03172666259915,
      "grad_norm": 2.9343104362487793,
      "learning_rate": 0.00013075409836065576,
      "loss": 0.3878,
      "step": 10660
    },
    {
      "epoch": 52.12934716290421,
      "grad_norm": 3.9840517044067383,
      "learning_rate": 0.00013062295081967213,
      "loss": 0.3334,
      "step": 10680
    },
    {
      "epoch": 52.22696766320927,
      "grad_norm": 2.7341692447662354,
      "learning_rate": 0.00013049180327868854,
      "loss": 0.3643,
      "step": 10700
    },
    {
      "epoch": 52.324588163514335,
      "grad_norm": 4.158634662628174,
      "learning_rate": 0.00013036065573770493,
      "loss": 0.3497,
      "step": 10720
    },
    {
      "epoch": 52.422208663819404,
      "grad_norm": 3.440031051635742,
      "learning_rate": 0.00013022950819672132,
      "loss": 0.3917,
      "step": 10740
    },
    {
      "epoch": 52.51982916412447,
      "grad_norm": 3.250298500061035,
      "learning_rate": 0.0001300983606557377,
      "loss": 0.3472,
      "step": 10760
    },
    {
      "epoch": 52.61744966442953,
      "grad_norm": 3.672611951828003,
      "learning_rate": 0.0001299672131147541,
      "loss": 0.3315,
      "step": 10780
    },
    {
      "epoch": 52.71507016473459,
      "grad_norm": 2.9064407348632812,
      "learning_rate": 0.00012983606557377052,
      "loss": 0.4121,
      "step": 10800
    },
    {
      "epoch": 52.81269066503966,
      "grad_norm": 3.6620004177093506,
      "learning_rate": 0.00012970491803278688,
      "loss": 0.3402,
      "step": 10820
    },
    {
      "epoch": 52.91031116534472,
      "grad_norm": 4.237580299377441,
      "learning_rate": 0.0001295737704918033,
      "loss": 0.4054,
      "step": 10840
    },
    {
      "epoch": 53.007931665649785,
      "grad_norm": 5.529778957366943,
      "learning_rate": 0.00012944262295081968,
      "loss": 0.359,
      "step": 10860
    },
    {
      "epoch": 53.10555216595485,
      "grad_norm": 3.2727859020233154,
      "learning_rate": 0.00012931147540983607,
      "loss": 0.3431,
      "step": 10880
    },
    {
      "epoch": 53.20317266625992,
      "grad_norm": 3.652461290359497,
      "learning_rate": 0.00012918032786885246,
      "loss": 0.3441,
      "step": 10900
    },
    {
      "epoch": 53.30079316656498,
      "grad_norm": 3.462115526199341,
      "learning_rate": 0.00012904918032786885,
      "loss": 0.3754,
      "step": 10920
    },
    {
      "epoch": 53.39841366687004,
      "grad_norm": 2.771514654159546,
      "learning_rate": 0.00012891803278688527,
      "loss": 0.3568,
      "step": 10940
    },
    {
      "epoch": 53.496034167175104,
      "grad_norm": 3.806295394897461,
      "learning_rate": 0.00012878688524590166,
      "loss": 0.3357,
      "step": 10960
    },
    {
      "epoch": 53.59365466748017,
      "grad_norm": 4.707630634307861,
      "learning_rate": 0.00012865573770491802,
      "loss": 0.3461,
      "step": 10980
    },
    {
      "epoch": 53.691275167785236,
      "grad_norm": 4.196589946746826,
      "learning_rate": 0.00012852459016393444,
      "loss": 0.3578,
      "step": 11000
    },
    {
      "epoch": 53.7888956680903,
      "grad_norm": 3.576612949371338,
      "learning_rate": 0.00012839344262295083,
      "loss": 0.3946,
      "step": 11020
    },
    {
      "epoch": 53.88651616839536,
      "grad_norm": 3.4126014709472656,
      "learning_rate": 0.00012826229508196722,
      "loss": 0.3441,
      "step": 11040
    },
    {
      "epoch": 53.98413666870043,
      "grad_norm": 3.8211019039154053,
      "learning_rate": 0.0001281311475409836,
      "loss": 0.3652,
      "step": 11060
    },
    {
      "epoch": 54.08175716900549,
      "grad_norm": 3.447636127471924,
      "learning_rate": 0.00012800000000000002,
      "loss": 0.3315,
      "step": 11080
    },
    {
      "epoch": 54.179377669310554,
      "grad_norm": 4.556658744812012,
      "learning_rate": 0.0001278688524590164,
      "loss": 0.37,
      "step": 11100
    },
    {
      "epoch": 54.27699816961562,
      "grad_norm": 3.292505979537964,
      "learning_rate": 0.00012773770491803277,
      "loss": 0.3386,
      "step": 11120
    },
    {
      "epoch": 54.374618669920686,
      "grad_norm": 2.7513251304626465,
      "learning_rate": 0.0001276065573770492,
      "loss": 0.3497,
      "step": 11140
    },
    {
      "epoch": 54.47223917022575,
      "grad_norm": 3.0105299949645996,
      "learning_rate": 0.00012747540983606558,
      "loss": 0.3235,
      "step": 11160
    },
    {
      "epoch": 54.56985967053081,
      "grad_norm": 3.2643628120422363,
      "learning_rate": 0.000127344262295082,
      "loss": 0.3673,
      "step": 11180
    },
    {
      "epoch": 54.66748017083587,
      "grad_norm": 3.5350589752197266,
      "learning_rate": 0.00012721311475409836,
      "loss": 0.3733,
      "step": 11200
    },
    {
      "epoch": 54.76510067114094,
      "grad_norm": 5.337443828582764,
      "learning_rate": 0.00012708196721311477,
      "loss": 0.3376,
      "step": 11220
    },
    {
      "epoch": 54.862721171446005,
      "grad_norm": 3.618621826171875,
      "learning_rate": 0.00012695081967213116,
      "loss": 0.3372,
      "step": 11240
    },
    {
      "epoch": 54.96034167175107,
      "grad_norm": 3.5537171363830566,
      "learning_rate": 0.00012681967213114753,
      "loss": 0.3639,
      "step": 11260
    },
    {
      "epoch": 55.05796217205613,
      "grad_norm": 3.529486656188965,
      "learning_rate": 0.00012668852459016394,
      "loss": 0.3512,
      "step": 11280
    },
    {
      "epoch": 55.1555826723612,
      "grad_norm": 4.277002811431885,
      "learning_rate": 0.00012655737704918033,
      "loss": 0.3841,
      "step": 11300
    },
    {
      "epoch": 55.25320317266626,
      "grad_norm": 3.5343832969665527,
      "learning_rate": 0.00012642622950819675,
      "loss": 0.3757,
      "step": 11320
    },
    {
      "epoch": 55.35082367297132,
      "grad_norm": 3.350728988647461,
      "learning_rate": 0.0001262950819672131,
      "loss": 0.3179,
      "step": 11340
    },
    {
      "epoch": 55.448444173276386,
      "grad_norm": 4.037693500518799,
      "learning_rate": 0.0001261639344262295,
      "loss": 0.3685,
      "step": 11360
    },
    {
      "epoch": 55.546064673581455,
      "grad_norm": 4.390751838684082,
      "learning_rate": 0.00012603278688524592,
      "loss": 0.3161,
      "step": 11380
    },
    {
      "epoch": 55.64368517388652,
      "grad_norm": 4.671621322631836,
      "learning_rate": 0.0001259016393442623,
      "loss": 0.3268,
      "step": 11400
    },
    {
      "epoch": 55.74130567419158,
      "grad_norm": 4.637950897216797,
      "learning_rate": 0.0001257704918032787,
      "loss": 0.357,
      "step": 11420
    },
    {
      "epoch": 55.83892617449664,
      "grad_norm": 5.1185455322265625,
      "learning_rate": 0.00012563934426229508,
      "loss": 0.3628,
      "step": 11440
    },
    {
      "epoch": 55.93654667480171,
      "grad_norm": 5.654429912567139,
      "learning_rate": 0.0001255081967213115,
      "loss": 0.3519,
      "step": 11460
    },
    {
      "epoch": 56.034167175106774,
      "grad_norm": 3.041135549545288,
      "learning_rate": 0.00012537704918032786,
      "loss": 0.3529,
      "step": 11480
    },
    {
      "epoch": 56.131787675411836,
      "grad_norm": 4.020240783691406,
      "learning_rate": 0.00012524590163934425,
      "loss": 0.3597,
      "step": 11500
    },
    {
      "epoch": 56.2294081757169,
      "grad_norm": 3.730290412902832,
      "learning_rate": 0.00012511475409836067,
      "loss": 0.3137,
      "step": 11520
    },
    {
      "epoch": 56.32702867602197,
      "grad_norm": 3.314856767654419,
      "learning_rate": 0.00012498360655737706,
      "loss": 0.3122,
      "step": 11540
    },
    {
      "epoch": 56.42464917632703,
      "grad_norm": 4.214298248291016,
      "learning_rate": 0.00012485245901639345,
      "loss": 0.343,
      "step": 11560
    },
    {
      "epoch": 56.52226967663209,
      "grad_norm": 4.5516791343688965,
      "learning_rate": 0.00012472131147540984,
      "loss": 0.3714,
      "step": 11580
    },
    {
      "epoch": 56.619890176937155,
      "grad_norm": 5.098419666290283,
      "learning_rate": 0.00012459016393442625,
      "loss": 0.3489,
      "step": 11600
    },
    {
      "epoch": 56.717510677242224,
      "grad_norm": 2.3853330612182617,
      "learning_rate": 0.00012445901639344262,
      "loss": 0.3389,
      "step": 11620
    },
    {
      "epoch": 56.815131177547286,
      "grad_norm": 4.595917224884033,
      "learning_rate": 0.000124327868852459,
      "loss": 0.3392,
      "step": 11640
    },
    {
      "epoch": 56.91275167785235,
      "grad_norm": 4.49993896484375,
      "learning_rate": 0.00012419672131147542,
      "loss": 0.3419,
      "step": 11660
    },
    {
      "epoch": 57.01037217815741,
      "grad_norm": 3.2568199634552,
      "learning_rate": 0.0001240655737704918,
      "loss": 0.3614,
      "step": 11680
    },
    {
      "epoch": 57.10799267846248,
      "grad_norm": 4.653031826019287,
      "learning_rate": 0.0001239344262295082,
      "loss": 0.3096,
      "step": 11700
    },
    {
      "epoch": 57.20561317876754,
      "grad_norm": 2.5655269622802734,
      "learning_rate": 0.0001238032786885246,
      "loss": 0.3306,
      "step": 11720
    },
    {
      "epoch": 57.303233679072605,
      "grad_norm": 3.4758191108703613,
      "learning_rate": 0.000123672131147541,
      "loss": 0.3513,
      "step": 11740
    },
    {
      "epoch": 57.40085417937767,
      "grad_norm": 3.107544183731079,
      "learning_rate": 0.0001235409836065574,
      "loss": 0.3231,
      "step": 11760
    },
    {
      "epoch": 57.49847467968274,
      "grad_norm": 3.983060121536255,
      "learning_rate": 0.00012340983606557376,
      "loss": 0.3366,
      "step": 11780
    },
    {
      "epoch": 57.5960951799878,
      "grad_norm": 3.987680673599243,
      "learning_rate": 0.00012327868852459017,
      "loss": 0.3365,
      "step": 11800
    },
    {
      "epoch": 57.69371568029286,
      "grad_norm": 3.7637107372283936,
      "learning_rate": 0.00012314754098360656,
      "loss": 0.3432,
      "step": 11820
    },
    {
      "epoch": 57.79133618059792,
      "grad_norm": 3.193894386291504,
      "learning_rate": 0.00012301639344262295,
      "loss": 0.3668,
      "step": 11840
    },
    {
      "epoch": 57.88895668090299,
      "grad_norm": 3.078507423400879,
      "learning_rate": 0.00012288524590163934,
      "loss": 0.3922,
      "step": 11860
    },
    {
      "epoch": 57.986577181208055,
      "grad_norm": 3.561068058013916,
      "learning_rate": 0.00012275409836065573,
      "loss": 0.3285,
      "step": 11880
    },
    {
      "epoch": 58.08419768151312,
      "grad_norm": 2.9965078830718994,
      "learning_rate": 0.00012262295081967215,
      "loss": 0.2991,
      "step": 11900
    },
    {
      "epoch": 58.18181818181818,
      "grad_norm": 3.8305752277374268,
      "learning_rate": 0.0001224918032786885,
      "loss": 0.3728,
      "step": 11920
    },
    {
      "epoch": 58.27943868212325,
      "grad_norm": 3.708678960800171,
      "learning_rate": 0.00012236065573770493,
      "loss": 0.3078,
      "step": 11940
    },
    {
      "epoch": 58.37705918242831,
      "grad_norm": 3.8988072872161865,
      "learning_rate": 0.00012222950819672132,
      "loss": 0.3389,
      "step": 11960
    },
    {
      "epoch": 58.474679682733374,
      "grad_norm": 3.550420045852661,
      "learning_rate": 0.00012209836065573773,
      "loss": 0.4069,
      "step": 11980
    },
    {
      "epoch": 58.572300183038436,
      "grad_norm": 3.799747943878174,
      "learning_rate": 0.0001219672131147541,
      "loss": 0.3058,
      "step": 12000
    },
    {
      "epoch": 58.669920683343506,
      "grad_norm": 4.734778881072998,
      "learning_rate": 0.0001218360655737705,
      "loss": 0.3199,
      "step": 12020
    },
    {
      "epoch": 58.76754118364857,
      "grad_norm": 3.449169158935547,
      "learning_rate": 0.0001217049180327869,
      "loss": 0.3309,
      "step": 12040
    },
    {
      "epoch": 58.86516168395363,
      "grad_norm": 3.164651393890381,
      "learning_rate": 0.00012157377049180328,
      "loss": 0.3338,
      "step": 12060
    },
    {
      "epoch": 58.96278218425869,
      "grad_norm": 3.976644515991211,
      "learning_rate": 0.00012144262295081968,
      "loss": 0.3207,
      "step": 12080
    },
    {
      "epoch": 59.060402684563755,
      "grad_norm": 4.6017632484436035,
      "learning_rate": 0.00012131147540983607,
      "loss": 0.317,
      "step": 12100
    },
    {
      "epoch": 59.158023184868824,
      "grad_norm": 6.806497573852539,
      "learning_rate": 0.00012118032786885247,
      "loss": 0.2943,
      "step": 12120
    },
    {
      "epoch": 59.25564368517389,
      "grad_norm": 3.545241355895996,
      "learning_rate": 0.00012104918032786885,
      "loss": 0.3093,
      "step": 12140
    },
    {
      "epoch": 59.35326418547895,
      "grad_norm": 2.435683488845825,
      "learning_rate": 0.00012091803278688525,
      "loss": 0.3235,
      "step": 12160
    },
    {
      "epoch": 59.45088468578401,
      "grad_norm": 3.507638931274414,
      "learning_rate": 0.00012078688524590165,
      "loss": 0.3405,
      "step": 12180
    },
    {
      "epoch": 59.54850518608908,
      "grad_norm": 3.9438552856445312,
      "learning_rate": 0.00012065573770491804,
      "loss": 0.3311,
      "step": 12200
    },
    {
      "epoch": 59.64612568639414,
      "grad_norm": 5.4473652839660645,
      "learning_rate": 0.00012052459016393443,
      "loss": 0.3424,
      "step": 12220
    },
    {
      "epoch": 59.743746186699205,
      "grad_norm": 3.204779624938965,
      "learning_rate": 0.00012039344262295082,
      "loss": 0.3297,
      "step": 12240
    },
    {
      "epoch": 59.84136668700427,
      "grad_norm": 3.356764793395996,
      "learning_rate": 0.00012026229508196722,
      "loss": 0.3698,
      "step": 12260
    },
    {
      "epoch": 59.93898718730934,
      "grad_norm": 3.6063504219055176,
      "learning_rate": 0.0001201311475409836,
      "loss": 0.3479,
      "step": 12280
    },
    {
      "epoch": 60.0366076876144,
      "grad_norm": 3.49246883392334,
      "learning_rate": 0.00012,
      "loss": 0.3394,
      "step": 12300
    },
    {
      "epoch": 60.13422818791946,
      "grad_norm": 4.206517219543457,
      "learning_rate": 0.0001198688524590164,
      "loss": 0.2825,
      "step": 12320
    },
    {
      "epoch": 60.231848688224524,
      "grad_norm": 3.532618999481201,
      "learning_rate": 0.0001197377049180328,
      "loss": 0.3162,
      "step": 12340
    },
    {
      "epoch": 60.32946918852959,
      "grad_norm": 3.6355433464050293,
      "learning_rate": 0.00011960655737704917,
      "loss": 0.3138,
      "step": 12360
    },
    {
      "epoch": 60.427089688834656,
      "grad_norm": 3.1386282444000244,
      "learning_rate": 0.00011947540983606557,
      "loss": 0.335,
      "step": 12380
    },
    {
      "epoch": 60.52471018913972,
      "grad_norm": 2.946789026260376,
      "learning_rate": 0.00011934426229508198,
      "loss": 0.321,
      "step": 12400
    },
    {
      "epoch": 60.62233068944478,
      "grad_norm": 4.056975841522217,
      "learning_rate": 0.00011921311475409838,
      "loss": 0.3647,
      "step": 12420
    },
    {
      "epoch": 60.71995118974985,
      "grad_norm": 3.8449249267578125,
      "learning_rate": 0.00011908196721311476,
      "loss": 0.3189,
      "step": 12440
    },
    {
      "epoch": 60.81757169005491,
      "grad_norm": 3.464470148086548,
      "learning_rate": 0.00011895081967213116,
      "loss": 0.3711,
      "step": 12460
    },
    {
      "epoch": 60.915192190359974,
      "grad_norm": 3.6908376216888428,
      "learning_rate": 0.00011881967213114755,
      "loss": 0.3303,
      "step": 12480
    },
    {
      "epoch": 61.01281269066504,
      "grad_norm": 4.723835468292236,
      "learning_rate": 0.00011868852459016392,
      "loss": 0.305,
      "step": 12500
    },
    {
      "epoch": 61.110433190970106,
      "grad_norm": 3.6280064582824707,
      "learning_rate": 0.00011855737704918033,
      "loss": 0.3103,
      "step": 12520
    },
    {
      "epoch": 61.20805369127517,
      "grad_norm": 3.640411138534546,
      "learning_rate": 0.00011842622950819673,
      "loss": 0.3017,
      "step": 12540
    },
    {
      "epoch": 61.30567419158023,
      "grad_norm": 4.304122447967529,
      "learning_rate": 0.00011829508196721313,
      "loss": 0.3452,
      "step": 12560
    },
    {
      "epoch": 61.40329469188529,
      "grad_norm": 3.4708502292633057,
      "learning_rate": 0.00011816393442622951,
      "loss": 0.3195,
      "step": 12580
    },
    {
      "epoch": 61.50091519219036,
      "grad_norm": 3.257230758666992,
      "learning_rate": 0.00011803278688524591,
      "loss": 0.3115,
      "step": 12600
    },
    {
      "epoch": 61.598535692495425,
      "grad_norm": 3.0247697830200195,
      "learning_rate": 0.0001179016393442623,
      "loss": 0.314,
      "step": 12620
    },
    {
      "epoch": 61.69615619280049,
      "grad_norm": 2.5768277645111084,
      "learning_rate": 0.0001177704918032787,
      "loss": 0.3195,
      "step": 12640
    },
    {
      "epoch": 61.79377669310555,
      "grad_norm": 4.08043909072876,
      "learning_rate": 0.00011763934426229508,
      "loss": 0.3312,
      "step": 12660
    },
    {
      "epoch": 61.89139719341062,
      "grad_norm": 5.277688503265381,
      "learning_rate": 0.00011750819672131148,
      "loss": 0.3607,
      "step": 12680
    },
    {
      "epoch": 61.98901769371568,
      "grad_norm": 4.2203288078308105,
      "learning_rate": 0.00011737704918032789,
      "loss": 0.3233,
      "step": 12700
    },
    {
      "epoch": 62.08663819402074,
      "grad_norm": 3.8487565517425537,
      "learning_rate": 0.00011724590163934426,
      "loss": 0.2777,
      "step": 12720
    },
    {
      "epoch": 62.184258694325806,
      "grad_norm": 3.509904623031616,
      "learning_rate": 0.00011711475409836066,
      "loss": 0.3418,
      "step": 12740
    },
    {
      "epoch": 62.281879194630875,
      "grad_norm": 3.7442731857299805,
      "learning_rate": 0.00011698360655737705,
      "loss": 0.3115,
      "step": 12760
    },
    {
      "epoch": 62.37949969493594,
      "grad_norm": 3.3927197456359863,
      "learning_rate": 0.00011685245901639346,
      "loss": 0.3304,
      "step": 12780
    },
    {
      "epoch": 62.477120195241,
      "grad_norm": 4.529331684112549,
      "learning_rate": 0.00011672131147540983,
      "loss": 0.3246,
      "step": 12800
    },
    {
      "epoch": 62.57474069554606,
      "grad_norm": 2.9640207290649414,
      "learning_rate": 0.00011659016393442623,
      "loss": 0.3228,
      "step": 12820
    },
    {
      "epoch": 62.67236119585113,
      "grad_norm": 4.769125938415527,
      "learning_rate": 0.00011645901639344264,
      "loss": 0.3552,
      "step": 12840
    },
    {
      "epoch": 62.769981696156194,
      "grad_norm": 4.817038536071777,
      "learning_rate": 0.00011632786885245903,
      "loss": 0.269,
      "step": 12860
    },
    {
      "epoch": 62.867602196461256,
      "grad_norm": 4.221793174743652,
      "learning_rate": 0.0001161967213114754,
      "loss": 0.3192,
      "step": 12880
    },
    {
      "epoch": 62.96522269676632,
      "grad_norm": 3.8652241230010986,
      "learning_rate": 0.0001160655737704918,
      "loss": 0.3415,
      "step": 12900
    },
    {
      "epoch": 63.06284319707139,
      "grad_norm": 3.7543272972106934,
      "learning_rate": 0.00011593442622950821,
      "loss": 0.2707,
      "step": 12920
    },
    {
      "epoch": 63.16046369737645,
      "grad_norm": 3.2344672679901123,
      "learning_rate": 0.00011580327868852458,
      "loss": 0.3133,
      "step": 12940
    },
    {
      "epoch": 63.25808419768151,
      "grad_norm": 2.9289095401763916,
      "learning_rate": 0.00011567213114754099,
      "loss": 0.3111,
      "step": 12960
    },
    {
      "epoch": 63.355704697986575,
      "grad_norm": 4.034417629241943,
      "learning_rate": 0.00011554098360655739,
      "loss": 0.3443,
      "step": 12980
    },
    {
      "epoch": 63.453325198291644,
      "grad_norm": 3.571948528289795,
      "learning_rate": 0.00011540983606557378,
      "loss": 0.3061,
      "step": 13000
    },
    {
      "epoch": 63.550945698596706,
      "grad_norm": 3.8213961124420166,
      "learning_rate": 0.00011527868852459016,
      "loss": 0.2943,
      "step": 13020
    },
    {
      "epoch": 63.64856619890177,
      "grad_norm": 4.343628883361816,
      "learning_rate": 0.00011514754098360656,
      "loss": 0.3499,
      "step": 13040
    },
    {
      "epoch": 63.74618669920683,
      "grad_norm": 3.373922348022461,
      "learning_rate": 0.00011501639344262296,
      "loss": 0.3193,
      "step": 13060
    },
    {
      "epoch": 63.8438071995119,
      "grad_norm": 4.19200325012207,
      "learning_rate": 0.00011488524590163936,
      "loss": 0.3221,
      "step": 13080
    },
    {
      "epoch": 63.94142769981696,
      "grad_norm": 3.28054141998291,
      "learning_rate": 0.00011475409836065574,
      "loss": 0.328,
      "step": 13100
    },
    {
      "epoch": 64.03904820012202,
      "grad_norm": 3.887531042098999,
      "learning_rate": 0.00011462295081967214,
      "loss": 0.3161,
      "step": 13120
    },
    {
      "epoch": 64.1366687004271,
      "grad_norm": 3.633239507675171,
      "learning_rate": 0.00011449180327868853,
      "loss": 0.3001,
      "step": 13140
    },
    {
      "epoch": 64.23428920073215,
      "grad_norm": 3.5353002548217773,
      "learning_rate": 0.00011436065573770491,
      "loss": 0.3085,
      "step": 13160
    },
    {
      "epoch": 64.33190970103722,
      "grad_norm": 3.3285086154937744,
      "learning_rate": 0.00011422950819672131,
      "loss": 0.2807,
      "step": 13180
    },
    {
      "epoch": 64.42953020134229,
      "grad_norm": 4.186200141906738,
      "learning_rate": 0.00011409836065573771,
      "loss": 0.3072,
      "step": 13200
    },
    {
      "epoch": 64.52715070164734,
      "grad_norm": 4.000847339630127,
      "learning_rate": 0.00011396721311475412,
      "loss": 0.2794,
      "step": 13220
    },
    {
      "epoch": 64.62477120195241,
      "grad_norm": 3.643327474594116,
      "learning_rate": 0.00011383606557377049,
      "loss": 0.3292,
      "step": 13240
    },
    {
      "epoch": 64.72239170225747,
      "grad_norm": 4.381191730499268,
      "learning_rate": 0.00011370491803278688,
      "loss": 0.3311,
      "step": 13260
    },
    {
      "epoch": 64.82001220256254,
      "grad_norm": 3.9231462478637695,
      "learning_rate": 0.00011357377049180329,
      "loss": 0.3239,
      "step": 13280
    },
    {
      "epoch": 64.9176327028676,
      "grad_norm": 3.68272066116333,
      "learning_rate": 0.00011344262295081969,
      "loss": 0.3573,
      "step": 13300
    },
    {
      "epoch": 65.01525320317266,
      "grad_norm": 3.5405211448669434,
      "learning_rate": 0.00011331147540983606,
      "loss": 0.3026,
      "step": 13320
    },
    {
      "epoch": 65.11287370347773,
      "grad_norm": 4.24603796005249,
      "learning_rate": 0.00011318032786885247,
      "loss": 0.2984,
      "step": 13340
    },
    {
      "epoch": 65.2104942037828,
      "grad_norm": 3.9128637313842773,
      "learning_rate": 0.00011304918032786887,
      "loss": 0.3203,
      "step": 13360
    },
    {
      "epoch": 65.30811470408786,
      "grad_norm": 3.794922113418579,
      "learning_rate": 0.00011291803278688525,
      "loss": 0.2696,
      "step": 13380
    },
    {
      "epoch": 65.40573520439293,
      "grad_norm": 7.95392370223999,
      "learning_rate": 0.00011278688524590164,
      "loss": 0.3349,
      "step": 13400
    },
    {
      "epoch": 65.50335570469798,
      "grad_norm": 4.138427257537842,
      "learning_rate": 0.00011265573770491804,
      "loss": 0.303,
      "step": 13420
    },
    {
      "epoch": 65.60097620500305,
      "grad_norm": 3.470979690551758,
      "learning_rate": 0.00011252459016393444,
      "loss": 0.3019,
      "step": 13440
    },
    {
      "epoch": 65.69859670530812,
      "grad_norm": 3.5186235904693604,
      "learning_rate": 0.00011239344262295082,
      "loss": 0.3254,
      "step": 13460
    },
    {
      "epoch": 65.79621720561317,
      "grad_norm": 5.092533111572266,
      "learning_rate": 0.00011226229508196722,
      "loss": 0.3167,
      "step": 13480
    },
    {
      "epoch": 65.89383770591824,
      "grad_norm": 4.472383499145508,
      "learning_rate": 0.00011213114754098362,
      "loss": 0.3102,
      "step": 13500
    },
    {
      "epoch": 65.99145820622331,
      "grad_norm": 4.198476314544678,
      "learning_rate": 0.00011200000000000001,
      "loss": 0.3085,
      "step": 13520
    },
    {
      "epoch": 66.08907870652837,
      "grad_norm": 4.235730171203613,
      "learning_rate": 0.00011186885245901639,
      "loss": 0.2754,
      "step": 13540
    },
    {
      "epoch": 66.18669920683344,
      "grad_norm": 3.673928737640381,
      "learning_rate": 0.00011173770491803279,
      "loss": 0.2883,
      "step": 13560
    },
    {
      "epoch": 66.2843197071385,
      "grad_norm": 3.8741283416748047,
      "learning_rate": 0.0001116065573770492,
      "loss": 0.3055,
      "step": 13580
    },
    {
      "epoch": 66.38194020744356,
      "grad_norm": 3.2982194423675537,
      "learning_rate": 0.00011147540983606557,
      "loss": 0.3309,
      "step": 13600
    },
    {
      "epoch": 66.47956070774863,
      "grad_norm": 3.1477532386779785,
      "learning_rate": 0.00011134426229508197,
      "loss": 0.299,
      "step": 13620
    },
    {
      "epoch": 66.57718120805369,
      "grad_norm": 3.723947525024414,
      "learning_rate": 0.00011121311475409838,
      "loss": 0.3016,
      "step": 13640
    },
    {
      "epoch": 66.67480170835876,
      "grad_norm": 4.094952583312988,
      "learning_rate": 0.00011108196721311476,
      "loss": 0.3045,
      "step": 13660
    },
    {
      "epoch": 66.77242220866383,
      "grad_norm": 3.607595205307007,
      "learning_rate": 0.00011095081967213114,
      "loss": 0.2983,
      "step": 13680
    },
    {
      "epoch": 66.87004270896888,
      "grad_norm": 4.301547527313232,
      "learning_rate": 0.00011081967213114754,
      "loss": 0.2783,
      "step": 13700
    },
    {
      "epoch": 66.96766320927395,
      "grad_norm": 3.570128917694092,
      "learning_rate": 0.00011068852459016395,
      "loss": 0.3547,
      "step": 13720
    },
    {
      "epoch": 67.065283709579,
      "grad_norm": 4.2378621101379395,
      "learning_rate": 0.00011055737704918035,
      "loss": 0.3097,
      "step": 13740
    },
    {
      "epoch": 67.16290420988408,
      "grad_norm": 3.6283559799194336,
      "learning_rate": 0.00011042622950819672,
      "loss": 0.2773,
      "step": 13760
    },
    {
      "epoch": 67.26052471018915,
      "grad_norm": 3.317201614379883,
      "learning_rate": 0.00011029508196721311,
      "loss": 0.3179,
      "step": 13780
    },
    {
      "epoch": 67.3581452104942,
      "grad_norm": 3.120859146118164,
      "learning_rate": 0.00011016393442622952,
      "loss": 0.3239,
      "step": 13800
    },
    {
      "epoch": 67.45576571079927,
      "grad_norm": 3.163276433944702,
      "learning_rate": 0.00011003278688524589,
      "loss": 0.2875,
      "step": 13820
    },
    {
      "epoch": 67.55338621110434,
      "grad_norm": 3.597426652908325,
      "learning_rate": 0.0001099016393442623,
      "loss": 0.2739,
      "step": 13840
    },
    {
      "epoch": 67.6510067114094,
      "grad_norm": 3.572763204574585,
      "learning_rate": 0.0001097704918032787,
      "loss": 0.3009,
      "step": 13860
    },
    {
      "epoch": 67.74862721171446,
      "grad_norm": 3.660034656524658,
      "learning_rate": 0.0001096393442622951,
      "loss": 0.3089,
      "step": 13880
    },
    {
      "epoch": 67.84624771201952,
      "grad_norm": 4.177897930145264,
      "learning_rate": 0.00010950819672131148,
      "loss": 0.3205,
      "step": 13900
    },
    {
      "epoch": 67.94386821232459,
      "grad_norm": 5.586677074432373,
      "learning_rate": 0.00010937704918032787,
      "loss": 0.3263,
      "step": 13920
    },
    {
      "epoch": 68.04148871262966,
      "grad_norm": 2.8578503131866455,
      "learning_rate": 0.00010924590163934427,
      "loss": 0.3139,
      "step": 13940
    },
    {
      "epoch": 68.13910921293471,
      "grad_norm": 3.129765748977661,
      "learning_rate": 0.00010911475409836067,
      "loss": 0.2729,
      "step": 13960
    },
    {
      "epoch": 68.23672971323978,
      "grad_norm": 3.238567590713501,
      "learning_rate": 0.00010898360655737705,
      "loss": 0.3008,
      "step": 13980
    },
    {
      "epoch": 68.33435021354484,
      "grad_norm": 3.5786325931549072,
      "learning_rate": 0.00010885245901639345,
      "loss": 0.2818,
      "step": 14000
    },
    {
      "epoch": 68.4319707138499,
      "grad_norm": 3.5092949867248535,
      "learning_rate": 0.00010872131147540985,
      "loss": 0.3034,
      "step": 14020
    },
    {
      "epoch": 68.52959121415498,
      "grad_norm": 4.172147274017334,
      "learning_rate": 0.00010859016393442623,
      "loss": 0.2917,
      "step": 14040
    },
    {
      "epoch": 68.62721171446003,
      "grad_norm": 3.5723907947540283,
      "learning_rate": 0.00010845901639344262,
      "loss": 0.3119,
      "step": 14060
    },
    {
      "epoch": 68.7248322147651,
      "grad_norm": 3.2721340656280518,
      "learning_rate": 0.00010832786885245902,
      "loss": 0.2698,
      "step": 14080
    },
    {
      "epoch": 68.82245271507017,
      "grad_norm": 4.181795597076416,
      "learning_rate": 0.00010819672131147543,
      "loss": 0.3181,
      "step": 14100
    },
    {
      "epoch": 68.92007321537523,
      "grad_norm": 4.145654678344727,
      "learning_rate": 0.0001080655737704918,
      "loss": 0.2837,
      "step": 14120
    },
    {
      "epoch": 69.0176937156803,
      "grad_norm": 4.098430633544922,
      "learning_rate": 0.0001079344262295082,
      "loss": 0.3281,
      "step": 14140
    },
    {
      "epoch": 69.11531421598535,
      "grad_norm": 3.6833229064941406,
      "learning_rate": 0.00010780327868852461,
      "loss": 0.3059,
      "step": 14160
    },
    {
      "epoch": 69.21293471629042,
      "grad_norm": 3.802940845489502,
      "learning_rate": 0.00010767213114754098,
      "loss": 0.2648,
      "step": 14180
    },
    {
      "epoch": 69.31055521659549,
      "grad_norm": 2.9509127140045166,
      "learning_rate": 0.00010754098360655737,
      "loss": 0.2828,
      "step": 14200
    },
    {
      "epoch": 69.40817571690054,
      "grad_norm": 3.6474950313568115,
      "learning_rate": 0.00010740983606557378,
      "loss": 0.2848,
      "step": 14220
    },
    {
      "epoch": 69.50579621720561,
      "grad_norm": 3.762017250061035,
      "learning_rate": 0.00010727868852459018,
      "loss": 0.3305,
      "step": 14240
    },
    {
      "epoch": 69.60341671751068,
      "grad_norm": 3.3296637535095215,
      "learning_rate": 0.00010714754098360655,
      "loss": 0.2885,
      "step": 14260
    },
    {
      "epoch": 69.70103721781574,
      "grad_norm": 4.1584296226501465,
      "learning_rate": 0.00010701639344262296,
      "loss": 0.3162,
      "step": 14280
    },
    {
      "epoch": 69.79865771812081,
      "grad_norm": 3.365910768508911,
      "learning_rate": 0.00010688524590163935,
      "loss": 0.2855,
      "step": 14300
    },
    {
      "epoch": 69.89627821842586,
      "grad_norm": 3.9548261165618896,
      "learning_rate": 0.00010675409836065575,
      "loss": 0.3001,
      "step": 14320
    },
    {
      "epoch": 69.99389871873093,
      "grad_norm": 4.556044578552246,
      "learning_rate": 0.00010662295081967212,
      "loss": 0.2855,
      "step": 14340
    },
    {
      "epoch": 70.091519219036,
      "grad_norm": 4.781431198120117,
      "learning_rate": 0.00010649180327868853,
      "loss": 0.2741,
      "step": 14360
    },
    {
      "epoch": 70.18913971934106,
      "grad_norm": 2.645362377166748,
      "learning_rate": 0.00010636065573770493,
      "loss": 0.2885,
      "step": 14380
    },
    {
      "epoch": 70.28676021964613,
      "grad_norm": 4.382370471954346,
      "learning_rate": 0.0001062295081967213,
      "loss": 0.3038,
      "step": 14400
    },
    {
      "epoch": 70.3843807199512,
      "grad_norm": 2.908954381942749,
      "learning_rate": 0.00010609836065573771,
      "loss": 0.2765,
      "step": 14420
    },
    {
      "epoch": 70.48200122025625,
      "grad_norm": 3.298417091369629,
      "learning_rate": 0.0001059672131147541,
      "loss": 0.2986,
      "step": 14440
    },
    {
      "epoch": 70.57962172056132,
      "grad_norm": 3.324769973754883,
      "learning_rate": 0.0001058360655737705,
      "loss": 0.2712,
      "step": 14460
    },
    {
      "epoch": 70.67724222086638,
      "grad_norm": 4.584010124206543,
      "learning_rate": 0.00010570491803278688,
      "loss": 0.3049,
      "step": 14480
    },
    {
      "epoch": 70.77486272117144,
      "grad_norm": 3.648829221725464,
      "learning_rate": 0.00010557377049180328,
      "loss": 0.3169,
      "step": 14500
    },
    {
      "epoch": 70.87248322147651,
      "grad_norm": 3.729729175567627,
      "learning_rate": 0.00010544262295081968,
      "loss": 0.2818,
      "step": 14520
    },
    {
      "epoch": 70.97010372178157,
      "grad_norm": 3.572127103805542,
      "learning_rate": 0.00010531147540983609,
      "loss": 0.303,
      "step": 14540
    },
    {
      "epoch": 71.06772422208664,
      "grad_norm": 3.4848947525024414,
      "learning_rate": 0.00010518032786885246,
      "loss": 0.2796,
      "step": 14560
    },
    {
      "epoch": 71.16534472239171,
      "grad_norm": 3.478776454925537,
      "learning_rate": 0.00010504918032786885,
      "loss": 0.2718,
      "step": 14580
    },
    {
      "epoch": 71.26296522269676,
      "grad_norm": 3.6878888607025146,
      "learning_rate": 0.00010491803278688525,
      "loss": 0.2611,
      "step": 14600
    },
    {
      "epoch": 71.36058572300183,
      "grad_norm": 3.0455853939056396,
      "learning_rate": 0.00010478688524590163,
      "loss": 0.3063,
      "step": 14620
    },
    {
      "epoch": 71.45820622330689,
      "grad_norm": 4.242178916931152,
      "learning_rate": 0.00010465573770491803,
      "loss": 0.2811,
      "step": 14640
    },
    {
      "epoch": 71.55582672361196,
      "grad_norm": 2.9485058784484863,
      "learning_rate": 0.00010452459016393444,
      "loss": 0.278,
      "step": 14660
    },
    {
      "epoch": 71.65344722391703,
      "grad_norm": 4.489981651306152,
      "learning_rate": 0.00010439344262295083,
      "loss": 0.2833,
      "step": 14680
    },
    {
      "epoch": 71.75106772422208,
      "grad_norm": 3.748349666595459,
      "learning_rate": 0.00010426229508196721,
      "loss": 0.3362,
      "step": 14700
    },
    {
      "epoch": 71.84868822452715,
      "grad_norm": 3.5393285751342773,
      "learning_rate": 0.0001041311475409836,
      "loss": 0.2803,
      "step": 14720
    },
    {
      "epoch": 71.94630872483222,
      "grad_norm": 3.8631341457366943,
      "learning_rate": 0.00010400000000000001,
      "loss": 0.3103,
      "step": 14740
    },
    {
      "epoch": 72.04392922513728,
      "grad_norm": 3.3096232414245605,
      "learning_rate": 0.00010386885245901641,
      "loss": 0.3059,
      "step": 14760
    },
    {
      "epoch": 72.14154972544235,
      "grad_norm": 3.3514437675476074,
      "learning_rate": 0.00010373770491803279,
      "loss": 0.2832,
      "step": 14780
    },
    {
      "epoch": 72.2391702257474,
      "grad_norm": 3.548172950744629,
      "learning_rate": 0.00010360655737704919,
      "loss": 0.3087,
      "step": 14800
    },
    {
      "epoch": 72.33679072605247,
      "grad_norm": 2.9847452640533447,
      "learning_rate": 0.00010347540983606558,
      "loss": 0.2565,
      "step": 14820
    },
    {
      "epoch": 72.43441122635754,
      "grad_norm": 2.827442169189453,
      "learning_rate": 0.00010334426229508197,
      "loss": 0.2622,
      "step": 14840
    },
    {
      "epoch": 72.5320317266626,
      "grad_norm": 5.385785102844238,
      "learning_rate": 0.00010321311475409836,
      "loss": 0.2684,
      "step": 14860
    },
    {
      "epoch": 72.62965222696766,
      "grad_norm": 2.8758130073547363,
      "learning_rate": 0.00010308196721311476,
      "loss": 0.3032,
      "step": 14880
    },
    {
      "epoch": 72.72727272727273,
      "grad_norm": 3.0120010375976562,
      "learning_rate": 0.00010295081967213116,
      "loss": 0.2822,
      "step": 14900
    },
    {
      "epoch": 72.82489322757779,
      "grad_norm": 4.078627109527588,
      "learning_rate": 0.00010281967213114754,
      "loss": 0.2852,
      "step": 14920
    },
    {
      "epoch": 72.92251372788286,
      "grad_norm": 3.6309173107147217,
      "learning_rate": 0.00010268852459016394,
      "loss": 0.2961,
      "step": 14940
    },
    {
      "epoch": 73.02013422818791,
      "grad_norm": 5.154388427734375,
      "learning_rate": 0.00010255737704918033,
      "loss": 0.2926,
      "step": 14960
    },
    {
      "epoch": 73.11775472849298,
      "grad_norm": 3.6091132164001465,
      "learning_rate": 0.00010242622950819673,
      "loss": 0.2855,
      "step": 14980
    },
    {
      "epoch": 73.21537522879805,
      "grad_norm": 3.872344970703125,
      "learning_rate": 0.00010229508196721311,
      "loss": 0.2728,
      "step": 15000
    },
    {
      "epoch": 73.31299572910311,
      "grad_norm": 3.2357425689697266,
      "learning_rate": 0.00010216393442622951,
      "loss": 0.3022,
      "step": 15020
    },
    {
      "epoch": 73.41061622940818,
      "grad_norm": 3.10544490814209,
      "learning_rate": 0.00010203278688524592,
      "loss": 0.2792,
      "step": 15040
    },
    {
      "epoch": 73.50823672971325,
      "grad_norm": 3.796975612640381,
      "learning_rate": 0.00010190163934426229,
      "loss": 0.2975,
      "step": 15060
    },
    {
      "epoch": 73.6058572300183,
      "grad_norm": 4.492702484130859,
      "learning_rate": 0.0001017704918032787,
      "loss": 0.2628,
      "step": 15080
    },
    {
      "epoch": 73.70347773032337,
      "grad_norm": 3.9376978874206543,
      "learning_rate": 0.00010163934426229508,
      "loss": 0.2428,
      "step": 15100
    },
    {
      "epoch": 73.80109823062843,
      "grad_norm": 4.3148980140686035,
      "learning_rate": 0.00010150819672131149,
      "loss": 0.3151,
      "step": 15120
    },
    {
      "epoch": 73.8987187309335,
      "grad_norm": 3.7548837661743164,
      "learning_rate": 0.00010137704918032786,
      "loss": 0.2676,
      "step": 15140
    },
    {
      "epoch": 73.99633923123857,
      "grad_norm": 3.9053032398223877,
      "learning_rate": 0.00010124590163934427,
      "loss": 0.298,
      "step": 15160
    },
    {
      "epoch": 74.09395973154362,
      "grad_norm": 4.0978007316589355,
      "learning_rate": 0.00010111475409836067,
      "loss": 0.2597,
      "step": 15180
    },
    {
      "epoch": 74.19158023184869,
      "grad_norm": 2.8652732372283936,
      "learning_rate": 0.00010098360655737706,
      "loss": 0.2842,
      "step": 15200
    },
    {
      "epoch": 74.28920073215376,
      "grad_norm": 3.180609941482544,
      "learning_rate": 0.00010085245901639345,
      "loss": 0.2737,
      "step": 15220
    },
    {
      "epoch": 74.38682123245881,
      "grad_norm": 4.571623802185059,
      "learning_rate": 0.00010072131147540984,
      "loss": 0.3213,
      "step": 15240
    },
    {
      "epoch": 74.48444173276388,
      "grad_norm": 3.073421001434326,
      "learning_rate": 0.00010059016393442624,
      "loss": 0.2922,
      "step": 15260
    },
    {
      "epoch": 74.58206223306894,
      "grad_norm": 5.4209980964660645,
      "learning_rate": 0.00010045901639344261,
      "loss": 0.2575,
      "step": 15280
    },
    {
      "epoch": 74.67968273337401,
      "grad_norm": 4.5742621421813965,
      "learning_rate": 0.00010032786885245902,
      "loss": 0.3044,
      "step": 15300
    },
    {
      "epoch": 74.77730323367908,
      "grad_norm": 3.712390422821045,
      "learning_rate": 0.00010019672131147542,
      "loss": 0.2891,
      "step": 15320
    },
    {
      "epoch": 74.87492373398413,
      "grad_norm": 3.149919033050537,
      "learning_rate": 0.00010006557377049181,
      "loss": 0.275,
      "step": 15340
    },
    {
      "epoch": 74.9725442342892,
      "grad_norm": 4.481144428253174,
      "learning_rate": 9.99344262295082e-05,
      "loss": 0.2544,
      "step": 15360
    },
    {
      "epoch": 75.07016473459427,
      "grad_norm": 3.6176226139068604,
      "learning_rate": 9.980327868852459e-05,
      "loss": 0.2656,
      "step": 15380
    },
    {
      "epoch": 75.16778523489933,
      "grad_norm": 3.986781120300293,
      "learning_rate": 9.967213114754099e-05,
      "loss": 0.2712,
      "step": 15400
    },
    {
      "epoch": 75.2654057352044,
      "grad_norm": 2.7806153297424316,
      "learning_rate": 9.954098360655738e-05,
      "loss": 0.2542,
      "step": 15420
    },
    {
      "epoch": 75.36302623550945,
      "grad_norm": 4.493511199951172,
      "learning_rate": 9.940983606557378e-05,
      "loss": 0.2871,
      "step": 15440
    },
    {
      "epoch": 75.46064673581452,
      "grad_norm": 4.610682964324951,
      "learning_rate": 9.927868852459017e-05,
      "loss": 0.2699,
      "step": 15460
    },
    {
      "epoch": 75.55826723611959,
      "grad_norm": 3.7209839820861816,
      "learning_rate": 9.914754098360656e-05,
      "loss": 0.3086,
      "step": 15480
    },
    {
      "epoch": 75.65588773642465,
      "grad_norm": 4.1248931884765625,
      "learning_rate": 9.901639344262295e-05,
      "loss": 0.2744,
      "step": 15500
    },
    {
      "epoch": 75.75350823672972,
      "grad_norm": 3.7576661109924316,
      "learning_rate": 9.888524590163934e-05,
      "loss": 0.2791,
      "step": 15520
    },
    {
      "epoch": 75.85112873703477,
      "grad_norm": 3.6697514057159424,
      "learning_rate": 9.875409836065574e-05,
      "loss": 0.2779,
      "step": 15540
    },
    {
      "epoch": 75.94874923733984,
      "grad_norm": 4.156905651092529,
      "learning_rate": 9.862295081967213e-05,
      "loss": 0.2806,
      "step": 15560
    },
    {
      "epoch": 76.04636973764491,
      "grad_norm": 3.244154214859009,
      "learning_rate": 9.849180327868854e-05,
      "loss": 0.2772,
      "step": 15580
    },
    {
      "epoch": 76.14399023794996,
      "grad_norm": 4.677572250366211,
      "learning_rate": 9.836065573770493e-05,
      "loss": 0.2695,
      "step": 15600
    },
    {
      "epoch": 76.24161073825503,
      "grad_norm": 3.442591667175293,
      "learning_rate": 9.822950819672132e-05,
      "loss": 0.309,
      "step": 15620
    },
    {
      "epoch": 76.3392312385601,
      "grad_norm": 2.930734395980835,
      "learning_rate": 9.80983606557377e-05,
      "loss": 0.2543,
      "step": 15640
    },
    {
      "epoch": 76.43685173886516,
      "grad_norm": 3.7919161319732666,
      "learning_rate": 9.796721311475411e-05,
      "loss": 0.2802,
      "step": 15660
    },
    {
      "epoch": 76.53447223917023,
      "grad_norm": 3.598411798477173,
      "learning_rate": 9.78360655737705e-05,
      "loss": 0.2433,
      "step": 15680
    },
    {
      "epoch": 76.63209273947528,
      "grad_norm": 3.6771163940429688,
      "learning_rate": 9.770491803278689e-05,
      "loss": 0.2529,
      "step": 15700
    },
    {
      "epoch": 76.72971323978035,
      "grad_norm": 3.1646664142608643,
      "learning_rate": 9.757377049180329e-05,
      "loss": 0.2801,
      "step": 15720
    },
    {
      "epoch": 76.82733374008542,
      "grad_norm": 2.325819969177246,
      "learning_rate": 9.744262295081968e-05,
      "loss": 0.2609,
      "step": 15740
    },
    {
      "epoch": 76.92495424039048,
      "grad_norm": 4.598663330078125,
      "learning_rate": 9.731147540983607e-05,
      "loss": 0.2936,
      "step": 15760
    },
    {
      "epoch": 77.02257474069555,
      "grad_norm": 2.9897961616516113,
      "learning_rate": 9.718032786885246e-05,
      "loss": 0.2891,
      "step": 15780
    },
    {
      "epoch": 77.12019524100062,
      "grad_norm": 4.264781951904297,
      "learning_rate": 9.704918032786886e-05,
      "loss": 0.2677,
      "step": 15800
    },
    {
      "epoch": 77.21781574130567,
      "grad_norm": 3.616536855697632,
      "learning_rate": 9.691803278688525e-05,
      "loss": 0.2282,
      "step": 15820
    },
    {
      "epoch": 77.31543624161074,
      "grad_norm": 3.4226481914520264,
      "learning_rate": 9.678688524590165e-05,
      "loss": 0.2792,
      "step": 15840
    },
    {
      "epoch": 77.4130567419158,
      "grad_norm": 3.012357473373413,
      "learning_rate": 9.665573770491804e-05,
      "loss": 0.2644,
      "step": 15860
    },
    {
      "epoch": 77.51067724222086,
      "grad_norm": 3.582298517227173,
      "learning_rate": 9.652459016393443e-05,
      "loss": 0.2944,
      "step": 15880
    },
    {
      "epoch": 77.60829774252593,
      "grad_norm": 4.138154983520508,
      "learning_rate": 9.639344262295082e-05,
      "loss": 0.3097,
      "step": 15900
    },
    {
      "epoch": 77.70591824283099,
      "grad_norm": 4.919800281524658,
      "learning_rate": 9.626229508196721e-05,
      "loss": 0.2666,
      "step": 15920
    },
    {
      "epoch": 77.80353874313606,
      "grad_norm": 3.3038241863250732,
      "learning_rate": 9.613114754098361e-05,
      "loss": 0.2909,
      "step": 15940
    },
    {
      "epoch": 77.90115924344113,
      "grad_norm": 2.6817522048950195,
      "learning_rate": 9.6e-05,
      "loss": 0.2565,
      "step": 15960
    },
    {
      "epoch": 77.99877974374618,
      "grad_norm": 4.992214679718018,
      "learning_rate": 9.58688524590164e-05,
      "loss": 0.2767,
      "step": 15980
    },
    {
      "epoch": 78.09640024405125,
      "grad_norm": 3.589238166809082,
      "learning_rate": 9.57377049180328e-05,
      "loss": 0.2483,
      "step": 16000
    },
    {
      "epoch": 78.19402074435631,
      "grad_norm": 4.856943607330322,
      "learning_rate": 9.560655737704918e-05,
      "loss": 0.2698,
      "step": 16020
    },
    {
      "epoch": 78.29164124466138,
      "grad_norm": 3.963756561279297,
      "learning_rate": 9.547540983606557e-05,
      "loss": 0.2522,
      "step": 16040
    },
    {
      "epoch": 78.38926174496645,
      "grad_norm": 3.843501567840576,
      "learning_rate": 9.534426229508198e-05,
      "loss": 0.2502,
      "step": 16060
    },
    {
      "epoch": 78.4868822452715,
      "grad_norm": 3.1323421001434326,
      "learning_rate": 9.521311475409837e-05,
      "loss": 0.2693,
      "step": 16080
    },
    {
      "epoch": 78.58450274557657,
      "grad_norm": 3.3395233154296875,
      "learning_rate": 9.508196721311476e-05,
      "loss": 0.2806,
      "step": 16100
    },
    {
      "epoch": 78.68212324588164,
      "grad_norm": 3.2387075424194336,
      "learning_rate": 9.495081967213116e-05,
      "loss": 0.266,
      "step": 16120
    },
    {
      "epoch": 78.7797437461867,
      "grad_norm": 3.9725232124328613,
      "learning_rate": 9.481967213114755e-05,
      "loss": 0.2753,
      "step": 16140
    },
    {
      "epoch": 78.87736424649177,
      "grad_norm": 3.88059401512146,
      "learning_rate": 9.468852459016394e-05,
      "loss": 0.3028,
      "step": 16160
    },
    {
      "epoch": 78.97498474679682,
      "grad_norm": 4.623359680175781,
      "learning_rate": 9.455737704918033e-05,
      "loss": 0.2816,
      "step": 16180
    },
    {
      "epoch": 79.07260524710189,
      "grad_norm": 3.237917423248291,
      "learning_rate": 9.442622950819673e-05,
      "loss": 0.2545,
      "step": 16200
    },
    {
      "epoch": 79.17022574740696,
      "grad_norm": 3.4442942142486572,
      "learning_rate": 9.429508196721312e-05,
      "loss": 0.2596,
      "step": 16220
    },
    {
      "epoch": 79.26784624771201,
      "grad_norm": 4.398017406463623,
      "learning_rate": 9.416393442622952e-05,
      "loss": 0.2731,
      "step": 16240
    },
    {
      "epoch": 79.36546674801708,
      "grad_norm": 3.892063856124878,
      "learning_rate": 9.403278688524591e-05,
      "loss": 0.266,
      "step": 16260
    },
    {
      "epoch": 79.46308724832215,
      "grad_norm": 4.046624183654785,
      "learning_rate": 9.39016393442623e-05,
      "loss": 0.2368,
      "step": 16280
    },
    {
      "epoch": 79.56070774862721,
      "grad_norm": 5.123999118804932,
      "learning_rate": 9.377049180327869e-05,
      "loss": 0.2409,
      "step": 16300
    },
    {
      "epoch": 79.65832824893228,
      "grad_norm": 3.1601643562316895,
      "learning_rate": 9.363934426229508e-05,
      "loss": 0.2525,
      "step": 16320
    },
    {
      "epoch": 79.75594874923733,
      "grad_norm": 4.788114070892334,
      "learning_rate": 9.350819672131148e-05,
      "loss": 0.2998,
      "step": 16340
    },
    {
      "epoch": 79.8535692495424,
      "grad_norm": 3.9707436561584473,
      "learning_rate": 9.337704918032787e-05,
      "loss": 0.2724,
      "step": 16360
    },
    {
      "epoch": 79.95118974984747,
      "grad_norm": 3.7388081550598145,
      "learning_rate": 9.324590163934427e-05,
      "loss": 0.2823,
      "step": 16380
    },
    {
      "epoch": 80.04881025015253,
      "grad_norm": 2.7105419635772705,
      "learning_rate": 9.311475409836066e-05,
      "loss": 0.2602,
      "step": 16400
    },
    {
      "epoch": 80.1464307504576,
      "grad_norm": 3.460477113723755,
      "learning_rate": 9.298360655737705e-05,
      "loss": 0.2428,
      "step": 16420
    },
    {
      "epoch": 80.24405125076267,
      "grad_norm": 3.526585817337036,
      "learning_rate": 9.285245901639344e-05,
      "loss": 0.2982,
      "step": 16440
    },
    {
      "epoch": 80.34167175106772,
      "grad_norm": 3.746425151824951,
      "learning_rate": 9.272131147540985e-05,
      "loss": 0.2365,
      "step": 16460
    },
    {
      "epoch": 80.43929225137279,
      "grad_norm": 3.2804489135742188,
      "learning_rate": 9.259016393442623e-05,
      "loss": 0.2832,
      "step": 16480
    },
    {
      "epoch": 80.53691275167785,
      "grad_norm": 3.136016607284546,
      "learning_rate": 9.245901639344264e-05,
      "loss": 0.2744,
      "step": 16500
    },
    {
      "epoch": 80.63453325198292,
      "grad_norm": 4.361492156982422,
      "learning_rate": 9.232786885245903e-05,
      "loss": 0.2726,
      "step": 16520
    },
    {
      "epoch": 80.73215375228799,
      "grad_norm": 3.3838353157043457,
      "learning_rate": 9.21967213114754e-05,
      "loss": 0.2651,
      "step": 16540
    },
    {
      "epoch": 80.82977425259304,
      "grad_norm": 3.8092405796051025,
      "learning_rate": 9.20655737704918e-05,
      "loss": 0.2519,
      "step": 16560
    },
    {
      "epoch": 80.92739475289811,
      "grad_norm": 4.455862522125244,
      "learning_rate": 9.19344262295082e-05,
      "loss": 0.2518,
      "step": 16580
    },
    {
      "epoch": 81.02501525320318,
      "grad_norm": 3.073539972305298,
      "learning_rate": 9.18032786885246e-05,
      "loss": 0.272,
      "step": 16600
    },
    {
      "epoch": 81.12263575350823,
      "grad_norm": 4.361839294433594,
      "learning_rate": 9.167213114754099e-05,
      "loss": 0.2468,
      "step": 16620
    },
    {
      "epoch": 81.2202562538133,
      "grad_norm": 5.322413444519043,
      "learning_rate": 9.154098360655739e-05,
      "loss": 0.2339,
      "step": 16640
    },
    {
      "epoch": 81.31787675411836,
      "grad_norm": 4.6976704597473145,
      "learning_rate": 9.140983606557378e-05,
      "loss": 0.27,
      "step": 16660
    },
    {
      "epoch": 81.41549725442343,
      "grad_norm": 3.608119487762451,
      "learning_rate": 9.127868852459017e-05,
      "loss": 0.2615,
      "step": 16680
    },
    {
      "epoch": 81.5131177547285,
      "grad_norm": 3.837738513946533,
      "learning_rate": 9.114754098360656e-05,
      "loss": 0.2534,
      "step": 16700
    },
    {
      "epoch": 81.61073825503355,
      "grad_norm": 3.750638723373413,
      "learning_rate": 9.101639344262296e-05,
      "loss": 0.2533,
      "step": 16720
    },
    {
      "epoch": 81.70835875533862,
      "grad_norm": 3.916907548904419,
      "learning_rate": 9.088524590163935e-05,
      "loss": 0.2847,
      "step": 16740
    },
    {
      "epoch": 81.80597925564369,
      "grad_norm": 3.8545727729797363,
      "learning_rate": 9.075409836065574e-05,
      "loss": 0.2554,
      "step": 16760
    },
    {
      "epoch": 81.90359975594875,
      "grad_norm": 2.8878276348114014,
      "learning_rate": 9.062295081967214e-05,
      "loss": 0.2793,
      "step": 16780
    },
    {
      "epoch": 82.00122025625382,
      "grad_norm": 3.6857919692993164,
      "learning_rate": 9.049180327868852e-05,
      "loss": 0.2703,
      "step": 16800
    },
    {
      "epoch": 82.09884075655887,
      "grad_norm": 3.670179605484009,
      "learning_rate": 9.036065573770492e-05,
      "loss": 0.2641,
      "step": 16820
    },
    {
      "epoch": 82.19646125686394,
      "grad_norm": 3.4155802726745605,
      "learning_rate": 9.022950819672131e-05,
      "loss": 0.2528,
      "step": 16840
    },
    {
      "epoch": 82.29408175716901,
      "grad_norm": 3.139631986618042,
      "learning_rate": 9.009836065573771e-05,
      "loss": 0.2485,
      "step": 16860
    },
    {
      "epoch": 82.39170225747407,
      "grad_norm": 3.919480800628662,
      "learning_rate": 8.99672131147541e-05,
      "loss": 0.2586,
      "step": 16880
    },
    {
      "epoch": 82.48932275777914,
      "grad_norm": 4.068678379058838,
      "learning_rate": 8.98360655737705e-05,
      "loss": 0.2836,
      "step": 16900
    },
    {
      "epoch": 82.5869432580842,
      "grad_norm": 3.544682502746582,
      "learning_rate": 8.97049180327869e-05,
      "loss": 0.2515,
      "step": 16920
    },
    {
      "epoch": 82.68456375838926,
      "grad_norm": 3.8421828746795654,
      "learning_rate": 8.957377049180328e-05,
      "loss": 0.2654,
      "step": 16940
    },
    {
      "epoch": 82.78218425869433,
      "grad_norm": 3.0508570671081543,
      "learning_rate": 8.944262295081967e-05,
      "loss": 0.2742,
      "step": 16960
    },
    {
      "epoch": 82.87980475899938,
      "grad_norm": 3.601579427719116,
      "learning_rate": 8.931147540983606e-05,
      "loss": 0.2542,
      "step": 16980
    },
    {
      "epoch": 82.97742525930445,
      "grad_norm": 3.657724618911743,
      "learning_rate": 8.918032786885247e-05,
      "loss": 0.2253,
      "step": 17000
    },
    {
      "epoch": 83.07504575960952,
      "grad_norm": 5.093318462371826,
      "learning_rate": 8.904918032786886e-05,
      "loss": 0.2429,
      "step": 17020
    },
    {
      "epoch": 83.17266625991458,
      "grad_norm": 3.160510301589966,
      "learning_rate": 8.891803278688526e-05,
      "loss": 0.2416,
      "step": 17040
    },
    {
      "epoch": 83.27028676021965,
      "grad_norm": 3.9046666622161865,
      "learning_rate": 8.878688524590163e-05,
      "loss": 0.2526,
      "step": 17060
    },
    {
      "epoch": 83.36790726052472,
      "grad_norm": 4.196876525878906,
      "learning_rate": 8.865573770491804e-05,
      "loss": 0.2518,
      "step": 17080
    },
    {
      "epoch": 83.46552776082977,
      "grad_norm": 3.5083820819854736,
      "learning_rate": 8.852459016393443e-05,
      "loss": 0.2392,
      "step": 17100
    },
    {
      "epoch": 83.56314826113484,
      "grad_norm": 3.099393606185913,
      "learning_rate": 8.839344262295083e-05,
      "loss": 0.2783,
      "step": 17120
    },
    {
      "epoch": 83.6607687614399,
      "grad_norm": 3.532540798187256,
      "learning_rate": 8.826229508196722e-05,
      "loss": 0.261,
      "step": 17140
    },
    {
      "epoch": 83.75838926174497,
      "grad_norm": 3.2548046112060547,
      "learning_rate": 8.813114754098362e-05,
      "loss": 0.2622,
      "step": 17160
    },
    {
      "epoch": 83.85600976205004,
      "grad_norm": 4.520061492919922,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.272,
      "step": 17180
    },
    {
      "epoch": 83.95363026235509,
      "grad_norm": 3.1347196102142334,
      "learning_rate": 8.786885245901639e-05,
      "loss": 0.2504,
      "step": 17200
    },
    {
      "epoch": 84.05125076266016,
      "grad_norm": 3.6954307556152344,
      "learning_rate": 8.773770491803279e-05,
      "loss": 0.2566,
      "step": 17220
    },
    {
      "epoch": 84.14887126296523,
      "grad_norm": 4.145720481872559,
      "learning_rate": 8.760655737704918e-05,
      "loss": 0.2393,
      "step": 17240
    },
    {
      "epoch": 84.24649176327028,
      "grad_norm": 3.575308084487915,
      "learning_rate": 8.747540983606558e-05,
      "loss": 0.2667,
      "step": 17260
    },
    {
      "epoch": 84.34411226357535,
      "grad_norm": 4.094547271728516,
      "learning_rate": 8.734426229508197e-05,
      "loss": 0.244,
      "step": 17280
    },
    {
      "epoch": 84.44173276388041,
      "grad_norm": 3.583008050918579,
      "learning_rate": 8.721311475409837e-05,
      "loss": 0.2262,
      "step": 17300
    },
    {
      "epoch": 84.53935326418548,
      "grad_norm": 4.159909248352051,
      "learning_rate": 8.708196721311475e-05,
      "loss": 0.2625,
      "step": 17320
    },
    {
      "epoch": 84.63697376449055,
      "grad_norm": 3.5979392528533936,
      "learning_rate": 8.695081967213115e-05,
      "loss": 0.2421,
      "step": 17340
    },
    {
      "epoch": 84.7345942647956,
      "grad_norm": 2.0888795852661133,
      "learning_rate": 8.681967213114754e-05,
      "loss": 0.2555,
      "step": 17360
    },
    {
      "epoch": 84.83221476510067,
      "grad_norm": 3.8699913024902344,
      "learning_rate": 8.668852459016393e-05,
      "loss": 0.2733,
      "step": 17380
    },
    {
      "epoch": 84.92983526540573,
      "grad_norm": 3.1710896492004395,
      "learning_rate": 8.655737704918033e-05,
      "loss": 0.2824,
      "step": 17400
    },
    {
      "epoch": 85.0274557657108,
      "grad_norm": 3.3995399475097656,
      "learning_rate": 8.642622950819672e-05,
      "loss": 0.2555,
      "step": 17420
    },
    {
      "epoch": 85.12507626601587,
      "grad_norm": 4.362491130828857,
      "learning_rate": 8.629508196721313e-05,
      "loss": 0.2535,
      "step": 17440
    },
    {
      "epoch": 85.22269676632092,
      "grad_norm": 3.141589641571045,
      "learning_rate": 8.61639344262295e-05,
      "loss": 0.2368,
      "step": 17460
    },
    {
      "epoch": 85.32031726662599,
      "grad_norm": 2.7880699634552,
      "learning_rate": 8.60327868852459e-05,
      "loss": 0.2382,
      "step": 17480
    },
    {
      "epoch": 85.41793776693106,
      "grad_norm": 3.9995205402374268,
      "learning_rate": 8.59016393442623e-05,
      "loss": 0.2484,
      "step": 17500
    },
    {
      "epoch": 85.51555826723612,
      "grad_norm": 4.39112663269043,
      "learning_rate": 8.57704918032787e-05,
      "loss": 0.254,
      "step": 17520
    },
    {
      "epoch": 85.61317876754119,
      "grad_norm": 3.7383909225463867,
      "learning_rate": 8.563934426229509e-05,
      "loss": 0.2685,
      "step": 17540
    },
    {
      "epoch": 85.71079926784624,
      "grad_norm": 2.9596993923187256,
      "learning_rate": 8.550819672131149e-05,
      "loss": 0.2725,
      "step": 17560
    },
    {
      "epoch": 85.80841976815131,
      "grad_norm": 3.4363205432891846,
      "learning_rate": 8.537704918032787e-05,
      "loss": 0.2518,
      "step": 17580
    },
    {
      "epoch": 85.90604026845638,
      "grad_norm": 4.1404709815979,
      "learning_rate": 8.524590163934426e-05,
      "loss": 0.2409,
      "step": 17600
    },
    {
      "epoch": 86.00366076876143,
      "grad_norm": 3.080606460571289,
      "learning_rate": 8.511475409836066e-05,
      "loss": 0.2646,
      "step": 17620
    },
    {
      "epoch": 86.1012812690665,
      "grad_norm": 4.42800760269165,
      "learning_rate": 8.498360655737705e-05,
      "loss": 0.2274,
      "step": 17640
    },
    {
      "epoch": 86.19890176937157,
      "grad_norm": 3.3812572956085205,
      "learning_rate": 8.485245901639345e-05,
      "loss": 0.2165,
      "step": 17660
    },
    {
      "epoch": 86.29652226967663,
      "grad_norm": 3.873788833618164,
      "learning_rate": 8.472131147540984e-05,
      "loss": 0.2464,
      "step": 17680
    },
    {
      "epoch": 86.3941427699817,
      "grad_norm": 4.275656700134277,
      "learning_rate": 8.459016393442624e-05,
      "loss": 0.2236,
      "step": 17700
    },
    {
      "epoch": 86.49176327028675,
      "grad_norm": 3.0242347717285156,
      "learning_rate": 8.445901639344262e-05,
      "loss": 0.2518,
      "step": 17720
    },
    {
      "epoch": 86.58938377059182,
      "grad_norm": 3.052992105484009,
      "learning_rate": 8.432786885245902e-05,
      "loss": 0.2659,
      "step": 17740
    },
    {
      "epoch": 86.68700427089689,
      "grad_norm": 5.510742664337158,
      "learning_rate": 8.419672131147541e-05,
      "loss": 0.269,
      "step": 17760
    },
    {
      "epoch": 86.78462477120195,
      "grad_norm": 3.8788983821868896,
      "learning_rate": 8.406557377049181e-05,
      "loss": 0.2548,
      "step": 17780
    },
    {
      "epoch": 86.88224527150702,
      "grad_norm": 4.494417190551758,
      "learning_rate": 8.39344262295082e-05,
      "loss": 0.2558,
      "step": 17800
    },
    {
      "epoch": 86.97986577181209,
      "grad_norm": 4.6547040939331055,
      "learning_rate": 8.380327868852459e-05,
      "loss": 0.2775,
      "step": 17820
    },
    {
      "epoch": 87.07748627211714,
      "grad_norm": 3.511051654815674,
      "learning_rate": 8.367213114754098e-05,
      "loss": 0.2233,
      "step": 17840
    },
    {
      "epoch": 87.17510677242221,
      "grad_norm": 4.116011142730713,
      "learning_rate": 8.354098360655737e-05,
      "loss": 0.2246,
      "step": 17860
    },
    {
      "epoch": 87.27272727272727,
      "grad_norm": 3.525118589401245,
      "learning_rate": 8.340983606557377e-05,
      "loss": 0.2058,
      "step": 17880
    },
    {
      "epoch": 87.37034777303234,
      "grad_norm": 3.5380094051361084,
      "learning_rate": 8.327868852459016e-05,
      "loss": 0.2467,
      "step": 17900
    },
    {
      "epoch": 87.4679682733374,
      "grad_norm": 3.6875052452087402,
      "learning_rate": 8.314754098360657e-05,
      "loss": 0.2967,
      "step": 17920
    },
    {
      "epoch": 87.56558877364246,
      "grad_norm": 3.565765142440796,
      "learning_rate": 8.301639344262296e-05,
      "loss": 0.2434,
      "step": 17940
    },
    {
      "epoch": 87.66320927394753,
      "grad_norm": 3.3109848499298096,
      "learning_rate": 8.288524590163935e-05,
      "loss": 0.2538,
      "step": 17960
    },
    {
      "epoch": 87.7608297742526,
      "grad_norm": 3.278052568435669,
      "learning_rate": 8.275409836065573e-05,
      "loss": 0.2739,
      "step": 17980
    },
    {
      "epoch": 87.85845027455765,
      "grad_norm": 5.82271671295166,
      "learning_rate": 8.262295081967214e-05,
      "loss": 0.244,
      "step": 18000
    },
    {
      "epoch": 87.95607077486272,
      "grad_norm": 4.266513824462891,
      "learning_rate": 8.249180327868853e-05,
      "loss": 0.2546,
      "step": 18020
    },
    {
      "epoch": 88.05369127516778,
      "grad_norm": 3.985555648803711,
      "learning_rate": 8.236065573770492e-05,
      "loss": 0.2673,
      "step": 18040
    },
    {
      "epoch": 88.15131177547285,
      "grad_norm": 3.4252424240112305,
      "learning_rate": 8.222950819672132e-05,
      "loss": 0.2308,
      "step": 18060
    },
    {
      "epoch": 88.24893227577792,
      "grad_norm": 3.1381208896636963,
      "learning_rate": 8.209836065573771e-05,
      "loss": 0.24,
      "step": 18080
    },
    {
      "epoch": 88.34655277608297,
      "grad_norm": 4.053617000579834,
      "learning_rate": 8.19672131147541e-05,
      "loss": 0.2646,
      "step": 18100
    },
    {
      "epoch": 88.44417327638804,
      "grad_norm": 3.525423288345337,
      "learning_rate": 8.183606557377049e-05,
      "loss": 0.2535,
      "step": 18120
    },
    {
      "epoch": 88.54179377669311,
      "grad_norm": 3.593766212463379,
      "learning_rate": 8.170491803278689e-05,
      "loss": 0.2426,
      "step": 18140
    },
    {
      "epoch": 88.63941427699817,
      "grad_norm": 4.087725639343262,
      "learning_rate": 8.157377049180328e-05,
      "loss": 0.2196,
      "step": 18160
    },
    {
      "epoch": 88.73703477730324,
      "grad_norm": 4.209127902984619,
      "learning_rate": 8.144262295081968e-05,
      "loss": 0.2593,
      "step": 18180
    },
    {
      "epoch": 88.83465527760829,
      "grad_norm": 3.9146687984466553,
      "learning_rate": 8.131147540983607e-05,
      "loss": 0.2492,
      "step": 18200
    },
    {
      "epoch": 88.93227577791336,
      "grad_norm": 3.4198620319366455,
      "learning_rate": 8.118032786885246e-05,
      "loss": 0.2569,
      "step": 18220
    },
    {
      "epoch": 89.02989627821843,
      "grad_norm": 3.5384342670440674,
      "learning_rate": 8.104918032786885e-05,
      "loss": 0.2298,
      "step": 18240
    },
    {
      "epoch": 89.12751677852349,
      "grad_norm": 3.252002239227295,
      "learning_rate": 8.091803278688524e-05,
      "loss": 0.2296,
      "step": 18260
    },
    {
      "epoch": 89.22513727882856,
      "grad_norm": 3.5200119018554688,
      "learning_rate": 8.078688524590164e-05,
      "loss": 0.2476,
      "step": 18280
    },
    {
      "epoch": 89.32275777913362,
      "grad_norm": 4.229335784912109,
      "learning_rate": 8.065573770491803e-05,
      "loss": 0.2427,
      "step": 18300
    },
    {
      "epoch": 89.42037827943868,
      "grad_norm": 3.418846368789673,
      "learning_rate": 8.052459016393444e-05,
      "loss": 0.248,
      "step": 18320
    },
    {
      "epoch": 89.51799877974375,
      "grad_norm": 4.674842357635498,
      "learning_rate": 8.039344262295082e-05,
      "loss": 0.2388,
      "step": 18340
    },
    {
      "epoch": 89.6156192800488,
      "grad_norm": 2.8320224285125732,
      "learning_rate": 8.026229508196721e-05,
      "loss": 0.2737,
      "step": 18360
    },
    {
      "epoch": 89.71323978035387,
      "grad_norm": 3.764437198638916,
      "learning_rate": 8.01311475409836e-05,
      "loss": 0.2217,
      "step": 18380
    },
    {
      "epoch": 89.81086028065894,
      "grad_norm": 3.7620530128479004,
      "learning_rate": 8e-05,
      "loss": 0.2339,
      "step": 18400
    },
    {
      "epoch": 89.908480780964,
      "grad_norm": 3.451263904571533,
      "learning_rate": 7.98688524590164e-05,
      "loss": 0.239,
      "step": 18420
    },
    {
      "epoch": 90.00610128126907,
      "grad_norm": 3.327611207962036,
      "learning_rate": 7.97377049180328e-05,
      "loss": 0.2509,
      "step": 18440
    },
    {
      "epoch": 90.10372178157414,
      "grad_norm": 3.0182056427001953,
      "learning_rate": 7.960655737704919e-05,
      "loss": 0.2173,
      "step": 18460
    },
    {
      "epoch": 90.20134228187919,
      "grad_norm": 4.298709869384766,
      "learning_rate": 7.947540983606558e-05,
      "loss": 0.2371,
      "step": 18480
    },
    {
      "epoch": 90.29896278218426,
      "grad_norm": 6.2249979972839355,
      "learning_rate": 7.934426229508197e-05,
      "loss": 0.2504,
      "step": 18500
    },
    {
      "epoch": 90.39658328248932,
      "grad_norm": 3.3702869415283203,
      "learning_rate": 7.921311475409836e-05,
      "loss": 0.2264,
      "step": 18520
    },
    {
      "epoch": 90.49420378279439,
      "grad_norm": 4.575166702270508,
      "learning_rate": 7.908196721311476e-05,
      "loss": 0.2296,
      "step": 18540
    },
    {
      "epoch": 90.59182428309946,
      "grad_norm": 3.4049549102783203,
      "learning_rate": 7.895081967213115e-05,
      "loss": 0.2488,
      "step": 18560
    },
    {
      "epoch": 90.68944478340451,
      "grad_norm": 3.5313448905944824,
      "learning_rate": 7.881967213114755e-05,
      "loss": 0.2414,
      "step": 18580
    },
    {
      "epoch": 90.78706528370958,
      "grad_norm": 3.3291146755218506,
      "learning_rate": 7.868852459016394e-05,
      "loss": 0.225,
      "step": 18600
    },
    {
      "epoch": 90.88468578401465,
      "grad_norm": 2.99589204788208,
      "learning_rate": 7.855737704918033e-05,
      "loss": 0.2548,
      "step": 18620
    },
    {
      "epoch": 90.9823062843197,
      "grad_norm": 3.2335972785949707,
      "learning_rate": 7.842622950819672e-05,
      "loss": 0.269,
      "step": 18640
    },
    {
      "epoch": 91.07992678462477,
      "grad_norm": 4.912237167358398,
      "learning_rate": 7.829508196721311e-05,
      "loss": 0.2392,
      "step": 18660
    },
    {
      "epoch": 91.17754728492983,
      "grad_norm": 4.516569137573242,
      "learning_rate": 7.816393442622951e-05,
      "loss": 0.2213,
      "step": 18680
    },
    {
      "epoch": 91.2751677852349,
      "grad_norm": 3.225470542907715,
      "learning_rate": 7.80327868852459e-05,
      "loss": 0.2414,
      "step": 18700
    },
    {
      "epoch": 91.37278828553997,
      "grad_norm": 3.231811761856079,
      "learning_rate": 7.79016393442623e-05,
      "loss": 0.2384,
      "step": 18720
    },
    {
      "epoch": 91.47040878584502,
      "grad_norm": 3.199504852294922,
      "learning_rate": 7.77704918032787e-05,
      "loss": 0.2336,
      "step": 18740
    },
    {
      "epoch": 91.5680292861501,
      "grad_norm": 3.4679319858551025,
      "learning_rate": 7.763934426229508e-05,
      "loss": 0.2136,
      "step": 18760
    },
    {
      "epoch": 91.66564978645516,
      "grad_norm": 4.474179267883301,
      "learning_rate": 7.750819672131147e-05,
      "loss": 0.2454,
      "step": 18780
    },
    {
      "epoch": 91.76327028676022,
      "grad_norm": 4.39286994934082,
      "learning_rate": 7.737704918032788e-05,
      "loss": 0.2377,
      "step": 18800
    },
    {
      "epoch": 91.86089078706529,
      "grad_norm": 2.924795627593994,
      "learning_rate": 7.724590163934426e-05,
      "loss": 0.2475,
      "step": 18820
    },
    {
      "epoch": 91.95851128737034,
      "grad_norm": 3.4778449535369873,
      "learning_rate": 7.711475409836067e-05,
      "loss": 0.2712,
      "step": 18840
    },
    {
      "epoch": 92.05613178767541,
      "grad_norm": 5.152170181274414,
      "learning_rate": 7.698360655737706e-05,
      "loss": 0.2336,
      "step": 18860
    },
    {
      "epoch": 92.15375228798048,
      "grad_norm": 2.8238580226898193,
      "learning_rate": 7.685245901639345e-05,
      "loss": 0.2504,
      "step": 18880
    },
    {
      "epoch": 92.25137278828554,
      "grad_norm": 3.870866537094116,
      "learning_rate": 7.672131147540984e-05,
      "loss": 0.25,
      "step": 18900
    },
    {
      "epoch": 92.3489932885906,
      "grad_norm": 5.052824020385742,
      "learning_rate": 7.659016393442622e-05,
      "loss": 0.2517,
      "step": 18920
    },
    {
      "epoch": 92.44661378889568,
      "grad_norm": 3.048907518386841,
      "learning_rate": 7.645901639344263e-05,
      "loss": 0.2347,
      "step": 18940
    },
    {
      "epoch": 92.54423428920073,
      "grad_norm": 3.2473690509796143,
      "learning_rate": 7.632786885245902e-05,
      "loss": 0.2362,
      "step": 18960
    },
    {
      "epoch": 92.6418547895058,
      "grad_norm": 4.635523319244385,
      "learning_rate": 7.619672131147542e-05,
      "loss": 0.2274,
      "step": 18980
    },
    {
      "epoch": 92.73947528981085,
      "grad_norm": 2.6211893558502197,
      "learning_rate": 7.606557377049181e-05,
      "loss": 0.2258,
      "step": 19000
    },
    {
      "epoch": 92.83709579011592,
      "grad_norm": 3.725900888442993,
      "learning_rate": 7.59344262295082e-05,
      "loss": 0.2407,
      "step": 19020
    },
    {
      "epoch": 92.934716290421,
      "grad_norm": 4.358538627624512,
      "learning_rate": 7.580327868852459e-05,
      "loss": 0.2268,
      "step": 19040
    },
    {
      "epoch": 93.03233679072605,
      "grad_norm": 3.608492851257324,
      "learning_rate": 7.567213114754099e-05,
      "loss": 0.2114,
      "step": 19060
    },
    {
      "epoch": 93.12995729103112,
      "grad_norm": 3.455113410949707,
      "learning_rate": 7.554098360655738e-05,
      "loss": 0.2194,
      "step": 19080
    },
    {
      "epoch": 93.22757779133617,
      "grad_norm": 3.443164825439453,
      "learning_rate": 7.540983606557377e-05,
      "loss": 0.2305,
      "step": 19100
    },
    {
      "epoch": 93.32519829164124,
      "grad_norm": 3.579775810241699,
      "learning_rate": 7.527868852459017e-05,
      "loss": 0.2358,
      "step": 19120
    },
    {
      "epoch": 93.42281879194631,
      "grad_norm": 2.8307318687438965,
      "learning_rate": 7.514754098360656e-05,
      "loss": 0.2289,
      "step": 19140
    },
    {
      "epoch": 93.52043929225137,
      "grad_norm": 4.2809834480285645,
      "learning_rate": 7.501639344262295e-05,
      "loss": 0.2246,
      "step": 19160
    },
    {
      "epoch": 93.61805979255644,
      "grad_norm": 4.093430042266846,
      "learning_rate": 7.488524590163934e-05,
      "loss": 0.2416,
      "step": 19180
    },
    {
      "epoch": 93.7156802928615,
      "grad_norm": 3.562998056411743,
      "learning_rate": 7.475409836065574e-05,
      "loss": 0.2453,
      "step": 19200
    },
    {
      "epoch": 93.81330079316656,
      "grad_norm": 3.611504554748535,
      "learning_rate": 7.462295081967213e-05,
      "loss": 0.2389,
      "step": 19220
    },
    {
      "epoch": 93.91092129347163,
      "grad_norm": 3.638408899307251,
      "learning_rate": 7.449180327868854e-05,
      "loss": 0.2354,
      "step": 19240
    },
    {
      "epoch": 94.00854179377669,
      "grad_norm": 2.646662473678589,
      "learning_rate": 7.436065573770493e-05,
      "loss": 0.2588,
      "step": 19260
    },
    {
      "epoch": 94.10616229408176,
      "grad_norm": 2.8512191772460938,
      "learning_rate": 7.422950819672131e-05,
      "loss": 0.2219,
      "step": 19280
    },
    {
      "epoch": 94.20378279438683,
      "grad_norm": 4.569155693054199,
      "learning_rate": 7.40983606557377e-05,
      "loss": 0.2151,
      "step": 19300
    },
    {
      "epoch": 94.30140329469188,
      "grad_norm": 3.036120653152466,
      "learning_rate": 7.39672131147541e-05,
      "loss": 0.2133,
      "step": 19320
    },
    {
      "epoch": 94.39902379499695,
      "grad_norm": 3.7911667823791504,
      "learning_rate": 7.38360655737705e-05,
      "loss": 0.206,
      "step": 19340
    },
    {
      "epoch": 94.49664429530202,
      "grad_norm": 3.4659411907196045,
      "learning_rate": 7.370491803278689e-05,
      "loss": 0.2175,
      "step": 19360
    },
    {
      "epoch": 94.59426479560707,
      "grad_norm": 3.7196578979492188,
      "learning_rate": 7.357377049180329e-05,
      "loss": 0.2508,
      "step": 19380
    },
    {
      "epoch": 94.69188529591214,
      "grad_norm": 3.499638557434082,
      "learning_rate": 7.344262295081968e-05,
      "loss": 0.2595,
      "step": 19400
    },
    {
      "epoch": 94.7895057962172,
      "grad_norm": 3.3197853565216064,
      "learning_rate": 7.331147540983607e-05,
      "loss": 0.2498,
      "step": 19420
    },
    {
      "epoch": 94.88712629652227,
      "grad_norm": 3.819153308868408,
      "learning_rate": 7.318032786885246e-05,
      "loss": 0.24,
      "step": 19440
    },
    {
      "epoch": 94.98474679682734,
      "grad_norm": 3.229252815246582,
      "learning_rate": 7.304918032786886e-05,
      "loss": 0.2402,
      "step": 19460
    },
    {
      "epoch": 95.0823672971324,
      "grad_norm": 3.8045654296875,
      "learning_rate": 7.291803278688525e-05,
      "loss": 0.2235,
      "step": 19480
    },
    {
      "epoch": 95.17998779743746,
      "grad_norm": 3.8064322471618652,
      "learning_rate": 7.278688524590165e-05,
      "loss": 0.2374,
      "step": 19500
    },
    {
      "epoch": 95.27760829774253,
      "grad_norm": 2.9170637130737305,
      "learning_rate": 7.265573770491804e-05,
      "loss": 0.223,
      "step": 19520
    },
    {
      "epoch": 95.37522879804759,
      "grad_norm": 3.6795055866241455,
      "learning_rate": 7.252459016393443e-05,
      "loss": 0.2256,
      "step": 19540
    },
    {
      "epoch": 95.47284929835266,
      "grad_norm": 3.1766109466552734,
      "learning_rate": 7.239344262295082e-05,
      "loss": 0.2536,
      "step": 19560
    },
    {
      "epoch": 95.57046979865771,
      "grad_norm": 3.862264633178711,
      "learning_rate": 7.226229508196721e-05,
      "loss": 0.2339,
      "step": 19580
    },
    {
      "epoch": 95.66809029896278,
      "grad_norm": 4.323207855224609,
      "learning_rate": 7.213114754098361e-05,
      "loss": 0.2434,
      "step": 19600
    },
    {
      "epoch": 95.76571079926785,
      "grad_norm": 2.973966598510742,
      "learning_rate": 7.2e-05,
      "loss": 0.219,
      "step": 19620
    },
    {
      "epoch": 95.8633312995729,
      "grad_norm": 3.5967094898223877,
      "learning_rate": 7.18688524590164e-05,
      "loss": 0.243,
      "step": 19640
    },
    {
      "epoch": 95.96095179987798,
      "grad_norm": 3.993166923522949,
      "learning_rate": 7.17377049180328e-05,
      "loss": 0.2263,
      "step": 19660
    },
    {
      "epoch": 96.05857230018304,
      "grad_norm": 3.425034284591675,
      "learning_rate": 7.160655737704918e-05,
      "loss": 0.2158,
      "step": 19680
    },
    {
      "epoch": 96.1561928004881,
      "grad_norm": 3.4514410495758057,
      "learning_rate": 7.147540983606557e-05,
      "loss": 0.1976,
      "step": 19700
    },
    {
      "epoch": 96.25381330079317,
      "grad_norm": 5.355749607086182,
      "learning_rate": 7.134426229508198e-05,
      "loss": 0.221,
      "step": 19720
    },
    {
      "epoch": 96.35143380109822,
      "grad_norm": 3.296389102935791,
      "learning_rate": 7.121311475409837e-05,
      "loss": 0.2606,
      "step": 19740
    },
    {
      "epoch": 96.4490543014033,
      "grad_norm": 3.1039505004882812,
      "learning_rate": 7.108196721311475e-05,
      "loss": 0.2532,
      "step": 19760
    },
    {
      "epoch": 96.54667480170836,
      "grad_norm": 3.3837499618530273,
      "learning_rate": 7.095081967213116e-05,
      "loss": 0.2351,
      "step": 19780
    },
    {
      "epoch": 96.64429530201342,
      "grad_norm": 4.072084426879883,
      "learning_rate": 7.081967213114755e-05,
      "loss": 0.2214,
      "step": 19800
    },
    {
      "epoch": 96.74191580231849,
      "grad_norm": 3.669661283493042,
      "learning_rate": 7.068852459016394e-05,
      "loss": 0.2168,
      "step": 19820
    },
    {
      "epoch": 96.83953630262356,
      "grad_norm": 3.449476957321167,
      "learning_rate": 7.055737704918033e-05,
      "loss": 0.2256,
      "step": 19840
    },
    {
      "epoch": 96.93715680292861,
      "grad_norm": 3.7562010288238525,
      "learning_rate": 7.042622950819673e-05,
      "loss": 0.233,
      "step": 19860
    },
    {
      "epoch": 97.03477730323368,
      "grad_norm": 3.0482330322265625,
      "learning_rate": 7.029508196721312e-05,
      "loss": 0.2388,
      "step": 19880
    },
    {
      "epoch": 97.13239780353874,
      "grad_norm": 2.8242592811584473,
      "learning_rate": 7.016393442622952e-05,
      "loss": 0.247,
      "step": 19900
    },
    {
      "epoch": 97.2300183038438,
      "grad_norm": 3.9347472190856934,
      "learning_rate": 7.003278688524591e-05,
      "loss": 0.214,
      "step": 19920
    },
    {
      "epoch": 97.32763880414888,
      "grad_norm": 3.7674787044525146,
      "learning_rate": 6.99016393442623e-05,
      "loss": 0.2321,
      "step": 19940
    },
    {
      "epoch": 97.42525930445393,
      "grad_norm": 3.0752851963043213,
      "learning_rate": 6.977049180327869e-05,
      "loss": 0.228,
      "step": 19960
    },
    {
      "epoch": 97.522879804759,
      "grad_norm": 4.3538899421691895,
      "learning_rate": 6.963934426229508e-05,
      "loss": 0.2196,
      "step": 19980
    },
    {
      "epoch": 97.62050030506407,
      "grad_norm": 2.677072763442993,
      "learning_rate": 6.950819672131148e-05,
      "loss": 0.2519,
      "step": 20000
    },
    {
      "epoch": 97.71812080536913,
      "grad_norm": 3.771855592727661,
      "learning_rate": 6.937704918032787e-05,
      "loss": 0.1967,
      "step": 20020
    },
    {
      "epoch": 97.8157413056742,
      "grad_norm": 4.431488990783691,
      "learning_rate": 6.924590163934427e-05,
      "loss": 0.2112,
      "step": 20040
    },
    {
      "epoch": 97.91336180597925,
      "grad_norm": 3.970080852508545,
      "learning_rate": 6.911475409836066e-05,
      "loss": 0.2295,
      "step": 20060
    },
    {
      "epoch": 98.01098230628432,
      "grad_norm": 2.521176338195801,
      "learning_rate": 6.898360655737705e-05,
      "loss": 0.2495,
      "step": 20080
    },
    {
      "epoch": 98.10860280658939,
      "grad_norm": 3.8167226314544678,
      "learning_rate": 6.885245901639344e-05,
      "loss": 0.2123,
      "step": 20100
    },
    {
      "epoch": 98.20622330689444,
      "grad_norm": 3.245234727859497,
      "learning_rate": 6.872131147540984e-05,
      "loss": 0.229,
      "step": 20120
    },
    {
      "epoch": 98.30384380719951,
      "grad_norm": 3.270099401473999,
      "learning_rate": 6.859016393442623e-05,
      "loss": 0.1986,
      "step": 20140
    },
    {
      "epoch": 98.40146430750458,
      "grad_norm": 3.133777379989624,
      "learning_rate": 6.845901639344262e-05,
      "loss": 0.2407,
      "step": 20160
    },
    {
      "epoch": 98.49908480780964,
      "grad_norm": 4.500607490539551,
      "learning_rate": 6.832786885245903e-05,
      "loss": 0.223,
      "step": 20180
    },
    {
      "epoch": 98.59670530811471,
      "grad_norm": 3.4543895721435547,
      "learning_rate": 6.819672131147542e-05,
      "loss": 0.2373,
      "step": 20200
    },
    {
      "epoch": 98.69432580841976,
      "grad_norm": 3.2081384658813477,
      "learning_rate": 6.80655737704918e-05,
      "loss": 0.2262,
      "step": 20220
    },
    {
      "epoch": 98.79194630872483,
      "grad_norm": 4.9004926681518555,
      "learning_rate": 6.79344262295082e-05,
      "loss": 0.2335,
      "step": 20240
    },
    {
      "epoch": 98.8895668090299,
      "grad_norm": 3.9303877353668213,
      "learning_rate": 6.78032786885246e-05,
      "loss": 0.2061,
      "step": 20260
    },
    {
      "epoch": 98.98718730933496,
      "grad_norm": 3.722957134246826,
      "learning_rate": 6.767213114754099e-05,
      "loss": 0.2518,
      "step": 20280
    },
    {
      "epoch": 99.08480780964003,
      "grad_norm": 4.127925872802734,
      "learning_rate": 6.754098360655739e-05,
      "loss": 0.2133,
      "step": 20300
    },
    {
      "epoch": 99.1824283099451,
      "grad_norm": 4.3639726638793945,
      "learning_rate": 6.740983606557378e-05,
      "loss": 0.2119,
      "step": 20320
    },
    {
      "epoch": 99.28004881025015,
      "grad_norm": 3.288351058959961,
      "learning_rate": 6.727868852459017e-05,
      "loss": 0.2307,
      "step": 20340
    },
    {
      "epoch": 99.37766931055522,
      "grad_norm": 3.5033068656921387,
      "learning_rate": 6.714754098360656e-05,
      "loss": 0.2227,
      "step": 20360
    },
    {
      "epoch": 99.47528981086027,
      "grad_norm": 3.1117262840270996,
      "learning_rate": 6.701639344262295e-05,
      "loss": 0.2257,
      "step": 20380
    },
    {
      "epoch": 99.57291031116534,
      "grad_norm": 4.022121906280518,
      "learning_rate": 6.688524590163935e-05,
      "loss": 0.2336,
      "step": 20400
    },
    {
      "epoch": 99.67053081147041,
      "grad_norm": 3.4611172676086426,
      "learning_rate": 6.675409836065574e-05,
      "loss": 0.2192,
      "step": 20420
    },
    {
      "epoch": 99.76815131177547,
      "grad_norm": 3.3648264408111572,
      "learning_rate": 6.662295081967214e-05,
      "loss": 0.2482,
      "step": 20440
    },
    {
      "epoch": 99.86577181208054,
      "grad_norm": 3.8535125255584717,
      "learning_rate": 6.649180327868853e-05,
      "loss": 0.2314,
      "step": 20460
    },
    {
      "epoch": 99.96339231238561,
      "grad_norm": 3.8780901432037354,
      "learning_rate": 6.636065573770492e-05,
      "loss": 0.2203,
      "step": 20480
    },
    {
      "epoch": 100.06101281269066,
      "grad_norm": 3.479278326034546,
      "learning_rate": 6.622950819672131e-05,
      "loss": 0.2003,
      "step": 20500
    },
    {
      "epoch": 100.15863331299573,
      "grad_norm": 3.530697822570801,
      "learning_rate": 6.609836065573771e-05,
      "loss": 0.2084,
      "step": 20520
    },
    {
      "epoch": 100.25625381330079,
      "grad_norm": 3.3677327632904053,
      "learning_rate": 6.59672131147541e-05,
      "loss": 0.2394,
      "step": 20540
    },
    {
      "epoch": 100.35387431360586,
      "grad_norm": 2.9369733333587646,
      "learning_rate": 6.58360655737705e-05,
      "loss": 0.2553,
      "step": 20560
    },
    {
      "epoch": 100.45149481391093,
      "grad_norm": 3.9915401935577393,
      "learning_rate": 6.57049180327869e-05,
      "loss": 0.2393,
      "step": 20580
    },
    {
      "epoch": 100.54911531421598,
      "grad_norm": 3.6804075241088867,
      "learning_rate": 6.557377049180327e-05,
      "loss": 0.2266,
      "step": 20600
    },
    {
      "epoch": 100.64673581452105,
      "grad_norm": 4.8830156326293945,
      "learning_rate": 6.544262295081967e-05,
      "loss": 0.19,
      "step": 20620
    },
    {
      "epoch": 100.74435631482612,
      "grad_norm": 4.2519731521606445,
      "learning_rate": 6.531147540983606e-05,
      "loss": 0.2416,
      "step": 20640
    },
    {
      "epoch": 100.84197681513118,
      "grad_norm": 3.413914203643799,
      "learning_rate": 6.518032786885247e-05,
      "loss": 0.2026,
      "step": 20660
    },
    {
      "epoch": 100.93959731543625,
      "grad_norm": 3.642609119415283,
      "learning_rate": 6.504918032786886e-05,
      "loss": 0.2227,
      "step": 20680
    },
    {
      "epoch": 101.0372178157413,
      "grad_norm": 4.770230770111084,
      "learning_rate": 6.491803278688526e-05,
      "loss": 0.2277,
      "step": 20700
    },
    {
      "epoch": 101.13483831604637,
      "grad_norm": 4.115867614746094,
      "learning_rate": 6.478688524590165e-05,
      "loss": 0.205,
      "step": 20720
    },
    {
      "epoch": 101.23245881635144,
      "grad_norm": 4.3617048263549805,
      "learning_rate": 6.465573770491804e-05,
      "loss": 0.2048,
      "step": 20740
    },
    {
      "epoch": 101.3300793166565,
      "grad_norm": 3.818500518798828,
      "learning_rate": 6.452459016393443e-05,
      "loss": 0.2245,
      "step": 20760
    },
    {
      "epoch": 101.42769981696156,
      "grad_norm": 3.3728935718536377,
      "learning_rate": 6.439344262295083e-05,
      "loss": 0.2342,
      "step": 20780
    },
    {
      "epoch": 101.52532031726662,
      "grad_norm": 3.2103302478790283,
      "learning_rate": 6.426229508196722e-05,
      "loss": 0.2024,
      "step": 20800
    },
    {
      "epoch": 101.62294081757169,
      "grad_norm": 3.5993459224700928,
      "learning_rate": 6.413114754098361e-05,
      "loss": 0.2272,
      "step": 20820
    },
    {
      "epoch": 101.72056131787676,
      "grad_norm": 3.478533983230591,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.236,
      "step": 20840
    },
    {
      "epoch": 101.81818181818181,
      "grad_norm": 4.245722770690918,
      "learning_rate": 6.386885245901639e-05,
      "loss": 0.2246,
      "step": 20860
    },
    {
      "epoch": 101.91580231848688,
      "grad_norm": 3.5051097869873047,
      "learning_rate": 6.373770491803279e-05,
      "loss": 0.2194,
      "step": 20880
    },
    {
      "epoch": 102.01342281879195,
      "grad_norm": 2.882301092147827,
      "learning_rate": 6.360655737704918e-05,
      "loss": 0.2196,
      "step": 20900
    },
    {
      "epoch": 102.111043319097,
      "grad_norm": 3.080702066421509,
      "learning_rate": 6.347540983606558e-05,
      "loss": 0.2109,
      "step": 20920
    },
    {
      "epoch": 102.20866381940208,
      "grad_norm": 2.9892094135284424,
      "learning_rate": 6.334426229508197e-05,
      "loss": 0.2086,
      "step": 20940
    },
    {
      "epoch": 102.30628431970713,
      "grad_norm": 3.401524782180786,
      "learning_rate": 6.321311475409837e-05,
      "loss": 0.202,
      "step": 20960
    },
    {
      "epoch": 102.4039048200122,
      "grad_norm": 3.0444400310516357,
      "learning_rate": 6.308196721311475e-05,
      "loss": 0.2071,
      "step": 20980
    },
    {
      "epoch": 102.50152532031727,
      "grad_norm": 3.028918743133545,
      "learning_rate": 6.295081967213115e-05,
      "loss": 0.226,
      "step": 21000
    },
    {
      "epoch": 102.59914582062233,
      "grad_norm": 3.2344133853912354,
      "learning_rate": 6.281967213114754e-05,
      "loss": 0.2417,
      "step": 21020
    },
    {
      "epoch": 102.6967663209274,
      "grad_norm": 3.1719000339508057,
      "learning_rate": 6.268852459016393e-05,
      "loss": 0.2363,
      "step": 21040
    },
    {
      "epoch": 102.79438682123246,
      "grad_norm": 3.474695920944214,
      "learning_rate": 6.255737704918033e-05,
      "loss": 0.2278,
      "step": 21060
    },
    {
      "epoch": 102.89200732153752,
      "grad_norm": 4.550293445587158,
      "learning_rate": 6.242622950819672e-05,
      "loss": 0.2087,
      "step": 21080
    },
    {
      "epoch": 102.98962782184259,
      "grad_norm": 3.7696001529693604,
      "learning_rate": 6.229508196721313e-05,
      "loss": 0.2305,
      "step": 21100
    },
    {
      "epoch": 103.08724832214764,
      "grad_norm": 3.388397216796875,
      "learning_rate": 6.21639344262295e-05,
      "loss": 0.2133,
      "step": 21120
    },
    {
      "epoch": 103.18486882245271,
      "grad_norm": 4.243392467498779,
      "learning_rate": 6.20327868852459e-05,
      "loss": 0.213,
      "step": 21140
    },
    {
      "epoch": 103.28248932275778,
      "grad_norm": 2.646786689758301,
      "learning_rate": 6.19016393442623e-05,
      "loss": 0.2132,
      "step": 21160
    },
    {
      "epoch": 103.38010982306284,
      "grad_norm": 3.157848596572876,
      "learning_rate": 6.17704918032787e-05,
      "loss": 0.1941,
      "step": 21180
    },
    {
      "epoch": 103.47773032336791,
      "grad_norm": 3.383357286453247,
      "learning_rate": 6.163934426229509e-05,
      "loss": 0.2211,
      "step": 21200
    },
    {
      "epoch": 103.57535082367298,
      "grad_norm": 2.940901279449463,
      "learning_rate": 6.150819672131148e-05,
      "loss": 0.1934,
      "step": 21220
    },
    {
      "epoch": 103.67297132397803,
      "grad_norm": 4.2973103523254395,
      "learning_rate": 6.137704918032787e-05,
      "loss": 0.2516,
      "step": 21240
    },
    {
      "epoch": 103.7705918242831,
      "grad_norm": 3.586219310760498,
      "learning_rate": 6.124590163934426e-05,
      "loss": 0.262,
      "step": 21260
    },
    {
      "epoch": 103.86821232458816,
      "grad_norm": 3.7726240158081055,
      "learning_rate": 6.111475409836066e-05,
      "loss": 0.1876,
      "step": 21280
    },
    {
      "epoch": 103.96583282489323,
      "grad_norm": 3.5038113594055176,
      "learning_rate": 6.098360655737705e-05,
      "loss": 0.2156,
      "step": 21300
    },
    {
      "epoch": 104.0634533251983,
      "grad_norm": 3.6055893898010254,
      "learning_rate": 6.085245901639345e-05,
      "loss": 0.2386,
      "step": 21320
    },
    {
      "epoch": 104.16107382550335,
      "grad_norm": 4.037646770477295,
      "learning_rate": 6.072131147540984e-05,
      "loss": 0.229,
      "step": 21340
    },
    {
      "epoch": 104.25869432580842,
      "grad_norm": 3.209284543991089,
      "learning_rate": 6.0590163934426236e-05,
      "loss": 0.2204,
      "step": 21360
    },
    {
      "epoch": 104.35631482611349,
      "grad_norm": 3.388456106185913,
      "learning_rate": 6.0459016393442625e-05,
      "loss": 0.2155,
      "step": 21380
    },
    {
      "epoch": 104.45393532641855,
      "grad_norm": 2.9914731979370117,
      "learning_rate": 6.032786885245902e-05,
      "loss": 0.2205,
      "step": 21400
    },
    {
      "epoch": 104.55155582672361,
      "grad_norm": 3.4753572940826416,
      "learning_rate": 6.019672131147541e-05,
      "loss": 0.226,
      "step": 21420
    },
    {
      "epoch": 104.64917632702867,
      "grad_norm": 3.400149345397949,
      "learning_rate": 6.00655737704918e-05,
      "loss": 0.2117,
      "step": 21440
    },
    {
      "epoch": 104.74679682733374,
      "grad_norm": 3.2313406467437744,
      "learning_rate": 5.99344262295082e-05,
      "loss": 0.216,
      "step": 21460
    },
    {
      "epoch": 104.84441732763881,
      "grad_norm": 4.102112770080566,
      "learning_rate": 5.9803278688524586e-05,
      "loss": 0.1962,
      "step": 21480
    },
    {
      "epoch": 104.94203782794386,
      "grad_norm": 3.6872830390930176,
      "learning_rate": 5.967213114754099e-05,
      "loss": 0.2212,
      "step": 21500
    },
    {
      "epoch": 105.03965832824893,
      "grad_norm": 3.1488096714019775,
      "learning_rate": 5.954098360655738e-05,
      "loss": 0.1919,
      "step": 21520
    },
    {
      "epoch": 105.137278828554,
      "grad_norm": 3.4388973712921143,
      "learning_rate": 5.9409836065573774e-05,
      "loss": 0.247,
      "step": 21540
    },
    {
      "epoch": 105.23489932885906,
      "grad_norm": 3.776465892791748,
      "learning_rate": 5.927868852459016e-05,
      "loss": 0.1856,
      "step": 21560
    },
    {
      "epoch": 105.33251982916413,
      "grad_norm": 3.552684783935547,
      "learning_rate": 5.9147540983606566e-05,
      "loss": 0.2042,
      "step": 21580
    },
    {
      "epoch": 105.43014032946918,
      "grad_norm": 2.8810436725616455,
      "learning_rate": 5.9016393442622956e-05,
      "loss": 0.2353,
      "step": 21600
    },
    {
      "epoch": 105.52776082977425,
      "grad_norm": 3.2408454418182373,
      "learning_rate": 5.888524590163935e-05,
      "loss": 0.2148,
      "step": 21620
    },
    {
      "epoch": 105.62538133007932,
      "grad_norm": 3.5531325340270996,
      "learning_rate": 5.875409836065574e-05,
      "loss": 0.2106,
      "step": 21640
    },
    {
      "epoch": 105.72300183038438,
      "grad_norm": 4.188174247741699,
      "learning_rate": 5.862295081967213e-05,
      "loss": 0.2211,
      "step": 21660
    },
    {
      "epoch": 105.82062233068945,
      "grad_norm": 4.232003688812256,
      "learning_rate": 5.849180327868853e-05,
      "loss": 0.2139,
      "step": 21680
    },
    {
      "epoch": 105.91824283099452,
      "grad_norm": 3.7038660049438477,
      "learning_rate": 5.8360655737704916e-05,
      "loss": 0.2048,
      "step": 21700
    },
    {
      "epoch": 106.01586333129957,
      "grad_norm": 2.9898719787597656,
      "learning_rate": 5.822950819672132e-05,
      "loss": 0.2184,
      "step": 21720
    },
    {
      "epoch": 106.11348383160464,
      "grad_norm": 3.482365846633911,
      "learning_rate": 5.80983606557377e-05,
      "loss": 0.2171,
      "step": 21740
    },
    {
      "epoch": 106.2111043319097,
      "grad_norm": 3.3595712184906006,
      "learning_rate": 5.7967213114754104e-05,
      "loss": 0.2041,
      "step": 21760
    },
    {
      "epoch": 106.30872483221476,
      "grad_norm": 3.7470903396606445,
      "learning_rate": 5.7836065573770494e-05,
      "loss": 0.1946,
      "step": 21780
    },
    {
      "epoch": 106.40634533251983,
      "grad_norm": 3.418549060821533,
      "learning_rate": 5.770491803278689e-05,
      "loss": 0.2013,
      "step": 21800
    },
    {
      "epoch": 106.50396583282489,
      "grad_norm": 3.115424394607544,
      "learning_rate": 5.757377049180328e-05,
      "loss": 0.2064,
      "step": 21820
    },
    {
      "epoch": 106.60158633312996,
      "grad_norm": 3.2718710899353027,
      "learning_rate": 5.744262295081968e-05,
      "loss": 0.2266,
      "step": 21840
    },
    {
      "epoch": 106.69920683343503,
      "grad_norm": 2.9478771686553955,
      "learning_rate": 5.731147540983607e-05,
      "loss": 0.2321,
      "step": 21860
    },
    {
      "epoch": 106.79682733374008,
      "grad_norm": 3.479456663131714,
      "learning_rate": 5.7180327868852454e-05,
      "loss": 0.2317,
      "step": 21880
    },
    {
      "epoch": 106.89444783404515,
      "grad_norm": 3.934882879257202,
      "learning_rate": 5.704918032786886e-05,
      "loss": 0.2074,
      "step": 21900
    },
    {
      "epoch": 106.99206833435021,
      "grad_norm": 3.40720272064209,
      "learning_rate": 5.6918032786885246e-05,
      "loss": 0.2123,
      "step": 21920
    },
    {
      "epoch": 107.08968883465528,
      "grad_norm": 2.825427770614624,
      "learning_rate": 5.678688524590164e-05,
      "loss": 0.1897,
      "step": 21940
    },
    {
      "epoch": 107.18730933496035,
      "grad_norm": 3.6434717178344727,
      "learning_rate": 5.665573770491803e-05,
      "loss": 0.207,
      "step": 21960
    },
    {
      "epoch": 107.2849298352654,
      "grad_norm": 4.119269847869873,
      "learning_rate": 5.6524590163934435e-05,
      "loss": 0.2086,
      "step": 21980
    },
    {
      "epoch": 107.38255033557047,
      "grad_norm": 2.835963487625122,
      "learning_rate": 5.639344262295082e-05,
      "loss": 0.1942,
      "step": 22000
    },
    {
      "epoch": 107.48017083587554,
      "grad_norm": 3.175858736038208,
      "learning_rate": 5.626229508196722e-05,
      "loss": 0.2089,
      "step": 22020
    },
    {
      "epoch": 107.5777913361806,
      "grad_norm": 3.6265509128570557,
      "learning_rate": 5.613114754098361e-05,
      "loss": 0.2164,
      "step": 22040
    },
    {
      "epoch": 107.67541183648567,
      "grad_norm": 2.90639591217041,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 0.2121,
      "step": 22060
    },
    {
      "epoch": 107.77303233679072,
      "grad_norm": 4.155264377593994,
      "learning_rate": 5.5868852459016395e-05,
      "loss": 0.219,
      "step": 22080
    },
    {
      "epoch": 107.87065283709579,
      "grad_norm": 3.730433940887451,
      "learning_rate": 5.5737704918032785e-05,
      "loss": 0.2369,
      "step": 22100
    },
    {
      "epoch": 107.96827333740086,
      "grad_norm": 2.8690969944000244,
      "learning_rate": 5.560655737704919e-05,
      "loss": 0.2189,
      "step": 22120
    },
    {
      "epoch": 108.06589383770591,
      "grad_norm": 3.061427116394043,
      "learning_rate": 5.547540983606557e-05,
      "loss": 0.2203,
      "step": 22140
    },
    {
      "epoch": 108.16351433801098,
      "grad_norm": 3.6155471801757812,
      "learning_rate": 5.534426229508197e-05,
      "loss": 0.2043,
      "step": 22160
    },
    {
      "epoch": 108.26113483831605,
      "grad_norm": 3.1074283123016357,
      "learning_rate": 5.521311475409836e-05,
      "loss": 0.2095,
      "step": 22180
    },
    {
      "epoch": 108.35875533862111,
      "grad_norm": 3.6168534755706787,
      "learning_rate": 5.508196721311476e-05,
      "loss": 0.2339,
      "step": 22200
    },
    {
      "epoch": 108.45637583892618,
      "grad_norm": 2.9254798889160156,
      "learning_rate": 5.495081967213115e-05,
      "loss": 0.214,
      "step": 22220
    },
    {
      "epoch": 108.55399633923123,
      "grad_norm": 3.2898197174072266,
      "learning_rate": 5.481967213114755e-05,
      "loss": 0.1981,
      "step": 22240
    },
    {
      "epoch": 108.6516168395363,
      "grad_norm": 3.0724892616271973,
      "learning_rate": 5.4688524590163933e-05,
      "loss": 0.1997,
      "step": 22260
    },
    {
      "epoch": 108.74923733984137,
      "grad_norm": 3.3344459533691406,
      "learning_rate": 5.4557377049180336e-05,
      "loss": 0.2243,
      "step": 22280
    },
    {
      "epoch": 108.84685784014643,
      "grad_norm": 4.20386266708374,
      "learning_rate": 5.4426229508196726e-05,
      "loss": 0.1973,
      "step": 22300
    },
    {
      "epoch": 108.9444783404515,
      "grad_norm": 3.0647547245025635,
      "learning_rate": 5.4295081967213115e-05,
      "loss": 0.1984,
      "step": 22320
    },
    {
      "epoch": 109.04209884075657,
      "grad_norm": 3.2792584896087646,
      "learning_rate": 5.416393442622951e-05,
      "loss": 0.2209,
      "step": 22340
    },
    {
      "epoch": 109.13971934106162,
      "grad_norm": 2.900493860244751,
      "learning_rate": 5.40327868852459e-05,
      "loss": 0.2123,
      "step": 22360
    },
    {
      "epoch": 109.23733984136669,
      "grad_norm": 2.754514217376709,
      "learning_rate": 5.3901639344262304e-05,
      "loss": 0.2049,
      "step": 22380
    },
    {
      "epoch": 109.33496034167175,
      "grad_norm": 2.955946445465088,
      "learning_rate": 5.3770491803278686e-05,
      "loss": 0.2164,
      "step": 22400
    },
    {
      "epoch": 109.43258084197682,
      "grad_norm": 3.0447773933410645,
      "learning_rate": 5.363934426229509e-05,
      "loss": 0.1903,
      "step": 22420
    },
    {
      "epoch": 109.53020134228188,
      "grad_norm": 2.8788065910339355,
      "learning_rate": 5.350819672131148e-05,
      "loss": 0.2018,
      "step": 22440
    },
    {
      "epoch": 109.62782184258694,
      "grad_norm": 3.766073703765869,
      "learning_rate": 5.3377049180327875e-05,
      "loss": 0.2092,
      "step": 22460
    },
    {
      "epoch": 109.72544234289201,
      "grad_norm": 4.014832973480225,
      "learning_rate": 5.3245901639344264e-05,
      "loss": 0.2138,
      "step": 22480
    },
    {
      "epoch": 109.82306284319706,
      "grad_norm": 2.987813949584961,
      "learning_rate": 5.311475409836065e-05,
      "loss": 0.2253,
      "step": 22500
    },
    {
      "epoch": 109.92068334350213,
      "grad_norm": 2.980419158935547,
      "learning_rate": 5.298360655737705e-05,
      "loss": 0.2159,
      "step": 22520
    },
    {
      "epoch": 110.0183038438072,
      "grad_norm": 3.1005942821502686,
      "learning_rate": 5.285245901639344e-05,
      "loss": 0.1907,
      "step": 22540
    },
    {
      "epoch": 110.11592434411226,
      "grad_norm": 3.2539725303649902,
      "learning_rate": 5.272131147540984e-05,
      "loss": 0.2023,
      "step": 22560
    },
    {
      "epoch": 110.21354484441733,
      "grad_norm": 3.4975571632385254,
      "learning_rate": 5.259016393442623e-05,
      "loss": 0.2101,
      "step": 22580
    },
    {
      "epoch": 110.3111653447224,
      "grad_norm": 4.542675018310547,
      "learning_rate": 5.245901639344263e-05,
      "loss": 0.2071,
      "step": 22600
    },
    {
      "epoch": 110.40878584502745,
      "grad_norm": 2.5775978565216064,
      "learning_rate": 5.2327868852459017e-05,
      "loss": 0.2027,
      "step": 22620
    },
    {
      "epoch": 110.50640634533252,
      "grad_norm": 2.964486598968506,
      "learning_rate": 5.219672131147541e-05,
      "loss": 0.2281,
      "step": 22640
    },
    {
      "epoch": 110.60402684563758,
      "grad_norm": 3.883513927459717,
      "learning_rate": 5.20655737704918e-05,
      "loss": 0.2083,
      "step": 22660
    },
    {
      "epoch": 110.70164734594265,
      "grad_norm": 3.696744441986084,
      "learning_rate": 5.1934426229508205e-05,
      "loss": 0.2044,
      "step": 22680
    },
    {
      "epoch": 110.79926784624772,
      "grad_norm": 3.104335308074951,
      "learning_rate": 5.1803278688524594e-05,
      "loss": 0.216,
      "step": 22700
    },
    {
      "epoch": 110.89688834655277,
      "grad_norm": 3.6240875720977783,
      "learning_rate": 5.1672131147540984e-05,
      "loss": 0.2086,
      "step": 22720
    },
    {
      "epoch": 110.99450884685784,
      "grad_norm": 4.600063323974609,
      "learning_rate": 5.154098360655738e-05,
      "loss": 0.1983,
      "step": 22740
    },
    {
      "epoch": 111.09212934716291,
      "grad_norm": 4.165120601654053,
      "learning_rate": 5.140983606557377e-05,
      "loss": 0.207,
      "step": 22760
    },
    {
      "epoch": 111.18974984746797,
      "grad_norm": 3.1178033351898193,
      "learning_rate": 5.1278688524590165e-05,
      "loss": 0.1836,
      "step": 22780
    },
    {
      "epoch": 111.28737034777303,
      "grad_norm": 3.8547523021698,
      "learning_rate": 5.1147540983606555e-05,
      "loss": 0.213,
      "step": 22800
    },
    {
      "epoch": 111.38499084807809,
      "grad_norm": 3.626835346221924,
      "learning_rate": 5.101639344262296e-05,
      "loss": 0.1901,
      "step": 22820
    },
    {
      "epoch": 111.48261134838316,
      "grad_norm": 3.89408540725708,
      "learning_rate": 5.088524590163935e-05,
      "loss": 0.2151,
      "step": 22840
    },
    {
      "epoch": 111.58023184868823,
      "grad_norm": 3.3434460163116455,
      "learning_rate": 5.075409836065574e-05,
      "loss": 0.2165,
      "step": 22860
    },
    {
      "epoch": 111.67785234899328,
      "grad_norm": 3.1831305027008057,
      "learning_rate": 5.062295081967213e-05,
      "loss": 0.1911,
      "step": 22880
    },
    {
      "epoch": 111.77547284929835,
      "grad_norm": 3.6153972148895264,
      "learning_rate": 5.049180327868853e-05,
      "loss": 0.2076,
      "step": 22900
    },
    {
      "epoch": 111.87309334960342,
      "grad_norm": 2.87998628616333,
      "learning_rate": 5.036065573770492e-05,
      "loss": 0.2121,
      "step": 22920
    },
    {
      "epoch": 111.97071384990848,
      "grad_norm": 2.9351823329925537,
      "learning_rate": 5.022950819672131e-05,
      "loss": 0.2126,
      "step": 22940
    },
    {
      "epoch": 112.06833435021355,
      "grad_norm": 3.098568916320801,
      "learning_rate": 5.009836065573771e-05,
      "loss": 0.2169,
      "step": 22960
    },
    {
      "epoch": 112.1659548505186,
      "grad_norm": 4.321131229400635,
      "learning_rate": 4.99672131147541e-05,
      "loss": 0.2028,
      "step": 22980
    },
    {
      "epoch": 112.26357535082367,
      "grad_norm": 2.1096384525299072,
      "learning_rate": 4.9836065573770496e-05,
      "loss": 0.2041,
      "step": 23000
    },
    {
      "epoch": 112.36119585112874,
      "grad_norm": 3.537277936935425,
      "learning_rate": 4.970491803278689e-05,
      "loss": 0.1988,
      "step": 23020
    },
    {
      "epoch": 112.4588163514338,
      "grad_norm": 4.58275842666626,
      "learning_rate": 4.957377049180328e-05,
      "loss": 0.1855,
      "step": 23040
    },
    {
      "epoch": 112.55643685173887,
      "grad_norm": 3.290548086166382,
      "learning_rate": 4.944262295081967e-05,
      "loss": 0.2041,
      "step": 23060
    },
    {
      "epoch": 112.65405735204394,
      "grad_norm": 4.11199951171875,
      "learning_rate": 4.931147540983607e-05,
      "loss": 0.2062,
      "step": 23080
    },
    {
      "epoch": 112.75167785234899,
      "grad_norm": 4.122567176818848,
      "learning_rate": 4.918032786885246e-05,
      "loss": 0.2003,
      "step": 23100
    },
    {
      "epoch": 112.84929835265406,
      "grad_norm": 2.775681495666504,
      "learning_rate": 4.904918032786885e-05,
      "loss": 0.2201,
      "step": 23120
    },
    {
      "epoch": 112.94691885295912,
      "grad_norm": 2.8952252864837646,
      "learning_rate": 4.891803278688525e-05,
      "loss": 0.2068,
      "step": 23140
    },
    {
      "epoch": 113.04453935326418,
      "grad_norm": 2.8353431224823,
      "learning_rate": 4.8786885245901645e-05,
      "loss": 0.2087,
      "step": 23160
    },
    {
      "epoch": 113.14215985356925,
      "grad_norm": 4.125601768493652,
      "learning_rate": 4.8655737704918034e-05,
      "loss": 0.1973,
      "step": 23180
    },
    {
      "epoch": 113.23978035387431,
      "grad_norm": 3.5090670585632324,
      "learning_rate": 4.852459016393443e-05,
      "loss": 0.1847,
      "step": 23200
    },
    {
      "epoch": 113.33740085417938,
      "grad_norm": 2.688890218734741,
      "learning_rate": 4.8393442622950826e-05,
      "loss": 0.1842,
      "step": 23220
    },
    {
      "epoch": 113.43502135448445,
      "grad_norm": 4.937670707702637,
      "learning_rate": 4.8262295081967216e-05,
      "loss": 0.214,
      "step": 23240
    },
    {
      "epoch": 113.5326418547895,
      "grad_norm": 4.070579528808594,
      "learning_rate": 4.8131147540983605e-05,
      "loss": 0.2203,
      "step": 23260
    },
    {
      "epoch": 113.63026235509457,
      "grad_norm": 3.052457809448242,
      "learning_rate": 4.8e-05,
      "loss": 0.2116,
      "step": 23280
    },
    {
      "epoch": 113.72788285539963,
      "grad_norm": 3.2423150539398193,
      "learning_rate": 4.78688524590164e-05,
      "loss": 0.192,
      "step": 23300
    },
    {
      "epoch": 113.8255033557047,
      "grad_norm": 2.9941837787628174,
      "learning_rate": 4.773770491803279e-05,
      "loss": 0.1916,
      "step": 23320
    },
    {
      "epoch": 113.92312385600977,
      "grad_norm": 3.3031139373779297,
      "learning_rate": 4.760655737704918e-05,
      "loss": 0.2261,
      "step": 23340
    },
    {
      "epoch": 114.02074435631482,
      "grad_norm": 3.531646490097046,
      "learning_rate": 4.747540983606558e-05,
      "loss": 0.2186,
      "step": 23360
    },
    {
      "epoch": 114.11836485661989,
      "grad_norm": 2.753261089324951,
      "learning_rate": 4.734426229508197e-05,
      "loss": 0.1979,
      "step": 23380
    },
    {
      "epoch": 114.21598535692496,
      "grad_norm": 3.2900078296661377,
      "learning_rate": 4.7213114754098365e-05,
      "loss": 0.1892,
      "step": 23400
    },
    {
      "epoch": 114.31360585723002,
      "grad_norm": 3.6558027267456055,
      "learning_rate": 4.708196721311476e-05,
      "loss": 0.2202,
      "step": 23420
    },
    {
      "epoch": 114.41122635753509,
      "grad_norm": 2.9026436805725098,
      "learning_rate": 4.695081967213115e-05,
      "loss": 0.2193,
      "step": 23440
    },
    {
      "epoch": 114.50884685784014,
      "grad_norm": 3.0958666801452637,
      "learning_rate": 4.681967213114754e-05,
      "loss": 0.1861,
      "step": 23460
    },
    {
      "epoch": 114.60646735814521,
      "grad_norm": 3.4162192344665527,
      "learning_rate": 4.6688524590163936e-05,
      "loss": 0.1813,
      "step": 23480
    },
    {
      "epoch": 114.70408785845028,
      "grad_norm": 3.4320363998413086,
      "learning_rate": 4.655737704918033e-05,
      "loss": 0.1849,
      "step": 23500
    },
    {
      "epoch": 114.80170835875533,
      "grad_norm": 3.931405544281006,
      "learning_rate": 4.642622950819672e-05,
      "loss": 0.2091,
      "step": 23520
    },
    {
      "epoch": 114.8993288590604,
      "grad_norm": 2.463747024536133,
      "learning_rate": 4.629508196721312e-05,
      "loss": 0.2013,
      "step": 23540
    },
    {
      "epoch": 114.99694935936547,
      "grad_norm": 3.4332265853881836,
      "learning_rate": 4.616393442622951e-05,
      "loss": 0.2224,
      "step": 23560
    },
    {
      "epoch": 115.09456985967053,
      "grad_norm": 4.516382217407227,
      "learning_rate": 4.60327868852459e-05,
      "loss": 0.191,
      "step": 23580
    },
    {
      "epoch": 115.1921903599756,
      "grad_norm": 3.7761762142181396,
      "learning_rate": 4.59016393442623e-05,
      "loss": 0.1972,
      "step": 23600
    },
    {
      "epoch": 115.28981086028065,
      "grad_norm": 3.425079584121704,
      "learning_rate": 4.5770491803278695e-05,
      "loss": 0.2053,
      "step": 23620
    },
    {
      "epoch": 115.38743136058572,
      "grad_norm": 3.098958969116211,
      "learning_rate": 4.5639344262295084e-05,
      "loss": 0.1994,
      "step": 23640
    },
    {
      "epoch": 115.48505186089079,
      "grad_norm": 4.242363452911377,
      "learning_rate": 4.550819672131148e-05,
      "loss": 0.2062,
      "step": 23660
    },
    {
      "epoch": 115.58267236119585,
      "grad_norm": 3.9641380310058594,
      "learning_rate": 4.537704918032787e-05,
      "loss": 0.1892,
      "step": 23680
    },
    {
      "epoch": 115.68029286150092,
      "grad_norm": 3.817915678024292,
      "learning_rate": 4.524590163934426e-05,
      "loss": 0.2076,
      "step": 23700
    },
    {
      "epoch": 115.77791336180599,
      "grad_norm": 4.9167680740356445,
      "learning_rate": 4.5114754098360655e-05,
      "loss": 0.1991,
      "step": 23720
    },
    {
      "epoch": 115.87553386211104,
      "grad_norm": 4.05020809173584,
      "learning_rate": 4.498360655737705e-05,
      "loss": 0.2154,
      "step": 23740
    },
    {
      "epoch": 115.97315436241611,
      "grad_norm": 5.213686466217041,
      "learning_rate": 4.485245901639345e-05,
      "loss": 0.1893,
      "step": 23760
    },
    {
      "epoch": 116.07077486272117,
      "grad_norm": 2.9788458347320557,
      "learning_rate": 4.472131147540984e-05,
      "loss": 0.2015,
      "step": 23780
    },
    {
      "epoch": 116.16839536302624,
      "grad_norm": 3.5112173557281494,
      "learning_rate": 4.459016393442623e-05,
      "loss": 0.2058,
      "step": 23800
    },
    {
      "epoch": 116.2660158633313,
      "grad_norm": 3.3048148155212402,
      "learning_rate": 4.445901639344263e-05,
      "loss": 0.2107,
      "step": 23820
    },
    {
      "epoch": 116.36363636363636,
      "grad_norm": 3.895615339279175,
      "learning_rate": 4.432786885245902e-05,
      "loss": 0.1744,
      "step": 23840
    },
    {
      "epoch": 116.46125686394143,
      "grad_norm": 3.7658941745758057,
      "learning_rate": 4.4196721311475415e-05,
      "loss": 0.1935,
      "step": 23860
    },
    {
      "epoch": 116.5588773642465,
      "grad_norm": 3.4818522930145264,
      "learning_rate": 4.406557377049181e-05,
      "loss": 0.2028,
      "step": 23880
    },
    {
      "epoch": 116.65649786455155,
      "grad_norm": 2.971785068511963,
      "learning_rate": 4.3934426229508194e-05,
      "loss": 0.1993,
      "step": 23900
    },
    {
      "epoch": 116.75411836485662,
      "grad_norm": 4.34529447555542,
      "learning_rate": 4.380327868852459e-05,
      "loss": 0.1887,
      "step": 23920
    },
    {
      "epoch": 116.85173886516168,
      "grad_norm": 3.1781437397003174,
      "learning_rate": 4.3672131147540986e-05,
      "loss": 0.2241,
      "step": 23940
    },
    {
      "epoch": 116.94935936546675,
      "grad_norm": 3.8372061252593994,
      "learning_rate": 4.3540983606557375e-05,
      "loss": 0.2103,
      "step": 23960
    },
    {
      "epoch": 117.04697986577182,
      "grad_norm": 3.981776237487793,
      "learning_rate": 4.340983606557377e-05,
      "loss": 0.1859,
      "step": 23980
    },
    {
      "epoch": 117.14460036607687,
      "grad_norm": 4.0308685302734375,
      "learning_rate": 4.327868852459017e-05,
      "loss": 0.1841,
      "step": 24000
    },
    {
      "epoch": 117.24222086638194,
      "grad_norm": 3.2919681072235107,
      "learning_rate": 4.3147540983606564e-05,
      "loss": 0.1964,
      "step": 24020
    },
    {
      "epoch": 117.33984136668701,
      "grad_norm": 4.155374050140381,
      "learning_rate": 4.301639344262295e-05,
      "loss": 0.1896,
      "step": 24040
    },
    {
      "epoch": 117.43746186699207,
      "grad_norm": 3.5654046535491943,
      "learning_rate": 4.288524590163935e-05,
      "loss": 0.1913,
      "step": 24060
    },
    {
      "epoch": 117.53508236729714,
      "grad_norm": 2.805769681930542,
      "learning_rate": 4.2754098360655745e-05,
      "loss": 0.1995,
      "step": 24080
    },
    {
      "epoch": 117.63270286760219,
      "grad_norm": 2.990482807159424,
      "learning_rate": 4.262295081967213e-05,
      "loss": 0.2034,
      "step": 24100
    },
    {
      "epoch": 117.73032336790726,
      "grad_norm": 3.236433506011963,
      "learning_rate": 4.2491803278688524e-05,
      "loss": 0.1948,
      "step": 24120
    },
    {
      "epoch": 117.82794386821233,
      "grad_norm": 4.235846996307373,
      "learning_rate": 4.236065573770492e-05,
      "loss": 0.2014,
      "step": 24140
    },
    {
      "epoch": 117.92556436851739,
      "grad_norm": 3.238041400909424,
      "learning_rate": 4.222950819672131e-05,
      "loss": 0.2106,
      "step": 24160
    },
    {
      "epoch": 118.02318486882245,
      "grad_norm": 3.18979811668396,
      "learning_rate": 4.2098360655737706e-05,
      "loss": 0.1918,
      "step": 24180
    },
    {
      "epoch": 118.12080536912751,
      "grad_norm": 2.699910879135132,
      "learning_rate": 4.19672131147541e-05,
      "loss": 0.2121,
      "step": 24200
    },
    {
      "epoch": 118.21842586943258,
      "grad_norm": 3.6096503734588623,
      "learning_rate": 4.183606557377049e-05,
      "loss": 0.2165,
      "step": 24220
    },
    {
      "epoch": 118.31604636973765,
      "grad_norm": 3.384746789932251,
      "learning_rate": 4.170491803278689e-05,
      "loss": 0.1884,
      "step": 24240
    },
    {
      "epoch": 118.4136668700427,
      "grad_norm": 3.606003761291504,
      "learning_rate": 4.1573770491803283e-05,
      "loss": 0.1932,
      "step": 24260
    },
    {
      "epoch": 118.51128737034777,
      "grad_norm": 2.9768433570861816,
      "learning_rate": 4.144262295081967e-05,
      "loss": 0.1802,
      "step": 24280
    },
    {
      "epoch": 118.60890787065284,
      "grad_norm": 3.784608840942383,
      "learning_rate": 4.131147540983607e-05,
      "loss": 0.1919,
      "step": 24300
    },
    {
      "epoch": 118.7065283709579,
      "grad_norm": 3.2883870601654053,
      "learning_rate": 4.118032786885246e-05,
      "loss": 0.2015,
      "step": 24320
    },
    {
      "epoch": 118.80414887126297,
      "grad_norm": 4.078542232513428,
      "learning_rate": 4.1049180327868854e-05,
      "loss": 0.1994,
      "step": 24340
    },
    {
      "epoch": 118.90176937156802,
      "grad_norm": 3.7962749004364014,
      "learning_rate": 4.0918032786885244e-05,
      "loss": 0.1958,
      "step": 24360
    },
    {
      "epoch": 118.99938987187309,
      "grad_norm": 3.4484541416168213,
      "learning_rate": 4.078688524590164e-05,
      "loss": 0.1926,
      "step": 24380
    },
    {
      "epoch": 119.09701037217816,
      "grad_norm": 2.9271483421325684,
      "learning_rate": 4.0655737704918036e-05,
      "loss": 0.1818,
      "step": 24400
    },
    {
      "epoch": 119.19463087248322,
      "grad_norm": 5.304340839385986,
      "learning_rate": 4.0524590163934425e-05,
      "loss": 0.192,
      "step": 24420
    },
    {
      "epoch": 119.29225137278829,
      "grad_norm": 4.88720178604126,
      "learning_rate": 4.039344262295082e-05,
      "loss": 0.1975,
      "step": 24440
    },
    {
      "epoch": 119.38987187309336,
      "grad_norm": 3.7572522163391113,
      "learning_rate": 4.026229508196722e-05,
      "loss": 0.1856,
      "step": 24460
    },
    {
      "epoch": 119.48749237339841,
      "grad_norm": 3.3659627437591553,
      "learning_rate": 4.013114754098361e-05,
      "loss": 0.1961,
      "step": 24480
    },
    {
      "epoch": 119.58511287370348,
      "grad_norm": 3.1397533416748047,
      "learning_rate": 4e-05,
      "loss": 0.1991,
      "step": 24500
    },
    {
      "epoch": 119.68273337400854,
      "grad_norm": 3.9070587158203125,
      "learning_rate": 3.98688524590164e-05,
      "loss": 0.1946,
      "step": 24520
    },
    {
      "epoch": 119.7803538743136,
      "grad_norm": 3.3265631198883057,
      "learning_rate": 3.973770491803279e-05,
      "loss": 0.194,
      "step": 24540
    },
    {
      "epoch": 119.87797437461867,
      "grad_norm": 3.5021114349365234,
      "learning_rate": 3.960655737704918e-05,
      "loss": 0.2084,
      "step": 24560
    },
    {
      "epoch": 119.97559487492373,
      "grad_norm": 3.7133572101593018,
      "learning_rate": 3.9475409836065574e-05,
      "loss": 0.2102,
      "step": 24580
    },
    {
      "epoch": 120.0732153752288,
      "grad_norm": 3.27691912651062,
      "learning_rate": 3.934426229508197e-05,
      "loss": 0.1864,
      "step": 24600
    },
    {
      "epoch": 120.17083587553387,
      "grad_norm": 4.74030065536499,
      "learning_rate": 3.921311475409836e-05,
      "loss": 0.1967,
      "step": 24620
    },
    {
      "epoch": 120.26845637583892,
      "grad_norm": 4.080429553985596,
      "learning_rate": 3.9081967213114756e-05,
      "loss": 0.1888,
      "step": 24640
    },
    {
      "epoch": 120.36607687614399,
      "grad_norm": 3.5443646907806396,
      "learning_rate": 3.895081967213115e-05,
      "loss": 0.199,
      "step": 24660
    },
    {
      "epoch": 120.46369737644905,
      "grad_norm": 4.0403923988342285,
      "learning_rate": 3.881967213114754e-05,
      "loss": 0.2002,
      "step": 24680
    },
    {
      "epoch": 120.56131787675412,
      "grad_norm": 4.144064426422119,
      "learning_rate": 3.868852459016394e-05,
      "loss": 0.193,
      "step": 24700
    },
    {
      "epoch": 120.65893837705919,
      "grad_norm": 5.478573799133301,
      "learning_rate": 3.8557377049180334e-05,
      "loss": 0.19,
      "step": 24720
    },
    {
      "epoch": 120.75655887736424,
      "grad_norm": 3.799931287765503,
      "learning_rate": 3.842622950819672e-05,
      "loss": 0.1876,
      "step": 24740
    },
    {
      "epoch": 120.85417937766931,
      "grad_norm": 3.1754050254821777,
      "learning_rate": 3.829508196721311e-05,
      "loss": 0.1881,
      "step": 24760
    },
    {
      "epoch": 120.95179987797438,
      "grad_norm": 3.4417614936828613,
      "learning_rate": 3.816393442622951e-05,
      "loss": 0.2023,
      "step": 24780
    },
    {
      "epoch": 121.04942037827944,
      "grad_norm": 2.689237594604492,
      "learning_rate": 3.8032786885245905e-05,
      "loss": 0.1918,
      "step": 24800
    },
    {
      "epoch": 121.1470408785845,
      "grad_norm": 2.6631858348846436,
      "learning_rate": 3.7901639344262294e-05,
      "loss": 0.1681,
      "step": 24820
    },
    {
      "epoch": 121.24466137888956,
      "grad_norm": 4.315155982971191,
      "learning_rate": 3.777049180327869e-05,
      "loss": 0.2037,
      "step": 24840
    },
    {
      "epoch": 121.34228187919463,
      "grad_norm": 3.0183377265930176,
      "learning_rate": 3.7639344262295086e-05,
      "loss": 0.1901,
      "step": 24860
    },
    {
      "epoch": 121.4399023794997,
      "grad_norm": 2.389106035232544,
      "learning_rate": 3.7508196721311476e-05,
      "loss": 0.2162,
      "step": 24880
    },
    {
      "epoch": 121.53752287980475,
      "grad_norm": 3.5406622886657715,
      "learning_rate": 3.737704918032787e-05,
      "loss": 0.1887,
      "step": 24900
    },
    {
      "epoch": 121.63514338010982,
      "grad_norm": 3.272151231765747,
      "learning_rate": 3.724590163934427e-05,
      "loss": 0.2156,
      "step": 24920
    },
    {
      "epoch": 121.7327638804149,
      "grad_norm": 3.1407511234283447,
      "learning_rate": 3.711475409836066e-05,
      "loss": 0.1833,
      "step": 24940
    },
    {
      "epoch": 121.83038438071995,
      "grad_norm": 3.3532192707061768,
      "learning_rate": 3.698360655737705e-05,
      "loss": 0.1858,
      "step": 24960
    },
    {
      "epoch": 121.92800488102502,
      "grad_norm": 3.1784684658050537,
      "learning_rate": 3.685245901639344e-05,
      "loss": 0.2,
      "step": 24980
    },
    {
      "epoch": 122.02562538133007,
      "grad_norm": 3.5837671756744385,
      "learning_rate": 3.672131147540984e-05,
      "loss": 0.2073,
      "step": 25000
    },
    {
      "epoch": 122.12324588163514,
      "grad_norm": 3.2040927410125732,
      "learning_rate": 3.659016393442623e-05,
      "loss": 0.1712,
      "step": 25020
    },
    {
      "epoch": 122.22086638194021,
      "grad_norm": 3.3119306564331055,
      "learning_rate": 3.6459016393442625e-05,
      "loss": 0.1879,
      "step": 25040
    },
    {
      "epoch": 122.31848688224527,
      "grad_norm": 3.2154297828674316,
      "learning_rate": 3.632786885245902e-05,
      "loss": 0.1976,
      "step": 25060
    },
    {
      "epoch": 122.41610738255034,
      "grad_norm": 2.9070613384246826,
      "learning_rate": 3.619672131147541e-05,
      "loss": 0.2077,
      "step": 25080
    },
    {
      "epoch": 122.5137278828554,
      "grad_norm": 3.866682529449463,
      "learning_rate": 3.6065573770491806e-05,
      "loss": 0.1673,
      "step": 25100
    },
    {
      "epoch": 122.61134838316046,
      "grad_norm": 4.236519813537598,
      "learning_rate": 3.59344262295082e-05,
      "loss": 0.2096,
      "step": 25120
    },
    {
      "epoch": 122.70896888346553,
      "grad_norm": 3.386503219604492,
      "learning_rate": 3.580327868852459e-05,
      "loss": 0.1962,
      "step": 25140
    },
    {
      "epoch": 122.80658938377059,
      "grad_norm": 2.7001540660858154,
      "learning_rate": 3.567213114754099e-05,
      "loss": 0.1731,
      "step": 25160
    },
    {
      "epoch": 122.90420988407566,
      "grad_norm": 3.281818389892578,
      "learning_rate": 3.554098360655738e-05,
      "loss": 0.2225,
      "step": 25180
    },
    {
      "epoch": 123.00183038438072,
      "grad_norm": 4.1742095947265625,
      "learning_rate": 3.5409836065573773e-05,
      "loss": 0.1947,
      "step": 25200
    },
    {
      "epoch": 123.09945088468578,
      "grad_norm": 3.328521490097046,
      "learning_rate": 3.527868852459016e-05,
      "loss": 0.1607,
      "step": 25220
    },
    {
      "epoch": 123.19707138499085,
      "grad_norm": 4.815507888793945,
      "learning_rate": 3.514754098360656e-05,
      "loss": 0.1977,
      "step": 25240
    },
    {
      "epoch": 123.29469188529592,
      "grad_norm": 3.736438512802124,
      "learning_rate": 3.5016393442622955e-05,
      "loss": 0.2105,
      "step": 25260
    },
    {
      "epoch": 123.39231238560097,
      "grad_norm": 3.4552454948425293,
      "learning_rate": 3.4885245901639344e-05,
      "loss": 0.2002,
      "step": 25280
    },
    {
      "epoch": 123.48993288590604,
      "grad_norm": 2.3649439811706543,
      "learning_rate": 3.475409836065574e-05,
      "loss": 0.1823,
      "step": 25300
    },
    {
      "epoch": 123.5875533862111,
      "grad_norm": 3.0043177604675293,
      "learning_rate": 3.462295081967214e-05,
      "loss": 0.1922,
      "step": 25320
    },
    {
      "epoch": 123.68517388651617,
      "grad_norm": 4.2065253257751465,
      "learning_rate": 3.4491803278688526e-05,
      "loss": 0.1943,
      "step": 25340
    },
    {
      "epoch": 123.78279438682124,
      "grad_norm": 3.961331605911255,
      "learning_rate": 3.436065573770492e-05,
      "loss": 0.1859,
      "step": 25360
    },
    {
      "epoch": 123.88041488712629,
      "grad_norm": 4.434045791625977,
      "learning_rate": 3.422950819672131e-05,
      "loss": 0.1985,
      "step": 25380
    },
    {
      "epoch": 123.97803538743136,
      "grad_norm": 2.87196946144104,
      "learning_rate": 3.409836065573771e-05,
      "loss": 0.2046,
      "step": 25400
    },
    {
      "epoch": 124.07565588773643,
      "grad_norm": 2.9702465534210205,
      "learning_rate": 3.39672131147541e-05,
      "loss": 0.1665,
      "step": 25420
    },
    {
      "epoch": 124.17327638804149,
      "grad_norm": 4.106092929840088,
      "learning_rate": 3.383606557377049e-05,
      "loss": 0.1966,
      "step": 25440
    },
    {
      "epoch": 124.27089688834656,
      "grad_norm": 3.1091065406799316,
      "learning_rate": 3.370491803278689e-05,
      "loss": 0.2078,
      "step": 25460
    },
    {
      "epoch": 124.36851738865161,
      "grad_norm": 3.4036500453948975,
      "learning_rate": 3.357377049180328e-05,
      "loss": 0.1814,
      "step": 25480
    },
    {
      "epoch": 124.46613788895668,
      "grad_norm": 3.7814347743988037,
      "learning_rate": 3.3442622950819675e-05,
      "loss": 0.195,
      "step": 25500
    },
    {
      "epoch": 124.56375838926175,
      "grad_norm": 4.475528240203857,
      "learning_rate": 3.331147540983607e-05,
      "loss": 0.1984,
      "step": 25520
    },
    {
      "epoch": 124.6613788895668,
      "grad_norm": 3.35787296295166,
      "learning_rate": 3.318032786885246e-05,
      "loss": 0.1809,
      "step": 25540
    },
    {
      "epoch": 124.75899938987187,
      "grad_norm": 3.594639301300049,
      "learning_rate": 3.3049180327868857e-05,
      "loss": 0.1844,
      "step": 25560
    },
    {
      "epoch": 124.85661989017694,
      "grad_norm": 3.5354974269866943,
      "learning_rate": 3.291803278688525e-05,
      "loss": 0.1953,
      "step": 25580
    },
    {
      "epoch": 124.954240390482,
      "grad_norm": 2.810798168182373,
      "learning_rate": 3.2786885245901635e-05,
      "loss": 0.1911,
      "step": 25600
    },
    {
      "epoch": 125.05186089078707,
      "grad_norm": 3.3332059383392334,
      "learning_rate": 3.265573770491803e-05,
      "loss": 0.1593,
      "step": 25620
    },
    {
      "epoch": 125.14948139109212,
      "grad_norm": 2.8333628177642822,
      "learning_rate": 3.252459016393443e-05,
      "loss": 0.1709,
      "step": 25640
    },
    {
      "epoch": 125.2471018913972,
      "grad_norm": 2.9144675731658936,
      "learning_rate": 3.2393442622950824e-05,
      "loss": 0.2073,
      "step": 25660
    },
    {
      "epoch": 125.34472239170226,
      "grad_norm": 3.185001850128174,
      "learning_rate": 3.226229508196721e-05,
      "loss": 0.1665,
      "step": 25680
    },
    {
      "epoch": 125.44234289200732,
      "grad_norm": 2.8228659629821777,
      "learning_rate": 3.213114754098361e-05,
      "loss": 0.163,
      "step": 25700
    },
    {
      "epoch": 125.53996339231239,
      "grad_norm": 2.8857967853546143,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.2057,
      "step": 25720
    },
    {
      "epoch": 125.63758389261746,
      "grad_norm": 2.9855597019195557,
      "learning_rate": 3.1868852459016395e-05,
      "loss": 0.1872,
      "step": 25740
    },
    {
      "epoch": 125.73520439292251,
      "grad_norm": 3.5702884197235107,
      "learning_rate": 3.173770491803279e-05,
      "loss": 0.1992,
      "step": 25760
    },
    {
      "epoch": 125.83282489322758,
      "grad_norm": 5.302943706512451,
      "learning_rate": 3.160655737704919e-05,
      "loss": 0.1995,
      "step": 25780
    },
    {
      "epoch": 125.93044539353264,
      "grad_norm": 3.9966931343078613,
      "learning_rate": 3.1475409836065576e-05,
      "loss": 0.2046,
      "step": 25800
    },
    {
      "epoch": 126.0280658938377,
      "grad_norm": 2.6985690593719482,
      "learning_rate": 3.1344262295081966e-05,
      "loss": 0.2084,
      "step": 25820
    },
    {
      "epoch": 126.12568639414278,
      "grad_norm": 3.317439079284668,
      "learning_rate": 3.121311475409836e-05,
      "loss": 0.1971,
      "step": 25840
    },
    {
      "epoch": 126.22330689444783,
      "grad_norm": 3.718867301940918,
      "learning_rate": 3.108196721311475e-05,
      "loss": 0.1759,
      "step": 25860
    },
    {
      "epoch": 126.3209273947529,
      "grad_norm": 3.7418856620788574,
      "learning_rate": 3.095081967213115e-05,
      "loss": 0.2067,
      "step": 25880
    },
    {
      "epoch": 126.41854789505797,
      "grad_norm": 3.9287869930267334,
      "learning_rate": 3.0819672131147544e-05,
      "loss": 0.2032,
      "step": 25900
    },
    {
      "epoch": 126.51616839536302,
      "grad_norm": 1.9036474227905273,
      "learning_rate": 3.068852459016393e-05,
      "loss": 0.1905,
      "step": 25920
    },
    {
      "epoch": 126.6137888956681,
      "grad_norm": 2.602092981338501,
      "learning_rate": 3.055737704918033e-05,
      "loss": 0.1851,
      "step": 25940
    },
    {
      "epoch": 126.71140939597315,
      "grad_norm": 3.2280685901641846,
      "learning_rate": 3.0426229508196725e-05,
      "loss": 0.1938,
      "step": 25960
    },
    {
      "epoch": 126.80902989627822,
      "grad_norm": 4.606971263885498,
      "learning_rate": 3.0295081967213118e-05,
      "loss": 0.1644,
      "step": 25980
    },
    {
      "epoch": 126.90665039658329,
      "grad_norm": 3.755833864212036,
      "learning_rate": 3.016393442622951e-05,
      "loss": 0.1836,
      "step": 26000
    },
    {
      "epoch": 127.00427089688834,
      "grad_norm": 3.572577953338623,
      "learning_rate": 3.00327868852459e-05,
      "loss": 0.1733,
      "step": 26020
    },
    {
      "epoch": 127.10189139719341,
      "grad_norm": 3.1435470581054688,
      "learning_rate": 2.9901639344262293e-05,
      "loss": 0.1815,
      "step": 26040
    },
    {
      "epoch": 127.19951189749847,
      "grad_norm": 3.3165206909179688,
      "learning_rate": 2.977049180327869e-05,
      "loss": 0.183,
      "step": 26060
    },
    {
      "epoch": 127.29713239780354,
      "grad_norm": 3.269935131072998,
      "learning_rate": 2.963934426229508e-05,
      "loss": 0.1854,
      "step": 26080
    },
    {
      "epoch": 127.3947528981086,
      "grad_norm": 3.6275577545166016,
      "learning_rate": 2.9508196721311478e-05,
      "loss": 0.1798,
      "step": 26100
    },
    {
      "epoch": 127.49237339841366,
      "grad_norm": 3.3832483291625977,
      "learning_rate": 2.937704918032787e-05,
      "loss": 0.1722,
      "step": 26120
    },
    {
      "epoch": 127.58999389871873,
      "grad_norm": 3.828364610671997,
      "learning_rate": 2.9245901639344263e-05,
      "loss": 0.1853,
      "step": 26140
    },
    {
      "epoch": 127.6876143990238,
      "grad_norm": 3.7207860946655273,
      "learning_rate": 2.911475409836066e-05,
      "loss": 0.1948,
      "step": 26160
    },
    {
      "epoch": 127.78523489932886,
      "grad_norm": 3.281031847000122,
      "learning_rate": 2.8983606557377052e-05,
      "loss": 0.2063,
      "step": 26180
    },
    {
      "epoch": 127.88285539963393,
      "grad_norm": 3.1116421222686768,
      "learning_rate": 2.8852459016393445e-05,
      "loss": 0.191,
      "step": 26200
    },
    {
      "epoch": 127.98047589993898,
      "grad_norm": 4.254022121429443,
      "learning_rate": 2.872131147540984e-05,
      "loss": 0.202,
      "step": 26220
    },
    {
      "epoch": 128.07809640024405,
      "grad_norm": 3.1279819011688232,
      "learning_rate": 2.8590163934426227e-05,
      "loss": 0.1934,
      "step": 26240
    },
    {
      "epoch": 128.1757169005491,
      "grad_norm": 3.3977596759796143,
      "learning_rate": 2.8459016393442623e-05,
      "loss": 0.1807,
      "step": 26260
    },
    {
      "epoch": 128.2733374008542,
      "grad_norm": 5.261218070983887,
      "learning_rate": 2.8327868852459016e-05,
      "loss": 0.17,
      "step": 26280
    },
    {
      "epoch": 128.37095790115924,
      "grad_norm": 4.153654098510742,
      "learning_rate": 2.819672131147541e-05,
      "loss": 0.1891,
      "step": 26300
    },
    {
      "epoch": 128.4685784014643,
      "grad_norm": 3.449397563934326,
      "learning_rate": 2.8065573770491805e-05,
      "loss": 0.1846,
      "step": 26320
    },
    {
      "epoch": 128.56619890176938,
      "grad_norm": 3.43442702293396,
      "learning_rate": 2.7934426229508198e-05,
      "loss": 0.1807,
      "step": 26340
    },
    {
      "epoch": 128.66381940207444,
      "grad_norm": 2.9243948459625244,
      "learning_rate": 2.7803278688524594e-05,
      "loss": 0.1704,
      "step": 26360
    },
    {
      "epoch": 128.7614399023795,
      "grad_norm": 3.9830613136291504,
      "learning_rate": 2.7672131147540987e-05,
      "loss": 0.1909,
      "step": 26380
    },
    {
      "epoch": 128.85906040268458,
      "grad_norm": 3.0765368938446045,
      "learning_rate": 2.754098360655738e-05,
      "loss": 0.1954,
      "step": 26400
    },
    {
      "epoch": 128.95668090298963,
      "grad_norm": 2.540853977203369,
      "learning_rate": 2.7409836065573775e-05,
      "loss": 0.2116,
      "step": 26420
    },
    {
      "epoch": 129.0543014032947,
      "grad_norm": 3.55985689163208,
      "learning_rate": 2.7278688524590168e-05,
      "loss": 0.193,
      "step": 26440
    },
    {
      "epoch": 129.15192190359974,
      "grad_norm": 3.974700450897217,
      "learning_rate": 2.7147540983606558e-05,
      "loss": 0.1673,
      "step": 26460
    },
    {
      "epoch": 129.24954240390483,
      "grad_norm": 4.614022731781006,
      "learning_rate": 2.701639344262295e-05,
      "loss": 0.1968,
      "step": 26480
    },
    {
      "epoch": 129.34716290420988,
      "grad_norm": 3.9277889728546143,
      "learning_rate": 2.6885245901639343e-05,
      "loss": 0.1643,
      "step": 26500
    },
    {
      "epoch": 129.44478340451494,
      "grad_norm": 3.4351913928985596,
      "learning_rate": 2.675409836065574e-05,
      "loss": 0.1903,
      "step": 26520
    },
    {
      "epoch": 129.54240390482002,
      "grad_norm": 3.3347392082214355,
      "learning_rate": 2.6622950819672132e-05,
      "loss": 0.1789,
      "step": 26540
    },
    {
      "epoch": 129.64002440512508,
      "grad_norm": 3.970414161682129,
      "learning_rate": 2.6491803278688525e-05,
      "loss": 0.1994,
      "step": 26560
    },
    {
      "epoch": 129.73764490543013,
      "grad_norm": 3.648883819580078,
      "learning_rate": 2.636065573770492e-05,
      "loss": 0.1877,
      "step": 26580
    },
    {
      "epoch": 129.8352654057352,
      "grad_norm": 3.3394792079925537,
      "learning_rate": 2.6229508196721314e-05,
      "loss": 0.1838,
      "step": 26600
    },
    {
      "epoch": 129.93288590604027,
      "grad_norm": 2.924798011779785,
      "learning_rate": 2.6098360655737706e-05,
      "loss": 0.1913,
      "step": 26620
    },
    {
      "epoch": 130.03050640634532,
      "grad_norm": 3.448457717895508,
      "learning_rate": 2.5967213114754103e-05,
      "loss": 0.1993,
      "step": 26640
    },
    {
      "epoch": 130.1281269066504,
      "grad_norm": 2.458868980407715,
      "learning_rate": 2.5836065573770492e-05,
      "loss": 0.1996,
      "step": 26660
    },
    {
      "epoch": 130.22574740695546,
      "grad_norm": 4.39287805557251,
      "learning_rate": 2.5704918032786885e-05,
      "loss": 0.1894,
      "step": 26680
    },
    {
      "epoch": 130.32336790726052,
      "grad_norm": 3.347745180130005,
      "learning_rate": 2.5573770491803277e-05,
      "loss": 0.1642,
      "step": 26700
    },
    {
      "epoch": 130.4209884075656,
      "grad_norm": 3.0466248989105225,
      "learning_rate": 2.5442622950819674e-05,
      "loss": 0.1976,
      "step": 26720
    },
    {
      "epoch": 130.51860890787066,
      "grad_norm": 2.580834150314331,
      "learning_rate": 2.5311475409836066e-05,
      "loss": 0.1764,
      "step": 26740
    },
    {
      "epoch": 130.6162294081757,
      "grad_norm": 3.6775128841400146,
      "learning_rate": 2.518032786885246e-05,
      "loss": 0.1725,
      "step": 26760
    },
    {
      "epoch": 130.71384990848077,
      "grad_norm": 3.829058885574341,
      "learning_rate": 2.5049180327868855e-05,
      "loss": 0.1776,
      "step": 26780
    },
    {
      "epoch": 130.81147040878585,
      "grad_norm": 4.047943592071533,
      "learning_rate": 2.4918032786885248e-05,
      "loss": 0.1943,
      "step": 26800
    },
    {
      "epoch": 130.9090909090909,
      "grad_norm": 3.2828705310821533,
      "learning_rate": 2.478688524590164e-05,
      "loss": 0.1864,
      "step": 26820
    },
    {
      "epoch": 131.00671140939596,
      "grad_norm": 4.2563958168029785,
      "learning_rate": 2.4655737704918033e-05,
      "loss": 0.1838,
      "step": 26840
    },
    {
      "epoch": 131.10433190970105,
      "grad_norm": 3.279503583908081,
      "learning_rate": 2.4524590163934426e-05,
      "loss": 0.2027,
      "step": 26860
    },
    {
      "epoch": 131.2019524100061,
      "grad_norm": 3.8052897453308105,
      "learning_rate": 2.4393442622950822e-05,
      "loss": 0.1916,
      "step": 26880
    },
    {
      "epoch": 131.29957291031116,
      "grad_norm": 3.12294602394104,
      "learning_rate": 2.4262295081967215e-05,
      "loss": 0.1542,
      "step": 26900
    },
    {
      "epoch": 131.39719341061624,
      "grad_norm": 4.636548042297363,
      "learning_rate": 2.4131147540983608e-05,
      "loss": 0.1677,
      "step": 26920
    },
    {
      "epoch": 131.4948139109213,
      "grad_norm": 2.8608415126800537,
      "learning_rate": 2.4e-05,
      "loss": 0.194,
      "step": 26940
    },
    {
      "epoch": 131.59243441122635,
      "grad_norm": 3.7946908473968506,
      "learning_rate": 2.3868852459016393e-05,
      "loss": 0.1764,
      "step": 26960
    },
    {
      "epoch": 131.69005491153143,
      "grad_norm": 3.1568832397460938,
      "learning_rate": 2.373770491803279e-05,
      "loss": 0.1865,
      "step": 26980
    },
    {
      "epoch": 131.7876754118365,
      "grad_norm": 4.103198528289795,
      "learning_rate": 2.3606557377049182e-05,
      "loss": 0.1753,
      "step": 27000
    },
    {
      "epoch": 131.88529591214154,
      "grad_norm": 3.655327796936035,
      "learning_rate": 2.3475409836065575e-05,
      "loss": 0.1965,
      "step": 27020
    },
    {
      "epoch": 131.98291641244663,
      "grad_norm": 3.908200263977051,
      "learning_rate": 2.3344262295081968e-05,
      "loss": 0.1791,
      "step": 27040
    },
    {
      "epoch": 132.08053691275168,
      "grad_norm": 3.2557129859924316,
      "learning_rate": 2.321311475409836e-05,
      "loss": 0.1715,
      "step": 27060
    },
    {
      "epoch": 132.17815741305674,
      "grad_norm": 3.102268695831299,
      "learning_rate": 2.3081967213114757e-05,
      "loss": 0.191,
      "step": 27080
    },
    {
      "epoch": 132.2757779133618,
      "grad_norm": 2.7028493881225586,
      "learning_rate": 2.295081967213115e-05,
      "loss": 0.1802,
      "step": 27100
    },
    {
      "epoch": 132.37339841366688,
      "grad_norm": 2.5240046977996826,
      "learning_rate": 2.2819672131147542e-05,
      "loss": 0.1618,
      "step": 27120
    },
    {
      "epoch": 132.47101891397193,
      "grad_norm": 3.0410265922546387,
      "learning_rate": 2.2688524590163935e-05,
      "loss": 0.1833,
      "step": 27140
    },
    {
      "epoch": 132.568639414277,
      "grad_norm": 3.677824020385742,
      "learning_rate": 2.2557377049180328e-05,
      "loss": 0.1804,
      "step": 27160
    },
    {
      "epoch": 132.66625991458207,
      "grad_norm": 2.826828718185425,
      "learning_rate": 2.2426229508196724e-05,
      "loss": 0.1915,
      "step": 27180
    },
    {
      "epoch": 132.76388041488713,
      "grad_norm": 4.393260955810547,
      "learning_rate": 2.2295081967213117e-05,
      "loss": 0.1952,
      "step": 27200
    },
    {
      "epoch": 132.86150091519218,
      "grad_norm": 3.1146349906921387,
      "learning_rate": 2.216393442622951e-05,
      "loss": 0.1973,
      "step": 27220
    },
    {
      "epoch": 132.95912141549726,
      "grad_norm": 3.343693494796753,
      "learning_rate": 2.2032786885245905e-05,
      "loss": 0.1822,
      "step": 27240
    },
    {
      "epoch": 133.05674191580232,
      "grad_norm": 3.0667552947998047,
      "learning_rate": 2.1901639344262295e-05,
      "loss": 0.1807,
      "step": 27260
    },
    {
      "epoch": 133.15436241610738,
      "grad_norm": 3.497859001159668,
      "learning_rate": 2.1770491803278688e-05,
      "loss": 0.1525,
      "step": 27280
    },
    {
      "epoch": 133.25198291641246,
      "grad_norm": 3.295478343963623,
      "learning_rate": 2.1639344262295084e-05,
      "loss": 0.1971,
      "step": 27300
    },
    {
      "epoch": 133.3496034167175,
      "grad_norm": 3.3247010707855225,
      "learning_rate": 2.1508196721311476e-05,
      "loss": 0.1963,
      "step": 27320
    },
    {
      "epoch": 133.44722391702257,
      "grad_norm": 4.269167900085449,
      "learning_rate": 2.1377049180327873e-05,
      "loss": 0.1967,
      "step": 27340
    },
    {
      "epoch": 133.54484441732765,
      "grad_norm": 3.7575721740722656,
      "learning_rate": 2.1245901639344262e-05,
      "loss": 0.1788,
      "step": 27360
    },
    {
      "epoch": 133.6424649176327,
      "grad_norm": 4.189979553222656,
      "learning_rate": 2.1114754098360655e-05,
      "loss": 0.1922,
      "step": 27380
    },
    {
      "epoch": 133.74008541793776,
      "grad_norm": 3.48610782623291,
      "learning_rate": 2.098360655737705e-05,
      "loss": 0.1749,
      "step": 27400
    },
    {
      "epoch": 133.83770591824282,
      "grad_norm": 3.320037364959717,
      "learning_rate": 2.0852459016393444e-05,
      "loss": 0.1886,
      "step": 27420
    },
    {
      "epoch": 133.9353264185479,
      "grad_norm": 3.347099781036377,
      "learning_rate": 2.0721311475409836e-05,
      "loss": 0.1717,
      "step": 27440
    },
    {
      "epoch": 134.03294691885296,
      "grad_norm": 2.9406418800354004,
      "learning_rate": 2.059016393442623e-05,
      "loss": 0.1702,
      "step": 27460
    },
    {
      "epoch": 134.130567419158,
      "grad_norm": 2.3794620037078857,
      "learning_rate": 2.0459016393442622e-05,
      "loss": 0.1872,
      "step": 27480
    },
    {
      "epoch": 134.2281879194631,
      "grad_norm": 3.1150660514831543,
      "learning_rate": 2.0327868852459018e-05,
      "loss": 0.2023,
      "step": 27500
    },
    {
      "epoch": 134.32580841976815,
      "grad_norm": 3.679694414138794,
      "learning_rate": 2.019672131147541e-05,
      "loss": 0.1858,
      "step": 27520
    },
    {
      "epoch": 134.4234289200732,
      "grad_norm": 2.665882110595703,
      "learning_rate": 2.0065573770491804e-05,
      "loss": 0.1782,
      "step": 27540
    },
    {
      "epoch": 134.5210494203783,
      "grad_norm": 3.0053212642669678,
      "learning_rate": 1.99344262295082e-05,
      "loss": 0.1773,
      "step": 27560
    },
    {
      "epoch": 134.61866992068335,
      "grad_norm": 2.689307451248169,
      "learning_rate": 1.980327868852459e-05,
      "loss": 0.1588,
      "step": 27580
    },
    {
      "epoch": 134.7162904209884,
      "grad_norm": 4.318088054656982,
      "learning_rate": 1.9672131147540985e-05,
      "loss": 0.1436,
      "step": 27600
    },
    {
      "epoch": 134.81391092129348,
      "grad_norm": 3.3378102779388428,
      "learning_rate": 1.9540983606557378e-05,
      "loss": 0.1919,
      "step": 27620
    },
    {
      "epoch": 134.91153142159854,
      "grad_norm": 3.2355871200561523,
      "learning_rate": 1.940983606557377e-05,
      "loss": 0.1828,
      "step": 27640
    },
    {
      "epoch": 135.0091519219036,
      "grad_norm": 3.1335229873657227,
      "learning_rate": 1.9278688524590167e-05,
      "loss": 0.1982,
      "step": 27660
    },
    {
      "epoch": 135.10677242220865,
      "grad_norm": 4.066319465637207,
      "learning_rate": 1.9147540983606556e-05,
      "loss": 0.1526,
      "step": 27680
    },
    {
      "epoch": 135.20439292251373,
      "grad_norm": 3.631089925765991,
      "learning_rate": 1.9016393442622952e-05,
      "loss": 0.1776,
      "step": 27700
    },
    {
      "epoch": 135.3020134228188,
      "grad_norm": 3.5840327739715576,
      "learning_rate": 1.8885245901639345e-05,
      "loss": 0.1922,
      "step": 27720
    },
    {
      "epoch": 135.39963392312384,
      "grad_norm": 2.926558256149292,
      "learning_rate": 1.8754098360655738e-05,
      "loss": 0.1847,
      "step": 27740
    },
    {
      "epoch": 135.49725442342893,
      "grad_norm": 4.487957000732422,
      "learning_rate": 1.8622950819672134e-05,
      "loss": 0.1896,
      "step": 27760
    },
    {
      "epoch": 135.59487492373398,
      "grad_norm": 3.209500789642334,
      "learning_rate": 1.8491803278688523e-05,
      "loss": 0.1827,
      "step": 27780
    },
    {
      "epoch": 135.69249542403904,
      "grad_norm": 2.8735058307647705,
      "learning_rate": 1.836065573770492e-05,
      "loss": 0.1671,
      "step": 27800
    },
    {
      "epoch": 135.79011592434412,
      "grad_norm": 3.221266508102417,
      "learning_rate": 1.8229508196721312e-05,
      "loss": 0.1861,
      "step": 27820
    },
    {
      "epoch": 135.88773642464918,
      "grad_norm": 3.1269659996032715,
      "learning_rate": 1.8098360655737705e-05,
      "loss": 0.1857,
      "step": 27840
    },
    {
      "epoch": 135.98535692495423,
      "grad_norm": 3.3478143215179443,
      "learning_rate": 1.79672131147541e-05,
      "loss": 0.1853,
      "step": 27860
    },
    {
      "epoch": 136.08297742525932,
      "grad_norm": 2.8102643489837646,
      "learning_rate": 1.7836065573770494e-05,
      "loss": 0.1691,
      "step": 27880
    },
    {
      "epoch": 136.18059792556437,
      "grad_norm": 2.9172909259796143,
      "learning_rate": 1.7704918032786887e-05,
      "loss": 0.1805,
      "step": 27900
    },
    {
      "epoch": 136.27821842586943,
      "grad_norm": 3.6803557872772217,
      "learning_rate": 1.757377049180328e-05,
      "loss": 0.1808,
      "step": 27920
    },
    {
      "epoch": 136.3758389261745,
      "grad_norm": 3.6122121810913086,
      "learning_rate": 1.7442622950819672e-05,
      "loss": 0.1825,
      "step": 27940
    },
    {
      "epoch": 136.47345942647956,
      "grad_norm": 2.836901903152466,
      "learning_rate": 1.731147540983607e-05,
      "loss": 0.1762,
      "step": 27960
    },
    {
      "epoch": 136.57107992678462,
      "grad_norm": 4.931893348693848,
      "learning_rate": 1.718032786885246e-05,
      "loss": 0.2097,
      "step": 27980
    },
    {
      "epoch": 136.66870042708968,
      "grad_norm": 3.634223461151123,
      "learning_rate": 1.7049180327868854e-05,
      "loss": 0.1764,
      "step": 28000
    },
    {
      "epoch": 136.76632092739476,
      "grad_norm": 4.417370319366455,
      "learning_rate": 1.6918032786885247e-05,
      "loss": 0.1695,
      "step": 28020
    },
    {
      "epoch": 136.8639414276998,
      "grad_norm": 3.6770431995391846,
      "learning_rate": 1.678688524590164e-05,
      "loss": 0.1762,
      "step": 28040
    },
    {
      "epoch": 136.96156192800487,
      "grad_norm": 3.134272813796997,
      "learning_rate": 1.6655737704918036e-05,
      "loss": 0.1603,
      "step": 28060
    },
    {
      "epoch": 137.05918242830995,
      "grad_norm": 3.993882179260254,
      "learning_rate": 1.6524590163934428e-05,
      "loss": 0.1927,
      "step": 28080
    },
    {
      "epoch": 137.156802928615,
      "grad_norm": 3.1614527702331543,
      "learning_rate": 1.6393442622950818e-05,
      "loss": 0.1739,
      "step": 28100
    },
    {
      "epoch": 137.25442342892006,
      "grad_norm": 2.9293642044067383,
      "learning_rate": 1.6262295081967214e-05,
      "loss": 0.1916,
      "step": 28120
    },
    {
      "epoch": 137.35204392922515,
      "grad_norm": 3.292917251586914,
      "learning_rate": 1.6131147540983607e-05,
      "loss": 0.1638,
      "step": 28140
    },
    {
      "epoch": 137.4496644295302,
      "grad_norm": 3.048471212387085,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.1877,
      "step": 28160
    },
    {
      "epoch": 137.54728492983526,
      "grad_norm": 3.398252487182617,
      "learning_rate": 1.5868852459016395e-05,
      "loss": 0.1685,
      "step": 28180
    },
    {
      "epoch": 137.64490543014034,
      "grad_norm": 3.906764268875122,
      "learning_rate": 1.5737704918032788e-05,
      "loss": 0.1772,
      "step": 28200
    },
    {
      "epoch": 137.7425259304454,
      "grad_norm": 3.1852598190307617,
      "learning_rate": 1.560655737704918e-05,
      "loss": 0.168,
      "step": 28220
    },
    {
      "epoch": 137.84014643075045,
      "grad_norm": 2.9285385608673096,
      "learning_rate": 1.5475409836065574e-05,
      "loss": 0.1733,
      "step": 28240
    },
    {
      "epoch": 137.93776693105553,
      "grad_norm": 3.9846913814544678,
      "learning_rate": 1.5344262295081966e-05,
      "loss": 0.1906,
      "step": 28260
    },
    {
      "epoch": 138.0353874313606,
      "grad_norm": 3.3251185417175293,
      "learning_rate": 1.5213114754098363e-05,
      "loss": 0.1872,
      "step": 28280
    },
    {
      "epoch": 138.13300793166565,
      "grad_norm": 3.7904114723205566,
      "learning_rate": 1.5081967213114755e-05,
      "loss": 0.1811,
      "step": 28300
    },
    {
      "epoch": 138.2306284319707,
      "grad_norm": 3.5363810062408447,
      "learning_rate": 1.4950819672131146e-05,
      "loss": 0.1803,
      "step": 28320
    },
    {
      "epoch": 138.32824893227578,
      "grad_norm": 3.430577516555786,
      "learning_rate": 1.481967213114754e-05,
      "loss": 0.1619,
      "step": 28340
    },
    {
      "epoch": 138.42586943258084,
      "grad_norm": 3.1190154552459717,
      "learning_rate": 1.4688524590163935e-05,
      "loss": 0.1909,
      "step": 28360
    },
    {
      "epoch": 138.5234899328859,
      "grad_norm": 2.656212091445923,
      "learning_rate": 1.455737704918033e-05,
      "loss": 0.1682,
      "step": 28380
    },
    {
      "epoch": 138.62111043319098,
      "grad_norm": 3.1637041568756104,
      "learning_rate": 1.4426229508196722e-05,
      "loss": 0.1729,
      "step": 28400
    },
    {
      "epoch": 138.71873093349603,
      "grad_norm": 3.313136100769043,
      "learning_rate": 1.4295081967213114e-05,
      "loss": 0.1853,
      "step": 28420
    },
    {
      "epoch": 138.8163514338011,
      "grad_norm": 3.63885498046875,
      "learning_rate": 1.4163934426229508e-05,
      "loss": 0.1783,
      "step": 28440
    },
    {
      "epoch": 138.91397193410617,
      "grad_norm": 4.251205921173096,
      "learning_rate": 1.4032786885245902e-05,
      "loss": 0.1703,
      "step": 28460
    },
    {
      "epoch": 139.01159243441123,
      "grad_norm": 3.76887583732605,
      "learning_rate": 1.3901639344262297e-05,
      "loss": 0.1871,
      "step": 28480
    },
    {
      "epoch": 139.10921293471628,
      "grad_norm": 3.2986671924591064,
      "learning_rate": 1.377049180327869e-05,
      "loss": 0.1728,
      "step": 28500
    },
    {
      "epoch": 139.20683343502137,
      "grad_norm": 2.5212690830230713,
      "learning_rate": 1.3639344262295084e-05,
      "loss": 0.175,
      "step": 28520
    },
    {
      "epoch": 139.30445393532642,
      "grad_norm": 4.492109298706055,
      "learning_rate": 1.3508196721311475e-05,
      "loss": 0.1838,
      "step": 28540
    },
    {
      "epoch": 139.40207443563148,
      "grad_norm": 3.0226120948791504,
      "learning_rate": 1.337704918032787e-05,
      "loss": 0.1753,
      "step": 28560
    },
    {
      "epoch": 139.49969493593656,
      "grad_norm": 2.4843361377716064,
      "learning_rate": 1.3245901639344262e-05,
      "loss": 0.1915,
      "step": 28580
    },
    {
      "epoch": 139.59731543624162,
      "grad_norm": 3.4304590225219727,
      "learning_rate": 1.3114754098360657e-05,
      "loss": 0.1703,
      "step": 28600
    },
    {
      "epoch": 139.69493593654667,
      "grad_norm": 3.242751121520996,
      "learning_rate": 1.2983606557377051e-05,
      "loss": 0.176,
      "step": 28620
    },
    {
      "epoch": 139.79255643685173,
      "grad_norm": 3.2713284492492676,
      "learning_rate": 1.2852459016393442e-05,
      "loss": 0.1838,
      "step": 28640
    },
    {
      "epoch": 139.8901769371568,
      "grad_norm": 2.9921929836273193,
      "learning_rate": 1.2721311475409837e-05,
      "loss": 0.1764,
      "step": 28660
    },
    {
      "epoch": 139.98779743746186,
      "grad_norm": 3.6933350563049316,
      "learning_rate": 1.259016393442623e-05,
      "loss": 0.1803,
      "step": 28680
    },
    {
      "epoch": 140.08541793776692,
      "grad_norm": 2.947892665863037,
      "learning_rate": 1.2459016393442624e-05,
      "loss": 0.1732,
      "step": 28700
    },
    {
      "epoch": 140.183038438072,
      "grad_norm": 3.5047738552093506,
      "learning_rate": 1.2327868852459017e-05,
      "loss": 0.1751,
      "step": 28720
    },
    {
      "epoch": 140.28065893837706,
      "grad_norm": 2.9315600395202637,
      "learning_rate": 1.2196721311475411e-05,
      "loss": 0.1879,
      "step": 28740
    },
    {
      "epoch": 140.3782794386821,
      "grad_norm": 3.6878082752227783,
      "learning_rate": 1.2065573770491804e-05,
      "loss": 0.1551,
      "step": 28760
    },
    {
      "epoch": 140.4758999389872,
      "grad_norm": 3.2162342071533203,
      "learning_rate": 1.1934426229508197e-05,
      "loss": 0.183,
      "step": 28780
    },
    {
      "epoch": 140.57352043929225,
      "grad_norm": 3.6583456993103027,
      "learning_rate": 1.1803278688524591e-05,
      "loss": 0.1784,
      "step": 28800
    },
    {
      "epoch": 140.6711409395973,
      "grad_norm": 2.780412435531616,
      "learning_rate": 1.1672131147540984e-05,
      "loss": 0.1885,
      "step": 28820
    },
    {
      "epoch": 140.7687614399024,
      "grad_norm": 2.3773672580718994,
      "learning_rate": 1.1540983606557378e-05,
      "loss": 0.1611,
      "step": 28840
    },
    {
      "epoch": 140.86638194020745,
      "grad_norm": 3.6276142597198486,
      "learning_rate": 1.1409836065573771e-05,
      "loss": 0.1806,
      "step": 28860
    },
    {
      "epoch": 140.9640024405125,
      "grad_norm": 3.3313121795654297,
      "learning_rate": 1.1278688524590164e-05,
      "loss": 0.1741,
      "step": 28880
    },
    {
      "epoch": 141.06162294081759,
      "grad_norm": 3.1109941005706787,
      "learning_rate": 1.1147540983606558e-05,
      "loss": 0.1747,
      "step": 28900
    },
    {
      "epoch": 141.15924344112264,
      "grad_norm": 2.3083291053771973,
      "learning_rate": 1.1016393442622953e-05,
      "loss": 0.16,
      "step": 28920
    },
    {
      "epoch": 141.2568639414277,
      "grad_norm": 3.6427536010742188,
      "learning_rate": 1.0885245901639344e-05,
      "loss": 0.1792,
      "step": 28940
    },
    {
      "epoch": 141.35448444173275,
      "grad_norm": 3.2836146354675293,
      "learning_rate": 1.0754098360655738e-05,
      "loss": 0.1921,
      "step": 28960
    },
    {
      "epoch": 141.45210494203783,
      "grad_norm": 3.712411642074585,
      "learning_rate": 1.0622950819672131e-05,
      "loss": 0.1834,
      "step": 28980
    },
    {
      "epoch": 141.5497254423429,
      "grad_norm": 2.6064016819000244,
      "learning_rate": 1.0491803278688525e-05,
      "loss": 0.1975,
      "step": 29000
    },
    {
      "epoch": 141.64734594264795,
      "grad_norm": 3.3907470703125,
      "learning_rate": 1.0360655737704918e-05,
      "loss": 0.172,
      "step": 29020
    },
    {
      "epoch": 141.74496644295303,
      "grad_norm": 3.0713050365448,
      "learning_rate": 1.0229508196721311e-05,
      "loss": 0.1547,
      "step": 29040
    },
    {
      "epoch": 141.84258694325808,
      "grad_norm": 5.218588352203369,
      "learning_rate": 1.0098360655737705e-05,
      "loss": 0.1712,
      "step": 29060
    },
    {
      "epoch": 141.94020744356314,
      "grad_norm": 3.205132007598877,
      "learning_rate": 9.9672131147541e-06,
      "loss": 0.1629,
      "step": 29080
    },
    {
      "epoch": 142.03782794386822,
      "grad_norm": 3.580003499984741,
      "learning_rate": 9.836065573770493e-06,
      "loss": 0.1877,
      "step": 29100
    },
    {
      "epoch": 142.13544844417328,
      "grad_norm": 3.393789529800415,
      "learning_rate": 9.704918032786885e-06,
      "loss": 0.1928,
      "step": 29120
    },
    {
      "epoch": 142.23306894447833,
      "grad_norm": 2.592445135116577,
      "learning_rate": 9.573770491803278e-06,
      "loss": 0.1638,
      "step": 29140
    },
    {
      "epoch": 142.33068944478342,
      "grad_norm": 4.577868938446045,
      "learning_rate": 9.442622950819673e-06,
      "loss": 0.1868,
      "step": 29160
    },
    {
      "epoch": 142.42830994508847,
      "grad_norm": 3.1620144844055176,
      "learning_rate": 9.311475409836067e-06,
      "loss": 0.1635,
      "step": 29180
    },
    {
      "epoch": 142.52593044539353,
      "grad_norm": 3.382749319076538,
      "learning_rate": 9.18032786885246e-06,
      "loss": 0.1469,
      "step": 29200
    },
    {
      "epoch": 142.6235509456986,
      "grad_norm": 3.314983606338501,
      "learning_rate": 9.049180327868853e-06,
      "loss": 0.1719,
      "step": 29220
    },
    {
      "epoch": 142.72117144600367,
      "grad_norm": 2.643578290939331,
      "learning_rate": 8.918032786885247e-06,
      "loss": 0.1587,
      "step": 29240
    },
    {
      "epoch": 142.81879194630872,
      "grad_norm": 2.4660592079162598,
      "learning_rate": 8.78688524590164e-06,
      "loss": 0.2031,
      "step": 29260
    },
    {
      "epoch": 142.91641244661378,
      "grad_norm": 3.2867209911346436,
      "learning_rate": 8.655737704918034e-06,
      "loss": 0.1679,
      "step": 29280
    },
    {
      "epoch": 143.01403294691886,
      "grad_norm": 3.2089104652404785,
      "learning_rate": 8.524590163934427e-06,
      "loss": 0.171,
      "step": 29300
    },
    {
      "epoch": 143.11165344722392,
      "grad_norm": 2.9183740615844727,
      "learning_rate": 8.39344262295082e-06,
      "loss": 0.1642,
      "step": 29320
    },
    {
      "epoch": 143.20927394752897,
      "grad_norm": 2.9158482551574707,
      "learning_rate": 8.262295081967214e-06,
      "loss": 0.1743,
      "step": 29340
    },
    {
      "epoch": 143.30689444783405,
      "grad_norm": 3.256065607070923,
      "learning_rate": 8.131147540983607e-06,
      "loss": 0.1816,
      "step": 29360
    },
    {
      "epoch": 143.4045149481391,
      "grad_norm": 2.916098117828369,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.1575,
      "step": 29380
    },
    {
      "epoch": 143.50213544844416,
      "grad_norm": 3.414485454559326,
      "learning_rate": 7.868852459016394e-06,
      "loss": 0.1584,
      "step": 29400
    },
    {
      "epoch": 143.59975594874925,
      "grad_norm": 3.5028018951416016,
      "learning_rate": 7.737704918032787e-06,
      "loss": 0.1731,
      "step": 29420
    },
    {
      "epoch": 143.6973764490543,
      "grad_norm": 2.7573652267456055,
      "learning_rate": 7.606557377049181e-06,
      "loss": 0.1848,
      "step": 29440
    },
    {
      "epoch": 143.79499694935936,
      "grad_norm": 3.1906893253326416,
      "learning_rate": 7.475409836065573e-06,
      "loss": 0.1842,
      "step": 29460
    },
    {
      "epoch": 143.89261744966444,
      "grad_norm": 2.384742021560669,
      "learning_rate": 7.344262295081968e-06,
      "loss": 0.1852,
      "step": 29480
    },
    {
      "epoch": 143.9902379499695,
      "grad_norm": 3.7481918334960938,
      "learning_rate": 7.213114754098361e-06,
      "loss": 0.1794,
      "step": 29500
    },
    {
      "epoch": 144.08785845027455,
      "grad_norm": 2.903989791870117,
      "learning_rate": 7.081967213114754e-06,
      "loss": 0.1745,
      "step": 29520
    },
    {
      "epoch": 144.1854789505796,
      "grad_norm": 3.4449713230133057,
      "learning_rate": 6.9508196721311484e-06,
      "loss": 0.1874,
      "step": 29540
    },
    {
      "epoch": 144.2830994508847,
      "grad_norm": 2.9290127754211426,
      "learning_rate": 6.819672131147542e-06,
      "loss": 0.1736,
      "step": 29560
    },
    {
      "epoch": 144.38071995118975,
      "grad_norm": 2.805908203125,
      "learning_rate": 6.688524590163935e-06,
      "loss": 0.187,
      "step": 29580
    },
    {
      "epoch": 144.4783404514948,
      "grad_norm": 2.9539241790771484,
      "learning_rate": 6.557377049180328e-06,
      "loss": 0.1856,
      "step": 29600
    },
    {
      "epoch": 144.57596095179989,
      "grad_norm": 2.8198204040527344,
      "learning_rate": 6.426229508196721e-06,
      "loss": 0.1733,
      "step": 29620
    },
    {
      "epoch": 144.67358145210494,
      "grad_norm": 3.3926849365234375,
      "learning_rate": 6.295081967213115e-06,
      "loss": 0.18,
      "step": 29640
    },
    {
      "epoch": 144.77120195241,
      "grad_norm": 4.100579261779785,
      "learning_rate": 6.163934426229508e-06,
      "loss": 0.1568,
      "step": 29660
    },
    {
      "epoch": 144.86882245271508,
      "grad_norm": 3.2875492572784424,
      "learning_rate": 6.032786885245902e-06,
      "loss": 0.1561,
      "step": 29680
    },
    {
      "epoch": 144.96644295302013,
      "grad_norm": 2.626185417175293,
      "learning_rate": 5.9016393442622956e-06,
      "loss": 0.1714,
      "step": 29700
    },
    {
      "epoch": 145.0640634533252,
      "grad_norm": 4.3447265625,
      "learning_rate": 5.770491803278689e-06,
      "loss": 0.1951,
      "step": 29720
    },
    {
      "epoch": 145.16168395363027,
      "grad_norm": 4.056821346282959,
      "learning_rate": 5.639344262295082e-06,
      "loss": 0.1621,
      "step": 29740
    },
    {
      "epoch": 145.25930445393533,
      "grad_norm": 3.4116666316986084,
      "learning_rate": 5.508196721311476e-06,
      "loss": 0.179,
      "step": 29760
    },
    {
      "epoch": 145.35692495424038,
      "grad_norm": 2.810452699661255,
      "learning_rate": 5.377049180327869e-06,
      "loss": 0.1723,
      "step": 29780
    },
    {
      "epoch": 145.45454545454547,
      "grad_norm": 3.708115816116333,
      "learning_rate": 5.245901639344263e-06,
      "loss": 0.1675,
      "step": 29800
    },
    {
      "epoch": 145.55216595485052,
      "grad_norm": 4.000546455383301,
      "learning_rate": 5.1147540983606555e-06,
      "loss": 0.1716,
      "step": 29820
    },
    {
      "epoch": 145.64978645515558,
      "grad_norm": 3.2421109676361084,
      "learning_rate": 4.98360655737705e-06,
      "loss": 0.1656,
      "step": 29840
    },
    {
      "epoch": 145.74740695546063,
      "grad_norm": 3.13706111907959,
      "learning_rate": 4.852459016393443e-06,
      "loss": 0.1822,
      "step": 29860
    },
    {
      "epoch": 145.84502745576572,
      "grad_norm": 3.364842176437378,
      "learning_rate": 4.721311475409836e-06,
      "loss": 0.1772,
      "step": 29880
    },
    {
      "epoch": 145.94264795607077,
      "grad_norm": 3.2013063430786133,
      "learning_rate": 4.59016393442623e-06,
      "loss": 0.1717,
      "step": 29900
    },
    {
      "epoch": 146.04026845637583,
      "grad_norm": 4.174123287200928,
      "learning_rate": 4.4590163934426235e-06,
      "loss": 0.1731,
      "step": 29920
    },
    {
      "epoch": 146.1378889566809,
      "grad_norm": 2.8885281085968018,
      "learning_rate": 4.327868852459017e-06,
      "loss": 0.1421,
      "step": 29940
    },
    {
      "epoch": 146.23550945698597,
      "grad_norm": 2.6078240871429443,
      "learning_rate": 4.19672131147541e-06,
      "loss": 0.1898,
      "step": 29960
    },
    {
      "epoch": 146.33312995729102,
      "grad_norm": 4.43600606918335,
      "learning_rate": 4.0655737704918034e-06,
      "loss": 0.1787,
      "step": 29980
    },
    {
      "epoch": 146.4307504575961,
      "grad_norm": 2.9905123710632324,
      "learning_rate": 3.934426229508197e-06,
      "loss": 0.1866,
      "step": 30000
    },
    {
      "epoch": 146.52837095790116,
      "grad_norm": 2.8058741092681885,
      "learning_rate": 3.8032786885245906e-06,
      "loss": 0.1751,
      "step": 30020
    },
    {
      "epoch": 146.62599145820622,
      "grad_norm": 2.752978563308716,
      "learning_rate": 3.672131147540984e-06,
      "loss": 0.1826,
      "step": 30040
    },
    {
      "epoch": 146.7236119585113,
      "grad_norm": 3.0315961837768555,
      "learning_rate": 3.540983606557377e-06,
      "loss": 0.1623,
      "step": 30060
    },
    {
      "epoch": 146.82123245881635,
      "grad_norm": 3.3782765865325928,
      "learning_rate": 3.409836065573771e-06,
      "loss": 0.1784,
      "step": 30080
    },
    {
      "epoch": 146.9188529591214,
      "grad_norm": 2.961002826690674,
      "learning_rate": 3.278688524590164e-06,
      "loss": 0.1764,
      "step": 30100
    },
    {
      "epoch": 147.0164734594265,
      "grad_norm": 3.775484561920166,
      "learning_rate": 3.1475409836065574e-06,
      "loss": 0.1595,
      "step": 30120
    },
    {
      "epoch": 147.11409395973155,
      "grad_norm": 4.18531608581543,
      "learning_rate": 3.016393442622951e-06,
      "loss": 0.1599,
      "step": 30140
    },
    {
      "epoch": 147.2117144600366,
      "grad_norm": 2.9723432064056396,
      "learning_rate": 2.8852459016393446e-06,
      "loss": 0.159,
      "step": 30160
    },
    {
      "epoch": 147.30933496034166,
      "grad_norm": 3.2833070755004883,
      "learning_rate": 2.754098360655738e-06,
      "loss": 0.1657,
      "step": 30180
    },
    {
      "epoch": 147.40695546064674,
      "grad_norm": 3.4174959659576416,
      "learning_rate": 2.6229508196721314e-06,
      "loss": 0.175,
      "step": 30200
    },
    {
      "epoch": 147.5045759609518,
      "grad_norm": 3.3127195835113525,
      "learning_rate": 2.491803278688525e-06,
      "loss": 0.1815,
      "step": 30220
    },
    {
      "epoch": 147.60219646125685,
      "grad_norm": 3.7137949466705322,
      "learning_rate": 2.360655737704918e-06,
      "loss": 0.198,
      "step": 30240
    },
    {
      "epoch": 147.69981696156194,
      "grad_norm": 2.630924701690674,
      "learning_rate": 2.2295081967213117e-06,
      "loss": 0.1687,
      "step": 30260
    },
    {
      "epoch": 147.797437461867,
      "grad_norm": 3.330245018005371,
      "learning_rate": 2.098360655737705e-06,
      "loss": 0.1581,
      "step": 30280
    },
    {
      "epoch": 147.89505796217205,
      "grad_norm": 3.237410068511963,
      "learning_rate": 1.9672131147540985e-06,
      "loss": 0.2022,
      "step": 30300
    },
    {
      "epoch": 147.99267846247713,
      "grad_norm": 2.633331537246704,
      "learning_rate": 1.836065573770492e-06,
      "loss": 0.1613,
      "step": 30320
    },
    {
      "epoch": 148.09029896278219,
      "grad_norm": 2.527902603149414,
      "learning_rate": 1.7049180327868855e-06,
      "loss": 0.1749,
      "step": 30340
    },
    {
      "epoch": 148.18791946308724,
      "grad_norm": 2.9230234622955322,
      "learning_rate": 1.5737704918032787e-06,
      "loss": 0.1464,
      "step": 30360
    },
    {
      "epoch": 148.28553996339232,
      "grad_norm": 2.591038703918457,
      "learning_rate": 1.4426229508196723e-06,
      "loss": 0.1819,
      "step": 30380
    },
    {
      "epoch": 148.38316046369738,
      "grad_norm": 3.6826913356781006,
      "learning_rate": 1.3114754098360657e-06,
      "loss": 0.1909,
      "step": 30400
    },
    {
      "epoch": 148.48078096400243,
      "grad_norm": 3.1828205585479736,
      "learning_rate": 1.180327868852459e-06,
      "loss": 0.1727,
      "step": 30420
    },
    {
      "epoch": 148.57840146430752,
      "grad_norm": 3.3356974124908447,
      "learning_rate": 1.0491803278688525e-06,
      "loss": 0.1624,
      "step": 30440
    },
    {
      "epoch": 148.67602196461257,
      "grad_norm": 3.1692721843719482,
      "learning_rate": 9.18032786885246e-07,
      "loss": 0.1769,
      "step": 30460
    },
    {
      "epoch": 148.77364246491763,
      "grad_norm": 2.968018054962158,
      "learning_rate": 7.868852459016393e-07,
      "loss": 0.1594,
      "step": 30480
    },
    {
      "epoch": 148.87126296522268,
      "grad_norm": 3.693136692047119,
      "learning_rate": 6.557377049180328e-07,
      "loss": 0.1927,
      "step": 30500
    }
  ],
  "logging_steps": 20,
  "max_steps": 30600,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 150,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.924112697660375e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}