{
  "best_metric": 0.33997730174492835,
  "best_model_checkpoint": "Mrohit01/1_lakh_cards-swinv2-base-patch4-window12to16-192to256-22kto1k-ft-finetuned/checkpoint-54461",
  "epoch": 9.99958686221855,
  "eval_steps": 500,
  "global_step": 60510,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016525511258004544,
      "grad_norm": 34.320098876953125,
      "learning_rate": 8.263097008758884e-08,
      "loss": 2.4664,
      "step": 10
    },
    {
      "epoch": 0.003305102251600909,
      "grad_norm": 319.06292724609375,
      "learning_rate": 1.6526194017517767e-07,
      "loss": 2.4352,
      "step": 20
    },
    {
      "epoch": 0.004957653377401363,
      "grad_norm": 57.93561553955078,
      "learning_rate": 2.478929102627665e-07,
      "loss": 2.3859,
      "step": 30
    },
    {
      "epoch": 0.006610204503201818,
      "grad_norm": 37.62372589111328,
      "learning_rate": 3.3052388035035534e-07,
      "loss": 2.3897,
      "step": 40
    },
    {
      "epoch": 0.008262755629002272,
      "grad_norm": 44.00041580200195,
      "learning_rate": 4.131548504379442e-07,
      "loss": 2.4616,
      "step": 50
    },
    {
      "epoch": 0.009915306754802726,
      "grad_norm": 25.988052368164062,
      "learning_rate": 4.95785820525533e-07,
      "loss": 2.427,
      "step": 60
    },
    {
      "epoch": 0.011567857880603181,
      "grad_norm": 105.19337463378906,
      "learning_rate": 5.784167906131218e-07,
      "loss": 2.4586,
      "step": 70
    },
    {
      "epoch": 0.013220409006403635,
      "grad_norm": 30.26386833190918,
      "learning_rate": 6.610477607007107e-07,
      "loss": 2.3847,
      "step": 80
    },
    {
      "epoch": 0.01487296013220409,
      "grad_norm": 35.36358642578125,
      "learning_rate": 7.436787307882995e-07,
      "loss": 2.3328,
      "step": 90
    },
    {
      "epoch": 0.016525511258004544,
      "grad_norm": 262.9168395996094,
      "learning_rate": 8.263097008758884e-07,
      "loss": 2.3209,
      "step": 100
    },
    {
      "epoch": 0.018178062383805,
      "grad_norm": 31.34343910217285,
      "learning_rate": 9.089406709634771e-07,
      "loss": 2.3854,
      "step": 110
    },
    {
      "epoch": 0.019830613509605452,
      "grad_norm": 21.935100555419922,
      "learning_rate": 9.91571641051066e-07,
      "loss": 2.3157,
      "step": 120
    },
    {
      "epoch": 0.021483164635405907,
      "grad_norm": 41.69834899902344,
      "learning_rate": 1.0742026111386547e-06,
      "loss": 2.3495,
      "step": 130
    },
    {
      "epoch": 0.023135715761206363,
      "grad_norm": 25.686676025390625,
      "learning_rate": 1.1568335812262436e-06,
      "loss": 2.2876,
      "step": 140
    },
    {
      "epoch": 0.02478826688700682,
      "grad_norm": 20.90071678161621,
      "learning_rate": 1.2394645513138325e-06,
      "loss": 2.333,
      "step": 150
    },
    {
      "epoch": 0.02644081801280727,
      "grad_norm": 30.158994674682617,
      "learning_rate": 1.3220955214014214e-06,
      "loss": 2.3064,
      "step": 160
    },
    {
      "epoch": 0.028093369138607726,
      "grad_norm": 29.495481491088867,
      "learning_rate": 1.4047264914890103e-06,
      "loss": 2.2628,
      "step": 170
    },
    {
      "epoch": 0.02974592026440818,
      "grad_norm": 20.178300857543945,
      "learning_rate": 1.487357461576599e-06,
      "loss": 2.2932,
      "step": 180
    },
    {
      "epoch": 0.031398471390208633,
      "grad_norm": 36.64740753173828,
      "learning_rate": 1.5699884316641878e-06,
      "loss": 2.3664,
      "step": 190
    },
    {
      "epoch": 0.03305102251600909,
      "grad_norm": 19.702604293823242,
      "learning_rate": 1.6526194017517767e-06,
      "loss": 2.3554,
      "step": 200
    },
    {
      "epoch": 0.034703573641809544,
      "grad_norm": 18.79967498779297,
      "learning_rate": 1.7352503718393656e-06,
      "loss": 2.32,
      "step": 210
    },
    {
      "epoch": 0.03635612476761,
      "grad_norm": 24.093412399291992,
      "learning_rate": 1.8178813419269543e-06,
      "loss": 2.2258,
      "step": 220
    },
    {
      "epoch": 0.038008675893410455,
      "grad_norm": 20.522418975830078,
      "learning_rate": 1.9005123120145432e-06,
      "loss": 2.2418,
      "step": 230
    },
    {
      "epoch": 0.039661227019210904,
      "grad_norm": 21.670164108276367,
      "learning_rate": 1.983143282102132e-06,
      "loss": 2.2425,
      "step": 240
    },
    {
      "epoch": 0.04131377814501136,
      "grad_norm": 24.221439361572266,
      "learning_rate": 2.0657742521897205e-06,
      "loss": 2.2311,
      "step": 250
    },
    {
      "epoch": 0.042966329270811815,
      "grad_norm": 26.810293197631836,
      "learning_rate": 2.1484052222773094e-06,
      "loss": 2.3007,
      "step": 260
    },
    {
      "epoch": 0.04461888039661227,
      "grad_norm": 33.146942138671875,
      "learning_rate": 2.2310361923648987e-06,
      "loss": 2.1901,
      "step": 270
    },
    {
      "epoch": 0.046271431522412726,
      "grad_norm": 24.769393920898438,
      "learning_rate": 2.313667162452487e-06,
      "loss": 2.2548,
      "step": 280
    },
    {
      "epoch": 0.04792398264821318,
      "grad_norm": 16.452192306518555,
      "learning_rate": 2.396298132540076e-06,
      "loss": 2.1442,
      "step": 290
    },
    {
      "epoch": 0.04957653377401364,
      "grad_norm": 17.729570388793945,
      "learning_rate": 2.478929102627665e-06,
      "loss": 2.1889,
      "step": 300
    },
    {
      "epoch": 0.051229084899814085,
      "grad_norm": 19.994186401367188,
      "learning_rate": 2.561560072715254e-06,
      "loss": 2.2463,
      "step": 310
    },
    {
      "epoch": 0.05288163602561454,
      "grad_norm": 27.13626480102539,
      "learning_rate": 2.6441910428028427e-06,
      "loss": 2.2024,
      "step": 320
    },
    {
      "epoch": 0.054534187151414996,
      "grad_norm": 20.60004997253418,
      "learning_rate": 2.7268220128904316e-06,
      "loss": 2.1578,
      "step": 330
    },
    {
      "epoch": 0.05618673827721545,
      "grad_norm": 20.759340286254883,
      "learning_rate": 2.8094529829780205e-06,
      "loss": 2.0856,
      "step": 340
    },
    {
      "epoch": 0.05783928940301591,
      "grad_norm": 34.48766326904297,
      "learning_rate": 2.892083953065609e-06,
      "loss": 2.1879,
      "step": 350
    },
    {
      "epoch": 0.05949184052881636,
      "grad_norm": 21.343008041381836,
      "learning_rate": 2.974714923153198e-06,
      "loss": 2.1215,
      "step": 360
    },
    {
      "epoch": 0.06114439165461681,
      "grad_norm": 20.634981155395508,
      "learning_rate": 3.0573458932407868e-06,
      "loss": 2.1139,
      "step": 370
    },
    {
      "epoch": 0.06279694278041727,
      "grad_norm": 23.69481086730957,
      "learning_rate": 3.1399768633283757e-06,
      "loss": 2.1182,
      "step": 380
    },
    {
      "epoch": 0.06444949390621772,
      "grad_norm": 21.073101043701172,
      "learning_rate": 3.2226078334159645e-06,
      "loss": 2.1002,
      "step": 390
    },
    {
      "epoch": 0.06610204503201818,
      "grad_norm": 74.09575653076172,
      "learning_rate": 3.3052388035035534e-06,
      "loss": 2.0813,
      "step": 400
    },
    {
      "epoch": 0.06775459615781863,
      "grad_norm": 18.449506759643555,
      "learning_rate": 3.3878697735911423e-06,
      "loss": 2.0566,
      "step": 410
    },
    {
      "epoch": 0.06940714728361909,
      "grad_norm": 28.852758407592773,
      "learning_rate": 3.470500743678731e-06,
      "loss": 1.9957,
      "step": 420
    },
    {
      "epoch": 0.07105969840941954,
      "grad_norm": 26.230737686157227,
      "learning_rate": 3.5531317137663197e-06,
      "loss": 1.9894,
      "step": 430
    },
    {
      "epoch": 0.07271224953522,
      "grad_norm": 24.988384246826172,
      "learning_rate": 3.6357626838539086e-06,
      "loss": 2.0197,
      "step": 440
    },
    {
      "epoch": 0.07436480066102046,
      "grad_norm": 101.62005615234375,
      "learning_rate": 3.7183936539414975e-06,
      "loss": 2.074,
      "step": 450
    },
    {
      "epoch": 0.07601735178682091,
      "grad_norm": 24.157245635986328,
      "learning_rate": 3.8010246240290863e-06,
      "loss": 2.1024,
      "step": 460
    },
    {
      "epoch": 0.07766990291262135,
      "grad_norm": 26.56095314025879,
      "learning_rate": 3.883655594116676e-06,
      "loss": 2.0546,
      "step": 470
    },
    {
      "epoch": 0.07932245403842181,
      "grad_norm": 59.286495208740234,
      "learning_rate": 3.966286564204264e-06,
      "loss": 2.0861,
      "step": 480
    },
    {
      "epoch": 0.08097500516422226,
      "grad_norm": 27.197782516479492,
      "learning_rate": 4.0489175342918534e-06,
      "loss": 2.0757,
      "step": 490
    },
    {
      "epoch": 0.08262755629002272,
      "grad_norm": 28.272613525390625,
      "learning_rate": 4.131548504379441e-06,
      "loss": 1.8954,
      "step": 500
    },
    {
      "epoch": 0.08428010741582317,
      "grad_norm": 23.627155303955078,
      "learning_rate": 4.21417947446703e-06,
      "loss": 1.9886,
      "step": 510
    },
    {
      "epoch": 0.08593265854162363,
      "grad_norm": 28.818801879882812,
      "learning_rate": 4.296810444554619e-06,
      "loss": 2.0049,
      "step": 520
    },
    {
      "epoch": 0.08758520966742409,
      "grad_norm": 31.500320434570312,
      "learning_rate": 4.379441414642208e-06,
      "loss": 1.9709,
      "step": 530
    },
    {
      "epoch": 0.08923776079322454,
      "grad_norm": 32.4036865234375,
      "learning_rate": 4.4620723847297975e-06,
      "loss": 1.957,
      "step": 540
    },
    {
      "epoch": 0.090890311919025,
      "grad_norm": 18.269474029541016,
      "learning_rate": 4.544703354817386e-06,
      "loss": 2.0274,
      "step": 550
    },
    {
      "epoch": 0.09254286304482545,
      "grad_norm": 25.879230499267578,
      "learning_rate": 4.627334324904974e-06,
      "loss": 1.982,
      "step": 560
    },
    {
      "epoch": 0.09419541417062591,
      "grad_norm": 78.83243560791016,
      "learning_rate": 4.709965294992563e-06,
      "loss": 1.9635,
      "step": 570
    },
    {
      "epoch": 0.09584796529642636,
      "grad_norm": 24.732942581176758,
      "learning_rate": 4.792596265080152e-06,
      "loss": 1.9254,
      "step": 580
    },
    {
      "epoch": 0.09750051642222682,
      "grad_norm": 23.506771087646484,
      "learning_rate": 4.8752272351677415e-06,
      "loss": 1.951,
      "step": 590
    },
    {
      "epoch": 0.09915306754802727,
      "grad_norm": 24.17493438720703,
      "learning_rate": 4.95785820525533e-06,
      "loss": 1.9573,
      "step": 600
    },
    {
      "epoch": 0.10080561867382772,
      "grad_norm": 25.89427375793457,
      "learning_rate": 5.040489175342919e-06,
      "loss": 2.0148,
      "step": 610
    },
    {
      "epoch": 0.10245816979962817,
      "grad_norm": 22.84876251220703,
      "learning_rate": 5.123120145430508e-06,
      "loss": 1.8556,
      "step": 620
    },
    {
      "epoch": 0.10411072092542863,
      "grad_norm": 229.06407165527344,
      "learning_rate": 5.205751115518096e-06,
      "loss": 1.9412,
      "step": 630
    },
    {
      "epoch": 0.10576327205122908,
      "grad_norm": 22.079832077026367,
      "learning_rate": 5.2883820856056855e-06,
      "loss": 1.8359,
      "step": 640
    },
    {
      "epoch": 0.10741582317702954,
      "grad_norm": 23.32947540283203,
      "learning_rate": 5.371013055693274e-06,
      "loss": 1.7962,
      "step": 650
    },
    {
      "epoch": 0.10906837430282999,
      "grad_norm": 25.554899215698242,
      "learning_rate": 5.453644025780863e-06,
      "loss": 1.8114,
      "step": 660
    },
    {
      "epoch": 0.11072092542863045,
      "grad_norm": 23.066320419311523,
      "learning_rate": 5.536274995868452e-06,
      "loss": 1.9601,
      "step": 670
    },
    {
      "epoch": 0.1123734765544309,
      "grad_norm": 29.641036987304688,
      "learning_rate": 5.618905965956041e-06,
      "loss": 1.8361,
      "step": 680
    },
    {
      "epoch": 0.11402602768023136,
      "grad_norm": 40.079246520996094,
      "learning_rate": 5.7015369360436295e-06,
      "loss": 1.7822,
      "step": 690
    },
    {
      "epoch": 0.11567857880603181,
      "grad_norm": 46.70073699951172,
      "learning_rate": 5.784167906131218e-06,
      "loss": 1.7975,
      "step": 700
    },
    {
      "epoch": 0.11733112993183227,
      "grad_norm": 27.957921981811523,
      "learning_rate": 5.866798876218807e-06,
      "loss": 1.9022,
      "step": 710
    },
    {
      "epoch": 0.11898368105763273,
      "grad_norm": 87.9314956665039,
      "learning_rate": 5.949429846306396e-06,
      "loss": 1.7631,
      "step": 720
    },
    {
      "epoch": 0.12063623218343318,
      "grad_norm": 28.518705368041992,
      "learning_rate": 6.032060816393985e-06,
      "loss": 1.8346,
      "step": 730
    },
    {
      "epoch": 0.12228878330923362,
      "grad_norm": 35.01418685913086,
      "learning_rate": 6.1146917864815735e-06,
      "loss": 1.7283,
      "step": 740
    },
    {
      "epoch": 0.12394133443503408,
      "grad_norm": 25.934650421142578,
      "learning_rate": 6.197322756569163e-06,
      "loss": 1.7489,
      "step": 750
    },
    {
      "epoch": 0.12559388556083453,
      "grad_norm": 49.37101364135742,
      "learning_rate": 6.279953726656751e-06,
      "loss": 1.7778,
      "step": 760
    },
    {
      "epoch": 0.127246436686635,
      "grad_norm": 40.169677734375,
      "learning_rate": 6.362584696744341e-06,
      "loss": 1.9885,
      "step": 770
    },
    {
      "epoch": 0.12889898781243544,
      "grad_norm": 39.495182037353516,
      "learning_rate": 6.445215666831929e-06,
      "loss": 1.8897,
      "step": 780
    },
    {
      "epoch": 0.1305515389382359,
      "grad_norm": 33.42755889892578,
      "learning_rate": 6.527846636919517e-06,
      "loss": 1.8193,
      "step": 790
    },
    {
      "epoch": 0.13220409006403636,
      "grad_norm": 53.90789794921875,
      "learning_rate": 6.610477607007107e-06,
      "loss": 1.7312,
      "step": 800
    },
    {
      "epoch": 0.1338566411898368,
      "grad_norm": 42.573890686035156,
      "learning_rate": 6.693108577094695e-06,
      "loss": 1.8778,
      "step": 810
    },
    {
      "epoch": 0.13550919231563727,
      "grad_norm": 20.9980525970459,
      "learning_rate": 6.775739547182285e-06,
      "loss": 1.7827,
      "step": 820
    },
    {
      "epoch": 0.13716174344143772,
      "grad_norm": 27.766277313232422,
      "learning_rate": 6.858370517269873e-06,
      "loss": 1.8191,
      "step": 830
    },
    {
      "epoch": 0.13881429456723818,
      "grad_norm": 26.448427200317383,
      "learning_rate": 6.941001487357462e-06,
      "loss": 1.8897,
      "step": 840
    },
    {
      "epoch": 0.14046684569303863,
      "grad_norm": 52.19208908081055,
      "learning_rate": 7.023632457445051e-06,
      "loss": 1.932,
      "step": 850
    },
    {
      "epoch": 0.1421193968188391,
      "grad_norm": 21.94788360595703,
      "learning_rate": 7.106263427532639e-06,
      "loss": 1.8672,
      "step": 860
    },
    {
      "epoch": 0.14377194794463954,
      "grad_norm": 40.3758430480957,
      "learning_rate": 7.188894397620229e-06,
      "loss": 1.8301,
      "step": 870
    },
    {
      "epoch": 0.14542449907044,
      "grad_norm": 22.807653427124023,
      "learning_rate": 7.271525367707817e-06,
      "loss": 1.8321,
      "step": 880
    },
    {
      "epoch": 0.14707705019624046,
      "grad_norm": 22.939292907714844,
      "learning_rate": 7.3541563377954064e-06,
      "loss": 1.7721,
      "step": 890
    },
    {
      "epoch": 0.1487296013220409,
      "grad_norm": 34.43024444580078,
      "learning_rate": 7.436787307882995e-06,
      "loss": 1.7898,
      "step": 900
    },
    {
      "epoch": 0.15038215244784137,
      "grad_norm": 19.193397521972656,
      "learning_rate": 7.519418277970584e-06,
      "loss": 1.8499,
      "step": 910
    },
    {
      "epoch": 0.15203470357364182,
      "grad_norm": 26.9038143157959,
      "learning_rate": 7.602049248058173e-06,
      "loss": 1.8749,
      "step": 920
    },
    {
      "epoch": 0.15368725469944228,
      "grad_norm": 30.20845603942871,
      "learning_rate": 7.684680218145761e-06,
      "loss": 1.7995,
      "step": 930
    },
    {
      "epoch": 0.1553398058252427,
      "grad_norm": 30.089397430419922,
      "learning_rate": 7.767311188233351e-06,
      "loss": 1.8802,
      "step": 940
    },
    {
      "epoch": 0.15699235695104316,
      "grad_norm": 18.006601333618164,
      "learning_rate": 7.849942158320938e-06,
      "loss": 1.8049,
      "step": 950
    },
    {
      "epoch": 0.15864490807684362,
      "grad_norm": 22.729703903198242,
      "learning_rate": 7.932573128408528e-06,
      "loss": 1.8031,
      "step": 960
    },
    {
      "epoch": 0.16029745920264407,
      "grad_norm": 22.711090087890625,
      "learning_rate": 8.015204098496117e-06,
      "loss": 1.7107,
      "step": 970
    },
    {
      "epoch": 0.16195001032844453,
      "grad_norm": 24.80642318725586,
      "learning_rate": 8.097835068583707e-06,
      "loss": 1.8531,
      "step": 980
    },
    {
      "epoch": 0.16360256145424498,
      "grad_norm": 28.32562255859375,
      "learning_rate": 8.180466038671294e-06,
      "loss": 1.927,
      "step": 990
    },
    {
      "epoch": 0.16525511258004544,
      "grad_norm": 20.05076789855957,
      "learning_rate": 8.263097008758882e-06,
      "loss": 1.7359,
      "step": 1000
    },
    {
      "epoch": 0.1669076637058459,
      "grad_norm": 32.90656661987305,
      "learning_rate": 8.345727978846472e-06,
      "loss": 1.7466,
      "step": 1010
    },
    {
      "epoch": 0.16856021483164635,
      "grad_norm": 44.9754524230957,
      "learning_rate": 8.42835894893406e-06,
      "loss": 1.8069,
      "step": 1020
    },
    {
      "epoch": 0.1702127659574468,
      "grad_norm": 201.73477172851562,
      "learning_rate": 8.510989919021651e-06,
      "loss": 1.7286,
      "step": 1030
    },
    {
      "epoch": 0.17186531708324726,
      "grad_norm": 18.8424015045166,
      "learning_rate": 8.593620889109238e-06,
      "loss": 1.8063,
      "step": 1040
    },
    {
      "epoch": 0.17351786820904772,
      "grad_norm": 26.58818244934082,
      "learning_rate": 8.676251859196826e-06,
      "loss": 1.9267,
      "step": 1050
    },
    {
      "epoch": 0.17517041933484817,
      "grad_norm": 37.0977897644043,
      "learning_rate": 8.758882829284416e-06,
      "loss": 1.7116,
      "step": 1060
    },
    {
      "epoch": 0.17682297046064863,
      "grad_norm": 19.845773696899414,
      "learning_rate": 8.841513799372005e-06,
      "loss": 1.7994,
      "step": 1070
    },
    {
      "epoch": 0.17847552158644908,
      "grad_norm": 93.75279235839844,
      "learning_rate": 8.924144769459595e-06,
      "loss": 1.8503,
      "step": 1080
    },
    {
      "epoch": 0.18012807271224954,
      "grad_norm": 50.5566291809082,
      "learning_rate": 9.006775739547182e-06,
      "loss": 1.7882,
      "step": 1090
    },
    {
      "epoch": 0.18178062383805,
      "grad_norm": 19.830806732177734,
      "learning_rate": 9.089406709634772e-06,
      "loss": 1.7492,
      "step": 1100
    },
    {
      "epoch": 0.18343317496385045,
      "grad_norm": 88.4405746459961,
      "learning_rate": 9.17203767972236e-06,
      "loss": 1.8902,
      "step": 1110
    },
    {
      "epoch": 0.1850857260896509,
      "grad_norm": 38.67186737060547,
      "learning_rate": 9.254668649809949e-06,
      "loss": 1.7064,
      "step": 1120
    },
    {
      "epoch": 0.18673827721545136,
      "grad_norm": 17.757261276245117,
      "learning_rate": 9.337299619897539e-06,
      "loss": 1.7849,
      "step": 1130
    },
    {
      "epoch": 0.18839082834125181,
      "grad_norm": 20.545978546142578,
      "learning_rate": 9.419930589985126e-06,
      "loss": 1.7574,
      "step": 1140
    },
    {
      "epoch": 0.19004337946705227,
      "grad_norm": 15.944082260131836,
      "learning_rate": 9.502561560072716e-06,
      "loss": 1.7188,
      "step": 1150
    },
    {
      "epoch": 0.19169593059285273,
      "grad_norm": 32.81608963012695,
      "learning_rate": 9.585192530160304e-06,
      "loss": 1.9046,
      "step": 1160
    },
    {
      "epoch": 0.19334848171865318,
      "grad_norm": 18.563947677612305,
      "learning_rate": 9.667823500247894e-06,
      "loss": 1.7162,
      "step": 1170
    },
    {
      "epoch": 0.19500103284445364,
      "grad_norm": 25.642000198364258,
      "learning_rate": 9.750454470335483e-06,
      "loss": 1.7451,
      "step": 1180
    },
    {
      "epoch": 0.1966535839702541,
      "grad_norm": 32.40825271606445,
      "learning_rate": 9.83308544042307e-06,
      "loss": 1.7892,
      "step": 1190
    },
    {
      "epoch": 0.19830613509605455,
      "grad_norm": 19.13682746887207,
      "learning_rate": 9.91571641051066e-06,
      "loss": 1.8015,
      "step": 1200
    },
    {
      "epoch": 0.19995868622185498,
      "grad_norm": 28.85736083984375,
      "learning_rate": 9.998347380598248e-06,
      "loss": 1.7506,
      "step": 1210
    },
    {
      "epoch": 0.20161123734765543,
      "grad_norm": 15.917028427124023,
      "learning_rate": 1.0080978350685838e-05,
      "loss": 1.7728,
      "step": 1220
    },
    {
      "epoch": 0.2032637884734559,
      "grad_norm": 29.365341186523438,
      "learning_rate": 1.0163609320773427e-05,
      "loss": 1.8862,
      "step": 1230
    },
    {
      "epoch": 0.20491633959925634,
      "grad_norm": 30.120737075805664,
      "learning_rate": 1.0246240290861015e-05,
      "loss": 1.7482,
      "step": 1240
    },
    {
      "epoch": 0.2065688907250568,
      "grad_norm": 19.36783790588379,
      "learning_rate": 1.0328871260948604e-05,
      "loss": 1.6657,
      "step": 1250
    },
    {
      "epoch": 0.20822144185085725,
      "grad_norm": 21.701364517211914,
      "learning_rate": 1.0411502231036192e-05,
      "loss": 1.8962,
      "step": 1260
    },
    {
      "epoch": 0.2098739929766577,
      "grad_norm": 23.340579986572266,
      "learning_rate": 1.0494133201123783e-05,
      "loss": 1.828,
      "step": 1270
    },
    {
      "epoch": 0.21152654410245816,
      "grad_norm": 22.94394874572754,
      "learning_rate": 1.0576764171211371e-05,
      "loss": 1.7493,
      "step": 1280
    },
    {
      "epoch": 0.21317909522825862,
      "grad_norm": 24.351810455322266,
      "learning_rate": 1.065939514129896e-05,
      "loss": 1.7058,
      "step": 1290
    },
    {
      "epoch": 0.21483164635405907,
      "grad_norm": 35.676246643066406,
      "learning_rate": 1.0742026111386548e-05,
      "loss": 1.6535,
      "step": 1300
    },
    {
      "epoch": 0.21648419747985953,
      "grad_norm": 19.702678680419922,
      "learning_rate": 1.0824657081474138e-05,
      "loss": 1.8294,
      "step": 1310
    },
    {
      "epoch": 0.21813674860565999,
      "grad_norm": 29.777185440063477,
      "learning_rate": 1.0907288051561727e-05,
      "loss": 1.7602,
      "step": 1320
    },
    {
      "epoch": 0.21978929973146044,
      "grad_norm": 41.81754684448242,
      "learning_rate": 1.0989919021649315e-05,
      "loss": 1.6949,
      "step": 1330
    },
    {
      "epoch": 0.2214418508572609,
      "grad_norm": 24.266855239868164,
      "learning_rate": 1.1072549991736903e-05,
      "loss": 1.7718,
      "step": 1340
    },
    {
      "epoch": 0.22309440198306135,
      "grad_norm": 36.88713073730469,
      "learning_rate": 1.1155180961824492e-05,
      "loss": 1.7773,
      "step": 1350
    },
    {
      "epoch": 0.2247469531088618,
      "grad_norm": 20.02907943725586,
      "learning_rate": 1.1237811931912082e-05,
      "loss": 1.8017,
      "step": 1360
    },
    {
      "epoch": 0.22639950423466226,
      "grad_norm": 32.76356887817383,
      "learning_rate": 1.132044290199967e-05,
      "loss": 1.6808,
      "step": 1370
    },
    {
      "epoch": 0.22805205536046272,
      "grad_norm": 28.06949234008789,
      "learning_rate": 1.1403073872087259e-05,
      "loss": 1.7669,
      "step": 1380
    },
    {
      "epoch": 0.22970460648626317,
      "grad_norm": 27.715717315673828,
      "learning_rate": 1.1485704842174847e-05,
      "loss": 1.8211,
      "step": 1390
    },
    {
      "epoch": 0.23135715761206363,
      "grad_norm": 14.42801570892334,
      "learning_rate": 1.1568335812262436e-05,
      "loss": 1.7236,
      "step": 1400
    },
    {
      "epoch": 0.23300970873786409,
      "grad_norm": 24.549047470092773,
      "learning_rate": 1.1650966782350026e-05,
      "loss": 1.7445,
      "step": 1410
    },
    {
      "epoch": 0.23466225986366454,
      "grad_norm": 19.45368003845215,
      "learning_rate": 1.1733597752437615e-05,
      "loss": 1.7746,
      "step": 1420
    },
    {
      "epoch": 0.236314810989465,
      "grad_norm": 25.40043830871582,
      "learning_rate": 1.1816228722525203e-05,
      "loss": 1.88,
      "step": 1430
    },
    {
      "epoch": 0.23796736211526545,
      "grad_norm": 80.87067413330078,
      "learning_rate": 1.1898859692612792e-05,
      "loss": 1.7234,
      "step": 1440
    },
    {
      "epoch": 0.2396199132410659,
      "grad_norm": 12.040018081665039,
      "learning_rate": 1.198149066270038e-05,
      "loss": 1.7625,
      "step": 1450
    },
    {
      "epoch": 0.24127246436686636,
      "grad_norm": 16.50065803527832,
      "learning_rate": 1.206412163278797e-05,
      "loss": 1.8548,
      "step": 1460
    },
    {
      "epoch": 0.24292501549266682,
      "grad_norm": 21.08123207092285,
      "learning_rate": 1.2146752602875559e-05,
      "loss": 1.7773,
      "step": 1470
    },
    {
      "epoch": 0.24457756661846725,
      "grad_norm": 13.436989784240723,
      "learning_rate": 1.2229383572963147e-05,
      "loss": 1.6914,
      "step": 1480
    },
    {
      "epoch": 0.2462301177442677,
      "grad_norm": 157.91610717773438,
      "learning_rate": 1.2312014543050736e-05,
      "loss": 1.7556,
      "step": 1490
    },
    {
      "epoch": 0.24788266887006816,
      "grad_norm": 37.71729278564453,
      "learning_rate": 1.2394645513138326e-05,
      "loss": 1.691,
      "step": 1500
    },
    {
      "epoch": 0.2495352199958686,
      "grad_norm": 73.44625854492188,
      "learning_rate": 1.2477276483225914e-05,
      "loss": 1.7093,
      "step": 1510
    },
    {
      "epoch": 0.25118777112166907,
      "grad_norm": 24.017974853515625,
      "learning_rate": 1.2559907453313503e-05,
      "loss": 1.691,
      "step": 1520
    },
    {
      "epoch": 0.25284032224746955,
      "grad_norm": 34.528289794921875,
      "learning_rate": 1.264253842340109e-05,
      "loss": 1.9174,
      "step": 1530
    },
    {
      "epoch": 0.25449287337327,
      "grad_norm": 16.658061981201172,
      "learning_rate": 1.2725169393488681e-05,
      "loss": 1.6134,
      "step": 1540
    },
    {
      "epoch": 0.25614542449907046,
      "grad_norm": 49.6403694152832,
      "learning_rate": 1.280780036357627e-05,
      "loss": 1.7812,
      "step": 1550
    },
    {
      "epoch": 0.2577979756248709,
      "grad_norm": 13.805188179016113,
      "learning_rate": 1.2890431333663858e-05,
      "loss": 1.7681,
      "step": 1560
    },
    {
      "epoch": 0.2594505267506714,
      "grad_norm": 51.3933219909668,
      "learning_rate": 1.2973062303751447e-05,
      "loss": 1.7339,
      "step": 1570
    },
    {
      "epoch": 0.2611030778764718,
      "grad_norm": 12.193310737609863,
      "learning_rate": 1.3055693273839033e-05,
      "loss": 1.8874,
      "step": 1580
    },
    {
      "epoch": 0.2627556290022723,
      "grad_norm": 19.450271606445312,
      "learning_rate": 1.3138324243926625e-05,
      "loss": 1.8667,
      "step": 1590
    },
    {
      "epoch": 0.2644081801280727,
      "grad_norm": 22.680065155029297,
      "learning_rate": 1.3220955214014214e-05,
      "loss": 1.8435,
      "step": 1600
    },
    {
      "epoch": 0.26606073125387314,
      "grad_norm": 43.32194900512695,
      "learning_rate": 1.3303586184101802e-05,
      "loss": 1.685,
      "step": 1610
    },
    {
      "epoch": 0.2677132823796736,
      "grad_norm": 14.092884063720703,
      "learning_rate": 1.338621715418939e-05,
      "loss": 1.7688,
      "step": 1620
    },
    {
      "epoch": 0.26936583350547405,
      "grad_norm": 20.905393600463867,
      "learning_rate": 1.346884812427698e-05,
      "loss": 1.8398,
      "step": 1630
    },
    {
      "epoch": 0.27101838463127453,
      "grad_norm": 12.114250183105469,
      "learning_rate": 1.355147909436457e-05,
      "loss": 1.7383,
      "step": 1640
    },
    {
      "epoch": 0.27267093575707496,
      "grad_norm": 16.64354133605957,
      "learning_rate": 1.3634110064452158e-05,
      "loss": 1.6308,
      "step": 1650
    },
    {
      "epoch": 0.27432348688287544,
      "grad_norm": 18.8261661529541,
      "learning_rate": 1.3716741034539746e-05,
      "loss": 1.7076,
      "step": 1660
    },
    {
      "epoch": 0.2759760380086759,
      "grad_norm": 11.279623031616211,
      "learning_rate": 1.3799372004627335e-05,
      "loss": 1.774,
      "step": 1670
    },
    {
      "epoch": 0.27762858913447636,
      "grad_norm": 12.221219062805176,
      "learning_rate": 1.3882002974714925e-05,
      "loss": 1.5898,
      "step": 1680
    },
    {
      "epoch": 0.2792811402602768,
      "grad_norm": 14.965773582458496,
      "learning_rate": 1.3964633944802513e-05,
      "loss": 1.6842,
      "step": 1690
    },
    {
      "epoch": 0.28093369138607727,
      "grad_norm": 28.31591796875,
      "learning_rate": 1.4047264914890102e-05,
      "loss": 1.6575,
      "step": 1700
    },
    {
      "epoch": 0.2825862425118777,
      "grad_norm": 12.584307670593262,
      "learning_rate": 1.412989588497769e-05,
      "loss": 1.6283,
      "step": 1710
    },
    {
      "epoch": 0.2842387936376782,
      "grad_norm": 30.381515502929688,
      "learning_rate": 1.4212526855065279e-05,
      "loss": 1.7115,
      "step": 1720
    },
    {
      "epoch": 0.2858913447634786,
      "grad_norm": 20.99606704711914,
      "learning_rate": 1.4295157825152869e-05,
      "loss": 1.7297,
      "step": 1730
    },
    {
      "epoch": 0.2875438958892791,
      "grad_norm": 91.88641357421875,
      "learning_rate": 1.4377788795240457e-05,
      "loss": 1.7907,
      "step": 1740
    },
    {
      "epoch": 0.2891964470150795,
      "grad_norm": 14.486876487731934,
      "learning_rate": 1.4460419765328046e-05,
      "loss": 1.763,
      "step": 1750
    },
    {
      "epoch": 0.29084899814088,
      "grad_norm": 32.83678436279297,
      "learning_rate": 1.4543050735415634e-05,
      "loss": 1.8511,
      "step": 1760
    },
    {
      "epoch": 0.2925015492666804,
      "grad_norm": 15.216378211975098,
      "learning_rate": 1.4625681705503224e-05,
      "loss": 1.8471,
      "step": 1770
    },
    {
      "epoch": 0.2941541003924809,
      "grad_norm": 34.238189697265625,
      "learning_rate": 1.4708312675590813e-05,
      "loss": 1.7476,
      "step": 1780
    },
    {
      "epoch": 0.29580665151828134,
      "grad_norm": 26.102210998535156,
      "learning_rate": 1.4790943645678401e-05,
      "loss": 1.5983,
      "step": 1790
    },
    {
      "epoch": 0.2974592026440818,
      "grad_norm": 23.120174407958984,
      "learning_rate": 1.487357461576599e-05,
      "loss": 1.7779,
      "step": 1800
    },
    {
      "epoch": 0.29911175376988225,
      "grad_norm": 14.618244171142578,
      "learning_rate": 1.4956205585853578e-05,
      "loss": 1.6681,
      "step": 1810
    },
    {
      "epoch": 0.30076430489568273,
      "grad_norm": 33.785186767578125,
      "learning_rate": 1.5038836555941168e-05,
      "loss": 1.7867,
      "step": 1820
    },
    {
      "epoch": 0.30241685602148316,
      "grad_norm": 11.777277946472168,
      "learning_rate": 1.5121467526028757e-05,
      "loss": 1.6726,
      "step": 1830
    },
    {
      "epoch": 0.30406940714728364,
      "grad_norm": 15.27955150604248,
      "learning_rate": 1.5204098496116345e-05,
      "loss": 1.7069,
      "step": 1840
    },
    {
      "epoch": 0.30572195827308407,
      "grad_norm": 25.062849044799805,
      "learning_rate": 1.5286729466203932e-05,
      "loss": 1.6681,
      "step": 1850
    },
    {
      "epoch": 0.30737450939888455,
      "grad_norm": 19.023670196533203,
      "learning_rate": 1.5369360436291522e-05,
      "loss": 1.7546,
      "step": 1860
    },
    {
      "epoch": 0.309027060524685,
      "grad_norm": 14.75047492980957,
      "learning_rate": 1.5451991406379112e-05,
      "loss": 1.7295,
      "step": 1870
    },
    {
      "epoch": 0.3106796116504854,
      "grad_norm": 12.37612247467041,
      "learning_rate": 1.5534622376466703e-05,
      "loss": 1.8116,
      "step": 1880
    },
    {
      "epoch": 0.3123321627762859,
      "grad_norm": 26.212312698364258,
      "learning_rate": 1.561725334655429e-05,
      "loss": 1.7477,
      "step": 1890
    },
    {
      "epoch": 0.3139847139020863,
      "grad_norm": 52.385616302490234,
      "learning_rate": 1.5699884316641876e-05,
      "loss": 1.6689,
      "step": 1900
    },
    {
      "epoch": 0.3156372650278868,
      "grad_norm": 15.535853385925293,
      "learning_rate": 1.5782515286729466e-05,
      "loss": 1.7874,
      "step": 1910
    },
    {
      "epoch": 0.31728981615368723,
      "grad_norm": 15.0889892578125,
      "learning_rate": 1.5865146256817056e-05,
      "loss": 1.812,
      "step": 1920
    },
    {
      "epoch": 0.3189423672794877,
      "grad_norm": 14.585269927978516,
      "learning_rate": 1.5947777226904643e-05,
      "loss": 1.7859,
      "step": 1930
    },
    {
      "epoch": 0.32059491840528814,
      "grad_norm": 26.54705810546875,
      "learning_rate": 1.6030408196992233e-05,
      "loss": 1.6718,
      "step": 1940
    },
    {
      "epoch": 0.3222474695310886,
      "grad_norm": 23.145902633666992,
      "learning_rate": 1.611303916707982e-05,
      "loss": 1.7847,
      "step": 1950
    },
    {
      "epoch": 0.32390002065688905,
      "grad_norm": 19.175052642822266,
      "learning_rate": 1.6195670137167414e-05,
      "loss": 1.895,
      "step": 1960
    },
    {
      "epoch": 0.32555257178268954,
      "grad_norm": 11.821075439453125,
      "learning_rate": 1.6278301107255e-05,
      "loss": 1.7336,
      "step": 1970
    },
    {
      "epoch": 0.32720512290848996,
      "grad_norm": 13.310040473937988,
      "learning_rate": 1.6360932077342587e-05,
      "loss": 1.6935,
      "step": 1980
    },
    {
      "epoch": 0.32885767403429045,
      "grad_norm": 20.371007919311523,
      "learning_rate": 1.6443563047430177e-05,
      "loss": 1.6693,
      "step": 1990
    },
    {
      "epoch": 0.3305102251600909,
      "grad_norm": 9.38290786743164,
      "learning_rate": 1.6526194017517764e-05,
      "loss": 1.8035,
      "step": 2000
    },
    {
      "epoch": 0.33216277628589136,
      "grad_norm": 11.759653091430664,
      "learning_rate": 1.6608824987605358e-05,
      "loss": 1.6504,
      "step": 2010
    },
    {
      "epoch": 0.3338153274116918,
      "grad_norm": 9.93994140625,
      "learning_rate": 1.6691455957692945e-05,
      "loss": 1.6363,
      "step": 2020
    },
    {
      "epoch": 0.33546787853749227,
      "grad_norm": 39.654541015625,
      "learning_rate": 1.677408692778053e-05,
      "loss": 1.6831,
      "step": 2030
    },
    {
      "epoch": 0.3371204296632927,
      "grad_norm": 15.763498306274414,
      "learning_rate": 1.685671789786812e-05,
      "loss": 1.7134,
      "step": 2040
    },
    {
      "epoch": 0.3387729807890932,
      "grad_norm": 13.894293785095215,
      "learning_rate": 1.6939348867955708e-05,
      "loss": 1.6449,
      "step": 2050
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 12.600088119506836,
      "learning_rate": 1.7021979838043302e-05,
      "loss": 1.8048,
      "step": 2060
    },
    {
      "epoch": 0.3420780830406941,
      "grad_norm": 49.38233947753906,
      "learning_rate": 1.710461080813089e-05,
      "loss": 1.669,
      "step": 2070
    },
    {
      "epoch": 0.3437306341664945,
      "grad_norm": 123.16156768798828,
      "learning_rate": 1.7187241778218475e-05,
      "loss": 1.6796,
      "step": 2080
    },
    {
      "epoch": 0.345383185292295,
      "grad_norm": 12.5492525100708,
      "learning_rate": 1.7269872748306065e-05,
      "loss": 1.5977,
      "step": 2090
    },
    {
      "epoch": 0.34703573641809543,
      "grad_norm": 27.242778778076172,
      "learning_rate": 1.7352503718393652e-05,
      "loss": 1.8755,
      "step": 2100
    },
    {
      "epoch": 0.3486882875438959,
      "grad_norm": 12.04410457611084,
      "learning_rate": 1.7435134688481246e-05,
      "loss": 1.7807,
      "step": 2110
    },
    {
      "epoch": 0.35034083866969634,
      "grad_norm": 37.36049270629883,
      "learning_rate": 1.7517765658568833e-05,
      "loss": 1.7438,
      "step": 2120
    },
    {
      "epoch": 0.3519933897954968,
      "grad_norm": 19.858129501342773,
      "learning_rate": 1.760039662865642e-05,
      "loss": 1.9511,
      "step": 2130
    },
    {
      "epoch": 0.35364594092129725,
      "grad_norm": 11.481547355651855,
      "learning_rate": 1.768302759874401e-05,
      "loss": 1.7014,
      "step": 2140
    },
    {
      "epoch": 0.3552984920470977,
      "grad_norm": 81.52596282958984,
      "learning_rate": 1.77656585688316e-05,
      "loss": 1.6589,
      "step": 2150
    },
    {
      "epoch": 0.35695104317289816,
      "grad_norm": 17.031217575073242,
      "learning_rate": 1.784828953891919e-05,
      "loss": 1.6914,
      "step": 2160
    },
    {
      "epoch": 0.3586035942986986,
      "grad_norm": 16.88812255859375,
      "learning_rate": 1.7930920509006777e-05,
      "loss": 1.8675,
      "step": 2170
    },
    {
      "epoch": 0.3602561454244991,
      "grad_norm": 10.55776309967041,
      "learning_rate": 1.8013551479094363e-05,
      "loss": 1.7589,
      "step": 2180
    },
    {
      "epoch": 0.3619086965502995,
      "grad_norm": 84.68812561035156,
      "learning_rate": 1.8096182449181954e-05,
      "loss": 1.6879,
      "step": 2190
    },
    {
      "epoch": 0.3635612476761,
      "grad_norm": 18.757328033447266,
      "learning_rate": 1.8178813419269544e-05,
      "loss": 1.7931,
      "step": 2200
    },
    {
      "epoch": 0.3652137988019004,
      "grad_norm": 23.67148780822754,
      "learning_rate": 1.8261444389357134e-05,
      "loss": 1.7531,
      "step": 2210
    },
    {
      "epoch": 0.3668663499277009,
      "grad_norm": 38.11347198486328,
      "learning_rate": 1.834407535944472e-05,
      "loss": 1.6541,
      "step": 2220
    },
    {
      "epoch": 0.3685189010535013,
      "grad_norm": 15.99669075012207,
      "learning_rate": 1.8426706329532307e-05,
      "loss": 1.7447,
      "step": 2230
    },
    {
      "epoch": 0.3701714521793018,
      "grad_norm": 16.732406616210938,
      "learning_rate": 1.8509337299619898e-05,
      "loss": 1.7383,
      "step": 2240
    },
    {
      "epoch": 0.37182400330510224,
      "grad_norm": 36.895389556884766,
      "learning_rate": 1.8591968269707488e-05,
      "loss": 1.7366,
      "step": 2250
    },
    {
      "epoch": 0.3734765544309027,
      "grad_norm": 14.426408767700195,
      "learning_rate": 1.8674599239795078e-05,
      "loss": 1.7979,
      "step": 2260
    },
    {
      "epoch": 0.37512910555670315,
      "grad_norm": 19.50206184387207,
      "learning_rate": 1.8757230209882665e-05,
      "loss": 1.7249,
      "step": 2270
    },
    {
      "epoch": 0.37678165668250363,
      "grad_norm": 33.446937561035156,
      "learning_rate": 1.883986117997025e-05,
      "loss": 1.8677,
      "step": 2280
    },
    {
      "epoch": 0.37843420780830406,
      "grad_norm": 16.420921325683594,
      "learning_rate": 1.8922492150057845e-05,
      "loss": 1.608,
      "step": 2290
    },
    {
      "epoch": 0.38008675893410454,
      "grad_norm": 33.9530029296875,
      "learning_rate": 1.9005123120145432e-05,
      "loss": 1.7731,
      "step": 2300
    },
    {
      "epoch": 0.38173931005990497,
      "grad_norm": 14.426531791687012,
      "learning_rate": 1.9087754090233022e-05,
      "loss": 1.7938,
      "step": 2310
    },
    {
      "epoch": 0.38339186118570545,
      "grad_norm": 18.52570152282715,
      "learning_rate": 1.917038506032061e-05,
      "loss": 1.7116,
      "step": 2320
    },
    {
      "epoch": 0.3850444123115059,
      "grad_norm": 11.881404876708984,
      "learning_rate": 1.9253016030408195e-05,
      "loss": 1.719,
      "step": 2330
    },
    {
      "epoch": 0.38669696343730636,
      "grad_norm": 13.673921585083008,
      "learning_rate": 1.933564700049579e-05,
      "loss": 1.6652,
      "step": 2340
    },
    {
      "epoch": 0.3883495145631068,
      "grad_norm": 12.703744888305664,
      "learning_rate": 1.9418277970583376e-05,
      "loss": 1.5334,
      "step": 2350
    },
    {
      "epoch": 0.3900020656889073,
      "grad_norm": 25.498388290405273,
      "learning_rate": 1.9500908940670966e-05,
      "loss": 1.7857,
      "step": 2360
    },
    {
      "epoch": 0.3916546168147077,
      "grad_norm": 10.993797302246094,
      "learning_rate": 1.9583539910758553e-05,
      "loss": 1.6038,
      "step": 2370
    },
    {
      "epoch": 0.3933071679405082,
      "grad_norm": 17.91847038269043,
      "learning_rate": 1.966617088084614e-05,
      "loss": 1.6546,
      "step": 2380
    },
    {
      "epoch": 0.3949597190663086,
      "grad_norm": 22.401233673095703,
      "learning_rate": 1.9748801850933733e-05,
      "loss": 1.7847,
      "step": 2390
    },
    {
      "epoch": 0.3966122701921091,
      "grad_norm": 20.50080108642578,
      "learning_rate": 1.983143282102132e-05,
      "loss": 1.6773,
      "step": 2400
    },
    {
      "epoch": 0.3982648213179095,
      "grad_norm": 22.017650604248047,
      "learning_rate": 1.991406379110891e-05,
      "loss": 1.7695,
      "step": 2410
    },
    {
      "epoch": 0.39991737244370995,
      "grad_norm": 10.337092399597168,
      "learning_rate": 1.9996694761196497e-05,
      "loss": 1.7029,
      "step": 2420
    },
    {
      "epoch": 0.40156992356951043,
      "grad_norm": 10.79664134979248,
      "learning_rate": 2.0079325731284083e-05,
      "loss": 1.6576,
      "step": 2430
    },
    {
      "epoch": 0.40322247469531086,
      "grad_norm": 33.853580474853516,
      "learning_rate": 2.0161956701371677e-05,
      "loss": 1.7565,
      "step": 2440
    },
    {
      "epoch": 0.40487502582111135,
      "grad_norm": 15.0186767578125,
      "learning_rate": 2.0244587671459264e-05,
      "loss": 1.7022,
      "step": 2450
    },
    {
      "epoch": 0.4065275769469118,
      "grad_norm": 21.504180908203125,
      "learning_rate": 2.0327218641546854e-05,
      "loss": 1.5559,
      "step": 2460
    },
    {
      "epoch": 0.40818012807271226,
      "grad_norm": 15.873148918151855,
      "learning_rate": 2.040984961163444e-05,
      "loss": 1.7129,
      "step": 2470
    },
    {
      "epoch": 0.4098326791985127,
      "grad_norm": 19.291330337524414,
      "learning_rate": 2.049248058172203e-05,
      "loss": 1.7669,
      "step": 2480
    },
    {
      "epoch": 0.41148523032431317,
      "grad_norm": 10.179187774658203,
      "learning_rate": 2.057511155180962e-05,
      "loss": 1.7552,
      "step": 2490
    },
    {
      "epoch": 0.4131377814501136,
      "grad_norm": 46.47639083862305,
      "learning_rate": 2.0657742521897208e-05,
      "loss": 1.757,
      "step": 2500
    },
    {
      "epoch": 0.4147903325759141,
      "grad_norm": 13.199604988098145,
      "learning_rate": 2.0740373491984798e-05,
      "loss": 1.6626,
      "step": 2510
    },
    {
      "epoch": 0.4164428837017145,
      "grad_norm": 9.901959419250488,
      "learning_rate": 2.0823004462072385e-05,
      "loss": 1.6169,
      "step": 2520
    },
    {
      "epoch": 0.418095434827515,
      "grad_norm": 21.13959312438965,
      "learning_rate": 2.0905635432159975e-05,
      "loss": 1.6024,
      "step": 2530
    },
    {
      "epoch": 0.4197479859533154,
      "grad_norm": 7.372568130493164,
      "learning_rate": 2.0988266402247565e-05,
      "loss": 1.7535,
      "step": 2540
    },
    {
      "epoch": 0.4214005370791159,
      "grad_norm": 61.1468620300293,
      "learning_rate": 2.1070897372335152e-05,
      "loss": 1.6967,
      "step": 2550
    },
    {
      "epoch": 0.4230530882049163,
      "grad_norm": 16.36248779296875,
      "learning_rate": 2.1153528342422742e-05,
      "loss": 1.6661,
      "step": 2560
    },
    {
      "epoch": 0.4247056393307168,
      "grad_norm": 13.132911682128906,
      "learning_rate": 2.123615931251033e-05,
      "loss": 1.7467,
      "step": 2570
    },
    {
      "epoch": 0.42635819045651724,
      "grad_norm": 17.8082218170166,
      "learning_rate": 2.131879028259792e-05,
      "loss": 1.7569,
      "step": 2580
    },
    {
      "epoch": 0.4280107415823177,
      "grad_norm": 48.110652923583984,
      "learning_rate": 2.140142125268551e-05,
      "loss": 1.6307,
      "step": 2590
    },
    {
      "epoch": 0.42966329270811815,
      "grad_norm": 32.73516082763672,
      "learning_rate": 2.1484052222773096e-05,
      "loss": 1.7427,
      "step": 2600
    },
    {
      "epoch": 0.43131584383391863,
      "grad_norm": 29.785804748535156,
      "learning_rate": 2.1566683192860686e-05,
      "loss": 1.8084,
      "step": 2610
    },
    {
      "epoch": 0.43296839495971906,
      "grad_norm": 21.294347763061523,
      "learning_rate": 2.1649314162948276e-05,
      "loss": 1.8063,
      "step": 2620
    },
    {
      "epoch": 0.43462094608551954,
      "grad_norm": 10.334474563598633,
      "learning_rate": 2.1731945133035863e-05,
      "loss": 1.7292,
      "step": 2630
    },
    {
      "epoch": 0.43627349721131997,
      "grad_norm": 37.316349029541016,
      "learning_rate": 2.1814576103123453e-05,
      "loss": 1.7178,
      "step": 2640
    },
    {
      "epoch": 0.43792604833712045,
      "grad_norm": 15.207347869873047,
      "learning_rate": 2.189720707321104e-05,
      "loss": 1.6157,
      "step": 2650
    },
    {
      "epoch": 0.4395785994629209,
      "grad_norm": 44.354366302490234,
      "learning_rate": 2.197983804329863e-05,
      "loss": 1.6401,
      "step": 2660
    },
    {
      "epoch": 0.44123115058872137,
      "grad_norm": 13.655917167663574,
      "learning_rate": 2.206246901338622e-05,
      "loss": 1.6388,
      "step": 2670
    },
    {
      "epoch": 0.4428837017145218,
      "grad_norm": 25.89215660095215,
      "learning_rate": 2.2145099983473807e-05,
      "loss": 1.7077,
      "step": 2680
    },
    {
      "epoch": 0.4445362528403222,
      "grad_norm": 9.225777626037598,
      "learning_rate": 2.2227730953561397e-05,
      "loss": 1.7651,
      "step": 2690
    },
    {
      "epoch": 0.4461888039661227,
      "grad_norm": 21.08796501159668,
      "learning_rate": 2.2310361923648984e-05,
      "loss": 1.6447,
      "step": 2700
    },
    {
      "epoch": 0.44784135509192313,
      "grad_norm": 16.045692443847656,
      "learning_rate": 2.2392992893736574e-05,
      "loss": 1.6521,
      "step": 2710
    },
    {
      "epoch": 0.4494939062177236,
      "grad_norm": 16.895357131958008,
      "learning_rate": 2.2475623863824164e-05,
      "loss": 1.5923,
      "step": 2720
    },
    {
      "epoch": 0.45114645734352404,
      "grad_norm": 19.075815200805664,
      "learning_rate": 2.255825483391175e-05,
      "loss": 1.6618,
      "step": 2730
    },
    {
      "epoch": 0.4527990084693245,
      "grad_norm": 12.131684303283691,
      "learning_rate": 2.264088580399934e-05,
      "loss": 1.7285,
      "step": 2740
    },
    {
      "epoch": 0.45445155959512495,
      "grad_norm": 60.209259033203125,
      "learning_rate": 2.2723516774086928e-05,
      "loss": 1.6896,
      "step": 2750
    },
    {
      "epoch": 0.45610411072092544,
      "grad_norm": 22.111814498901367,
      "learning_rate": 2.2806147744174518e-05,
      "loss": 1.7719,
      "step": 2760
    },
    {
      "epoch": 0.45775666184672587,
      "grad_norm": 9.202242851257324,
      "learning_rate": 2.2888778714262108e-05,
      "loss": 1.7777,
      "step": 2770
    },
    {
      "epoch": 0.45940921297252635,
      "grad_norm": 63.52800750732422,
      "learning_rate": 2.2971409684349695e-05,
      "loss": 1.6549,
      "step": 2780
    },
    {
      "epoch": 0.4610617640983268,
      "grad_norm": 12.614968299865723,
      "learning_rate": 2.3054040654437285e-05,
      "loss": 1.8158,
      "step": 2790
    },
    {
      "epoch": 0.46271431522412726,
      "grad_norm": 10.92754077911377,
      "learning_rate": 2.3136671624524872e-05,
      "loss": 1.6745,
      "step": 2800
    },
    {
      "epoch": 0.4643668663499277,
      "grad_norm": 57.359619140625,
      "learning_rate": 2.3219302594612462e-05,
      "loss": 1.7044,
      "step": 2810
    },
    {
      "epoch": 0.46601941747572817,
      "grad_norm": 19.017126083374023,
      "learning_rate": 2.3301933564700052e-05,
      "loss": 1.6981,
      "step": 2820
    },
    {
      "epoch": 0.4676719686015286,
      "grad_norm": 26.398160934448242,
      "learning_rate": 2.338456453478764e-05,
      "loss": 1.7188,
      "step": 2830
    },
    {
      "epoch": 0.4693245197273291,
      "grad_norm": 13.000535011291504,
      "learning_rate": 2.346719550487523e-05,
      "loss": 1.6572,
      "step": 2840
    },
    {
      "epoch": 0.4709770708531295,
      "grad_norm": 18.722341537475586,
      "learning_rate": 2.3549826474962816e-05,
      "loss": 1.8101,
      "step": 2850
    },
    {
      "epoch": 0.47262962197893,
      "grad_norm": 94.03982543945312,
      "learning_rate": 2.3632457445050406e-05,
      "loss": 1.5997,
      "step": 2860
    },
    {
      "epoch": 0.4742821731047304,
      "grad_norm": 45.00182342529297,
      "learning_rate": 2.3715088415137996e-05,
      "loss": 1.6486,
      "step": 2870
    },
    {
      "epoch": 0.4759347242305309,
      "grad_norm": 26.30855941772461,
      "learning_rate": 2.3797719385225583e-05,
      "loss": 1.7285,
      "step": 2880
    },
    {
      "epoch": 0.47758727535633133,
      "grad_norm": 127.1923599243164,
      "learning_rate": 2.3880350355313173e-05,
      "loss": 1.7588,
      "step": 2890
    },
    {
      "epoch": 0.4792398264821318,
      "grad_norm": 22.71497917175293,
      "learning_rate": 2.396298132540076e-05,
      "loss": 1.6556,
      "step": 2900
    },
    {
      "epoch": 0.48089237760793224,
      "grad_norm": 17.4209041595459,
      "learning_rate": 2.404561229548835e-05,
      "loss": 1.6399,
      "step": 2910
    },
    {
      "epoch": 0.4825449287337327,
      "grad_norm": 38.692264556884766,
      "learning_rate": 2.412824326557594e-05,
      "loss": 1.6147,
      "step": 2920
    },
    {
      "epoch": 0.48419747985953315,
      "grad_norm": 26.006572723388672,
      "learning_rate": 2.4210874235663527e-05,
      "loss": 1.6381,
      "step": 2930
    },
    {
      "epoch": 0.48585003098533364,
      "grad_norm": 13.967598915100098,
      "learning_rate": 2.4293505205751117e-05,
      "loss": 1.6556,
      "step": 2940
    },
    {
      "epoch": 0.48750258211113406,
      "grad_norm": 15.196572303771973,
      "learning_rate": 2.4376136175838707e-05,
      "loss": 1.6714,
      "step": 2950
    },
    {
      "epoch": 0.4891551332369345,
      "grad_norm": 23.98309326171875,
      "learning_rate": 2.4458767145926294e-05,
      "loss": 1.7578,
      "step": 2960
    },
    {
      "epoch": 0.490807684362735,
      "grad_norm": 19.997920989990234,
      "learning_rate": 2.4541398116013884e-05,
      "loss": 1.7737,
      "step": 2970
    },
    {
      "epoch": 0.4924602354885354,
      "grad_norm": 26.095102310180664,
      "learning_rate": 2.462402908610147e-05,
      "loss": 1.709,
      "step": 2980
    },
    {
      "epoch": 0.4941127866143359,
      "grad_norm": 60.75287628173828,
      "learning_rate": 2.470666005618906e-05,
      "loss": 1.7374,
      "step": 2990
    },
    {
      "epoch": 0.4957653377401363,
      "grad_norm": 12.193611145019531,
      "learning_rate": 2.478929102627665e-05,
      "loss": 1.7192,
      "step": 3000
    },
    {
      "epoch": 0.4974178888659368,
      "grad_norm": 21.721214294433594,
      "learning_rate": 2.4871921996364238e-05,
      "loss": 1.6808,
      "step": 3010
    },
    {
      "epoch": 0.4990704399917372,
      "grad_norm": 9.847155570983887,
      "learning_rate": 2.4954552966451828e-05,
      "loss": 1.6721,
      "step": 3020
    },
    {
      "epoch": 0.5007229911175377,
      "grad_norm": 12.325818061828613,
      "learning_rate": 2.5037183936539415e-05,
      "loss": 1.8163,
      "step": 3030
    },
    {
      "epoch": 0.5023755422433381,
      "grad_norm": 33.089630126953125,
      "learning_rate": 2.5119814906627005e-05,
      "loss": 1.8137,
      "step": 3040
    },
    {
      "epoch": 0.5040280933691386,
      "grad_norm": 47.662208557128906,
      "learning_rate": 2.5202445876714592e-05,
      "loss": 1.7631,
      "step": 3050
    },
    {
      "epoch": 0.5056806444949391,
      "grad_norm": 11.936186790466309,
      "learning_rate": 2.528507684680218e-05,
      "loss": 1.6799,
      "step": 3060
    },
    {
      "epoch": 0.5073331956207395,
      "grad_norm": 24.59088706970215,
      "learning_rate": 2.5367707816889776e-05,
      "loss": 1.7288,
      "step": 3070
    },
    {
      "epoch": 0.50898574674654,
      "grad_norm": 12.187572479248047,
      "learning_rate": 2.5450338786977362e-05,
      "loss": 1.7636,
      "step": 3080
    },
    {
      "epoch": 0.5106382978723404,
      "grad_norm": 12.487037658691406,
      "learning_rate": 2.553296975706495e-05,
      "loss": 1.6933,
      "step": 3090
    },
    {
      "epoch": 0.5122908489981409,
      "grad_norm": 16.490032196044922,
      "learning_rate": 2.561560072715254e-05,
      "loss": 1.8374,
      "step": 3100
    },
    {
      "epoch": 0.5139434001239414,
      "grad_norm": 17.2027530670166,
      "learning_rate": 2.5698231697240126e-05,
      "loss": 1.6313,
      "step": 3110
    },
    {
      "epoch": 0.5155959512497418,
      "grad_norm": 67.94872283935547,
      "learning_rate": 2.5780862667327716e-05,
      "loss": 1.6479,
      "step": 3120
    },
    {
      "epoch": 0.5172485023755422,
      "grad_norm": 17.46609878540039,
      "learning_rate": 2.5863493637415303e-05,
      "loss": 1.7751,
      "step": 3130
    },
    {
      "epoch": 0.5189010535013427,
      "grad_norm": 13.204253196716309,
      "learning_rate": 2.5946124607502893e-05,
      "loss": 1.6845,
      "step": 3140
    },
    {
      "epoch": 0.5205536046271432,
      "grad_norm": 10.950384140014648,
      "learning_rate": 2.602875557759048e-05,
      "loss": 1.6612,
      "step": 3150
    },
    {
      "epoch": 0.5222061557529436,
      "grad_norm": 25.03174591064453,
      "learning_rate": 2.6111386547678067e-05,
      "loss": 1.7298,
      "step": 3160
    },
    {
      "epoch": 0.523858706878744,
      "grad_norm": 41.82197952270508,
      "learning_rate": 2.6194017517765664e-05,
      "loss": 1.7301,
      "step": 3170
    },
    {
      "epoch": 0.5255112580045446,
      "grad_norm": 26.862686157226562,
      "learning_rate": 2.627664848785325e-05,
      "loss": 1.6881,
      "step": 3180
    },
    {
      "epoch": 0.527163809130345,
      "grad_norm": 13.66534423828125,
      "learning_rate": 2.6359279457940837e-05,
      "loss": 1.6658,
      "step": 3190
    },
    {
      "epoch": 0.5288163602561454,
      "grad_norm": 17.450504302978516,
      "learning_rate": 2.6441910428028427e-05,
      "loss": 1.6795,
      "step": 3200
    },
    {
      "epoch": 0.5304689113819459,
      "grad_norm": 12.588276863098145,
      "learning_rate": 2.6524541398116014e-05,
      "loss": 1.6239,
      "step": 3210
    },
    {
      "epoch": 0.5321214625077463,
      "grad_norm": 34.62938690185547,
      "learning_rate": 2.6607172368203604e-05,
      "loss": 1.5944,
      "step": 3220
    },
    {
      "epoch": 0.5337740136335468,
      "grad_norm": 22.4811954498291,
      "learning_rate": 2.668980333829119e-05,
      "loss": 1.6605,
      "step": 3230
    },
    {
      "epoch": 0.5354265647593472,
      "grad_norm": 13.54268741607666,
      "learning_rate": 2.677243430837878e-05,
      "loss": 1.6363,
      "step": 3240
    },
    {
      "epoch": 0.5370791158851477,
      "grad_norm": 13.956992149353027,
      "learning_rate": 2.6855065278466368e-05,
      "loss": 1.6435,
      "step": 3250
    },
    {
      "epoch": 0.5387316670109481,
      "grad_norm": 23.956907272338867,
      "learning_rate": 2.693769624855396e-05,
      "loss": 1.6956,
      "step": 3260
    },
    {
      "epoch": 0.5403842181367486,
      "grad_norm": 16.56192398071289,
      "learning_rate": 2.7020327218641552e-05,
      "loss": 1.6613,
      "step": 3270
    },
    {
      "epoch": 0.5420367692625491,
      "grad_norm": 21.902639389038086,
      "learning_rate": 2.710295818872914e-05,
      "loss": 1.7387,
      "step": 3280
    },
    {
      "epoch": 0.5436893203883495,
      "grad_norm": 24.28936004638672,
      "learning_rate": 2.7185589158816725e-05,
      "loss": 1.8334,
      "step": 3290
    },
    {
      "epoch": 0.5453418715141499,
      "grad_norm": 6.949965476989746,
      "learning_rate": 2.7268220128904315e-05,
      "loss": 1.6161,
      "step": 3300
    },
    {
      "epoch": 0.5469944226399505,
      "grad_norm": 10.36733627319336,
      "learning_rate": 2.7350851098991902e-05,
      "loss": 1.6539,
      "step": 3310
    },
    {
      "epoch": 0.5486469737657509,
      "grad_norm": 15.764904975891113,
      "learning_rate": 2.7433482069079492e-05,
      "loss": 1.7367,
      "step": 3320
    },
    {
      "epoch": 0.5502995248915513,
      "grad_norm": 30.997188568115234,
      "learning_rate": 2.751611303916708e-05,
      "loss": 1.7039,
      "step": 3330
    },
    {
      "epoch": 0.5519520760173517,
      "grad_norm": 63.2742919921875,
      "learning_rate": 2.759874400925467e-05,
      "loss": 1.6414,
      "step": 3340
    },
    {
      "epoch": 0.5536046271431523,
      "grad_norm": 13.54851245880127,
      "learning_rate": 2.7681374979342256e-05,
      "loss": 1.741,
      "step": 3350
    },
    {
      "epoch": 0.5552571782689527,
      "grad_norm": 31.921226501464844,
      "learning_rate": 2.776400594942985e-05,
      "loss": 1.65,
      "step": 3360
    },
    {
      "epoch": 0.5569097293947531,
      "grad_norm": 20.03881072998047,
      "learning_rate": 2.7846636919517436e-05,
      "loss": 1.6868,
      "step": 3370
    },
    {
      "epoch": 0.5585622805205536,
      "grad_norm": 11.299986839294434,
      "learning_rate": 2.7929267889605027e-05,
      "loss": 1.7552,
      "step": 3380
    },
    {
      "epoch": 0.5602148316463541,
      "grad_norm": 11.070747375488281,
      "learning_rate": 2.8011898859692613e-05,
      "loss": 1.6897,
      "step": 3390
    },
    {
      "epoch": 0.5618673827721545,
      "grad_norm": 8.979963302612305,
      "learning_rate": 2.8094529829780204e-05,
      "loss": 1.6199,
      "step": 3400
    },
    {
      "epoch": 0.563519933897955,
      "grad_norm": 11.309082984924316,
      "learning_rate": 2.817716079986779e-05,
      "loss": 1.8134,
      "step": 3410
    },
    {
      "epoch": 0.5651724850237554,
      "grad_norm": 17.12977409362793,
      "learning_rate": 2.825979176995538e-05,
      "loss": 1.6499,
      "step": 3420
    },
    {
      "epoch": 0.5668250361495559,
      "grad_norm": 6.682351589202881,
      "learning_rate": 2.8342422740042967e-05,
      "loss": 1.6524,
      "step": 3430
    },
    {
      "epoch": 0.5684775872753564,
      "grad_norm": 20.999141693115234,
      "learning_rate": 2.8425053710130557e-05,
      "loss": 1.6885,
      "step": 3440
    },
    {
      "epoch": 0.5701301384011568,
      "grad_norm": 10.989826202392578,
      "learning_rate": 2.850768468021815e-05,
      "loss": 1.5895,
      "step": 3450
    },
    {
      "epoch": 0.5717826895269572,
      "grad_norm": 11.188637733459473,
      "learning_rate": 2.8590315650305738e-05,
      "loss": 1.7236,
      "step": 3460
    },
    {
      "epoch": 0.5734352406527577,
      "grad_norm": 11.868768692016602,
      "learning_rate": 2.8672946620393324e-05,
      "loss": 1.7752,
      "step": 3470
    },
    {
      "epoch": 0.5750877917785582,
      "grad_norm": 13.934696197509766,
      "learning_rate": 2.8755577590480915e-05,
      "loss": 1.6321,
      "step": 3480
    },
    {
      "epoch": 0.5767403429043586,
      "grad_norm": 10.623023986816406,
      "learning_rate": 2.88382085605685e-05,
      "loss": 1.7567,
      "step": 3490
    },
    {
      "epoch": 0.578392894030159,
      "grad_norm": 21.444114685058594,
      "learning_rate": 2.892083953065609e-05,
      "loss": 1.7353,
      "step": 3500
    },
    {
      "epoch": 0.5800454451559595,
      "grad_norm": 22.53701400756836,
      "learning_rate": 2.900347050074368e-05,
      "loss": 1.6643,
      "step": 3510
    },
    {
      "epoch": 0.58169799628176,
      "grad_norm": 8.35075855255127,
      "learning_rate": 2.908610147083127e-05,
      "loss": 1.6099,
      "step": 3520
    },
    {
      "epoch": 0.5833505474075604,
      "grad_norm": 10.763632774353027,
      "learning_rate": 2.9168732440918855e-05,
      "loss": 1.7619,
      "step": 3530
    },
    {
      "epoch": 0.5850030985333609,
      "grad_norm": 10.737512588500977,
      "learning_rate": 2.925136341100645e-05,
      "loss": 1.6345,
      "step": 3540
    },
    {
      "epoch": 0.5866556496591613,
      "grad_norm": 15.459436416625977,
      "learning_rate": 2.933399438109404e-05,
      "loss": 1.6186,
      "step": 3550
    },
    {
      "epoch": 0.5883082007849618,
      "grad_norm": 16.63528823852539,
      "learning_rate": 2.9416625351181626e-05,
      "loss": 1.7349,
      "step": 3560
    },
    {
      "epoch": 0.5899607519107622,
      "grad_norm": 12.49666690826416,
      "learning_rate": 2.9499256321269213e-05,
      "loss": 1.7992,
      "step": 3570
    },
    {
      "epoch": 0.5916133030365627,
      "grad_norm": 30.694421768188477,
      "learning_rate": 2.9581887291356803e-05,
      "loss": 1.7004,
      "step": 3580
    },
    {
      "epoch": 0.5932658541623631,
      "grad_norm": 12.530024528503418,
      "learning_rate": 2.966451826144439e-05,
      "loss": 1.5653,
      "step": 3590
    },
    {
      "epoch": 0.5949184052881636,
      "grad_norm": 10.325366020202637,
      "learning_rate": 2.974714923153198e-05,
      "loss": 1.6872,
      "step": 3600
    },
    {
      "epoch": 0.5965709564139641,
      "grad_norm": 17.523683547973633,
      "learning_rate": 2.9829780201619566e-05,
      "loss": 1.5351,
      "step": 3610
    },
    {
      "epoch": 0.5982235075397645,
      "grad_norm": 12.359028816223145,
      "learning_rate": 2.9912411171707157e-05,
      "loss": 1.6894,
      "step": 3620
    },
    {
      "epoch": 0.5998760586655649,
      "grad_norm": 9.718979835510254,
      "learning_rate": 2.9995042141794743e-05,
      "loss": 1.4912,
      "step": 3630
    },
    {
      "epoch": 0.6015286097913655,
      "grad_norm": 12.050354957580566,
      "learning_rate": 3.0077673111882337e-05,
      "loss": 1.6147,
      "step": 3640
    },
    {
      "epoch": 0.6031811609171659,
      "grad_norm": 73.88978576660156,
      "learning_rate": 3.0160304081969927e-05,
      "loss": 1.639,
      "step": 3650
    },
    {
      "epoch": 0.6048337120429663,
      "grad_norm": 12.085061073303223,
      "learning_rate": 3.0242935052057514e-05,
      "loss": 1.7605,
      "step": 3660
    },
    {
      "epoch": 0.6064862631687667,
      "grad_norm": 12.490886688232422,
      "learning_rate": 3.03255660221451e-05,
      "loss": 1.6698,
      "step": 3670
    },
    {
      "epoch": 0.6081388142945673,
      "grad_norm": 11.048460006713867,
      "learning_rate": 3.040819699223269e-05,
      "loss": 1.7394,
      "step": 3680
    },
    {
      "epoch": 0.6097913654203677,
      "grad_norm": 9.499938011169434,
      "learning_rate": 3.0490827962320277e-05,
      "loss": 1.8048,
      "step": 3690
    },
    {
      "epoch": 0.6114439165461681,
      "grad_norm": 17.809703826904297,
      "learning_rate": 3.0573458932407864e-05,
      "loss": 1.768,
      "step": 3700
    },
    {
      "epoch": 0.6130964676719686,
      "grad_norm": 45.0641975402832,
      "learning_rate": 3.0656089902495454e-05,
      "loss": 1.7266,
      "step": 3710
    },
    {
      "epoch": 0.6147490187977691,
      "grad_norm": 37.341739654541016,
      "learning_rate": 3.0738720872583045e-05,
      "loss": 1.7567,
      "step": 3720
    },
    {
      "epoch": 0.6164015699235695,
      "grad_norm": 8.673922538757324,
      "learning_rate": 3.0821351842670635e-05,
      "loss": 1.8379,
      "step": 3730
    },
    {
      "epoch": 0.61805412104937,
      "grad_norm": 8.250535011291504,
      "learning_rate": 3.0903982812758225e-05,
      "loss": 1.7141,
      "step": 3740
    },
    {
      "epoch": 0.6197066721751704,
      "grad_norm": 22.169246673583984,
      "learning_rate": 3.0986613782845815e-05,
      "loss": 1.6403,
      "step": 3750
    },
    {
      "epoch": 0.6213592233009708,
      "grad_norm": 8.162503242492676,
      "learning_rate": 3.1069244752933405e-05,
      "loss": 1.7723,
      "step": 3760
    },
    {
      "epoch": 0.6230117744267714,
      "grad_norm": 20.341426849365234,
      "learning_rate": 3.115187572302099e-05,
      "loss": 1.7345,
      "step": 3770
    },
    {
      "epoch": 0.6246643255525718,
      "grad_norm": 10.553240776062012,
      "learning_rate": 3.123450669310858e-05,
      "loss": 1.8238,
      "step": 3780
    },
    {
      "epoch": 0.6263168766783722,
      "grad_norm": 15.004631042480469,
      "learning_rate": 3.131713766319617e-05,
      "loss": 1.5615,
      "step": 3790
    },
    {
      "epoch": 0.6279694278041726,
      "grad_norm": 9.84469223022461,
      "learning_rate": 3.139976863328375e-05,
      "loss": 1.724,
      "step": 3800
    },
    {
      "epoch": 0.6296219789299732,
      "grad_norm": 10.025922775268555,
      "learning_rate": 3.148239960337134e-05,
      "loss": 1.729,
      "step": 3810
    },
    {
      "epoch": 0.6312745300557736,
      "grad_norm": 11.77392292022705,
      "learning_rate": 3.156503057345893e-05,
      "loss": 1.7107,
      "step": 3820
    },
    {
      "epoch": 0.632927081181574,
      "grad_norm": 10.818305969238281,
      "learning_rate": 3.164766154354652e-05,
      "loss": 1.5897,
      "step": 3830
    },
    {
      "epoch": 0.6345796323073745,
      "grad_norm": 10.746159553527832,
      "learning_rate": 3.173029251363411e-05,
      "loss": 1.7638,
      "step": 3840
    },
    {
      "epoch": 0.636232183433175,
      "grad_norm": 13.713776588439941,
      "learning_rate": 3.18129234837217e-05,
      "loss": 1.7332,
      "step": 3850
    },
    {
      "epoch": 0.6378847345589754,
      "grad_norm": 10.594388008117676,
      "learning_rate": 3.1895554453809286e-05,
      "loss": 1.8471,
      "step": 3860
    },
    {
      "epoch": 0.6395372856847759,
      "grad_norm": 16.44990348815918,
      "learning_rate": 3.197818542389688e-05,
      "loss": 1.7023,
      "step": 3870
    },
    {
      "epoch": 0.6411898368105763,
      "grad_norm": 9.56428050994873,
      "learning_rate": 3.206081639398447e-05,
      "loss": 1.7214,
      "step": 3880
    },
    {
      "epoch": 0.6428423879363768,
      "grad_norm": 13.247419357299805,
      "learning_rate": 3.214344736407206e-05,
      "loss": 1.6734,
      "step": 3890
    },
    {
      "epoch": 0.6444949390621773,
      "grad_norm": 13.604035377502441,
      "learning_rate": 3.222607833415964e-05,
      "loss": 1.6755,
      "step": 3900
    },
    {
      "epoch": 0.6461474901879777,
      "grad_norm": 16.204240798950195,
      "learning_rate": 3.230870930424723e-05,
      "loss": 1.7442,
      "step": 3910
    },
    {
      "epoch": 0.6478000413137781,
      "grad_norm": 22.446441650390625,
      "learning_rate": 3.239134027433483e-05,
      "loss": 1.7222,
      "step": 3920
    },
    {
      "epoch": 0.6494525924395786,
      "grad_norm": 13.957686424255371,
      "learning_rate": 3.247397124442241e-05,
      "loss": 1.6194,
      "step": 3930
    },
    {
      "epoch": 0.6511051435653791,
      "grad_norm": 17.994192123413086,
      "learning_rate": 3.255660221451e-05,
      "loss": 1.7454,
      "step": 3940
    },
    {
      "epoch": 0.6527576946911795,
      "grad_norm": 10.868365287780762,
      "learning_rate": 3.263923318459759e-05,
      "loss": 1.707,
      "step": 3950
    },
    {
      "epoch": 0.6544102458169799,
      "grad_norm": 8.294291496276855,
      "learning_rate": 3.2721864154685175e-05,
      "loss": 1.7903,
      "step": 3960
    },
    {
      "epoch": 0.6560627969427805,
      "grad_norm": 10.930803298950195,
      "learning_rate": 3.2804495124772765e-05,
      "loss": 1.6172,
      "step": 3970
    },
    {
      "epoch": 0.6577153480685809,
      "grad_norm": 74.83100128173828,
      "learning_rate": 3.2887126094860355e-05,
      "loss": 1.7242,
      "step": 3980
    },
    {
      "epoch": 0.6593678991943813,
      "grad_norm": 10.721229553222656,
      "learning_rate": 3.2969757064947945e-05,
      "loss": 1.678,
      "step": 3990
    },
    {
      "epoch": 0.6610204503201818,
      "grad_norm": 22.046354293823242,
      "learning_rate": 3.305238803503553e-05,
      "loss": 1.5762,
      "step": 4000
    },
    {
      "epoch": 0.6626730014459823,
      "grad_norm": 8.417013168334961,
      "learning_rate": 3.313501900512312e-05,
      "loss": 1.6341,
      "step": 4010
    },
    {
      "epoch": 0.6643255525717827,
      "grad_norm": 9.018980026245117,
      "learning_rate": 3.3217649975210715e-05,
      "loss": 1.5761,
      "step": 4020
    },
    {
      "epoch": 0.6659781036975831,
      "grad_norm": 19.58799171447754,
      "learning_rate": 3.33002809452983e-05,
      "loss": 1.6891,
      "step": 4030
    },
    {
      "epoch": 0.6676306548233836,
      "grad_norm": 13.492208480834961,
      "learning_rate": 3.338291191538589e-05,
      "loss": 1.6399,
      "step": 4040
    },
    {
      "epoch": 0.669283205949184,
      "grad_norm": 10.820428848266602,
      "learning_rate": 3.346554288547348e-05,
      "loss": 1.7323,
      "step": 4050
    },
    {
      "epoch": 0.6709357570749845,
      "grad_norm": 24.102577209472656,
      "learning_rate": 3.354817385556106e-05,
      "loss": 1.7554,
      "step": 4060
    },
    {
      "epoch": 0.672588308200785,
      "grad_norm": 10.357276916503906,
      "learning_rate": 3.363080482564865e-05,
      "loss": 1.7642,
      "step": 4070
    },
    {
      "epoch": 0.6742408593265854,
      "grad_norm": 16.072227478027344,
      "learning_rate": 3.371343579573624e-05,
      "loss": 1.6417,
      "step": 4080
    },
    {
      "epoch": 0.6758934104523858,
      "grad_norm": 12.857048034667969,
      "learning_rate": 3.379606676582383e-05,
      "loss": 1.7107,
      "step": 4090
    },
    {
      "epoch": 0.6775459615781864,
      "grad_norm": 10.534919738769531,
      "learning_rate": 3.3878697735911416e-05,
      "loss": 1.7853,
      "step": 4100
    },
    {
      "epoch": 0.6791985127039868,
      "grad_norm": 22.40790367126465,
      "learning_rate": 3.396132870599901e-05,
      "loss": 1.7863,
      "step": 4110
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 19.036958694458008,
      "learning_rate": 3.4043959676086604e-05,
      "loss": 1.5541,
      "step": 4120
    },
    {
      "epoch": 0.6825036149555876,
      "grad_norm": 12.263296127319336,
      "learning_rate": 3.412659064617419e-05,
      "loss": 1.7194,
      "step": 4130
    },
    {
      "epoch": 0.6841561660813882,
      "grad_norm": 13.276509284973145,
      "learning_rate": 3.420922161626178e-05,
      "loss": 1.7551,
      "step": 4140
    },
    {
      "epoch": 0.6858087172071886,
      "grad_norm": 21.00751304626465,
      "learning_rate": 3.429185258634937e-05,
      "loss": 1.7179,
      "step": 4150
    },
    {
      "epoch": 0.687461268332989,
      "grad_norm": 27.047283172607422,
      "learning_rate": 3.437448355643695e-05,
      "loss": 1.7047,
      "step": 4160
    },
    {
      "epoch": 0.6891138194587895,
      "grad_norm": 12.669699668884277,
      "learning_rate": 3.445711452652454e-05,
      "loss": 1.6317,
      "step": 4170
    },
    {
      "epoch": 0.69076637058459,
      "grad_norm": 16.105510711669922,
      "learning_rate": 3.453974549661213e-05,
      "loss": 1.6754,
      "step": 4180
    },
    {
      "epoch": 0.6924189217103904,
      "grad_norm": 9.046321868896484,
      "learning_rate": 3.462237646669972e-05,
      "loss": 1.5878,
      "step": 4190
    },
    {
      "epoch": 0.6940714728361909,
      "grad_norm": 18.94280433654785,
      "learning_rate": 3.4705007436787304e-05,
      "loss": 1.6219,
      "step": 4200
    },
    {
      "epoch": 0.6957240239619913,
      "grad_norm": 11.86888313293457,
      "learning_rate": 3.47876384068749e-05,
      "loss": 1.867,
      "step": 4210
    },
    {
      "epoch": 0.6973765750877918,
      "grad_norm": 11.21365737915039,
      "learning_rate": 3.487026937696249e-05,
      "loss": 1.7011,
      "step": 4220
    },
    {
      "epoch": 0.6990291262135923,
      "grad_norm": 36.31560134887695,
      "learning_rate": 3.4952900347050075e-05,
      "loss": 1.6172,
      "step": 4230
    },
    {
      "epoch": 0.7006816773393927,
      "grad_norm": 7.454902648925781,
      "learning_rate": 3.5035531317137665e-05,
      "loss": 1.682,
      "step": 4240
    },
    {
      "epoch": 0.7023342284651931,
      "grad_norm": 8.024889945983887,
      "learning_rate": 3.5118162287225255e-05,
      "loss": 1.6474,
      "step": 4250
    },
    {
      "epoch": 0.7039867795909936,
      "grad_norm": 11.65849494934082,
      "learning_rate": 3.520079325731284e-05,
      "loss": 1.6873,
      "step": 4260
    },
    {
      "epoch": 0.7056393307167941,
      "grad_norm": 9.83031940460205,
      "learning_rate": 3.528342422740043e-05,
      "loss": 1.6696,
      "step": 4270
    },
    {
      "epoch": 0.7072918818425945,
      "grad_norm": 27.40468978881836,
      "learning_rate": 3.536605519748802e-05,
      "loss": 1.6058,
      "step": 4280
    },
    {
      "epoch": 0.7089444329683949,
      "grad_norm": 7.991322994232178,
      "learning_rate": 3.544868616757561e-05,
      "loss": 1.6647,
      "step": 4290
    },
    {
      "epoch": 0.7105969840941954,
      "grad_norm": 17.9750919342041,
      "learning_rate": 3.55313171376632e-05,
      "loss": 1.7095,
      "step": 4300
    },
    {
      "epoch": 0.7122495352199959,
      "grad_norm": 33.60334777832031,
      "learning_rate": 3.561394810775079e-05,
      "loss": 1.6818,
      "step": 4310
    },
    {
      "epoch": 0.7139020863457963,
      "grad_norm": 12.004261016845703,
      "learning_rate": 3.569657907783838e-05,
      "loss": 1.6787,
      "step": 4320
    },
    {
      "epoch": 0.7155546374715968,
      "grad_norm": 8.619003295898438,
      "learning_rate": 3.577921004792596e-05,
      "loss": 1.6197,
      "step": 4330
    },
    {
      "epoch": 0.7172071885973972,
      "grad_norm": 11.942529678344727,
      "learning_rate": 3.586184101801355e-05,
      "loss": 1.6431,
      "step": 4340
    },
    {
      "epoch": 0.7188597397231977,
      "grad_norm": 8.107340812683105,
      "learning_rate": 3.594447198810114e-05,
      "loss": 1.6827,
      "step": 4350
    },
    {
      "epoch": 0.7205122908489981,
      "grad_norm": 11.563468933105469,
      "learning_rate": 3.602710295818873e-05,
      "loss": 1.5576,
      "step": 4360
    },
    {
      "epoch": 0.7221648419747986,
      "grad_norm": 17.708297729492188,
      "learning_rate": 3.610973392827632e-05,
      "loss": 1.6501,
      "step": 4370
    },
    {
      "epoch": 0.723817393100599,
      "grad_norm": 9.903080940246582,
      "learning_rate": 3.619236489836391e-05,
      "loss": 1.604,
      "step": 4380
    },
    {
      "epoch": 0.7254699442263995,
      "grad_norm": 36.305904388427734,
      "learning_rate": 3.62749958684515e-05,
      "loss": 1.6622,
      "step": 4390
    },
    {
      "epoch": 0.7271224953522,
      "grad_norm": 17.942659378051758,
      "learning_rate": 3.635762683853909e-05,
      "loss": 1.5846,
      "step": 4400
    },
    {
      "epoch": 0.7287750464780004,
      "grad_norm": 15.10368537902832,
      "learning_rate": 3.644025780862668e-05,
      "loss": 1.6528,
      "step": 4410
    },
    {
      "epoch": 0.7304275976038008,
      "grad_norm": 11.274897575378418,
      "learning_rate": 3.652288877871427e-05,
      "loss": 1.6126,
      "step": 4420
    },
    {
      "epoch": 0.7320801487296014,
      "grad_norm": 9.983580589294434,
      "learning_rate": 3.660551974880185e-05,
      "loss": 1.7114,
      "step": 4430
    },
    {
      "epoch": 0.7337326998554018,
      "grad_norm": 6.765944480895996,
      "learning_rate": 3.668815071888944e-05,
      "loss": 1.6243,
      "step": 4440
    },
    {
      "epoch": 0.7353852509812022,
      "grad_norm": 6.97224760055542,
      "learning_rate": 3.677078168897703e-05,
      "loss": 1.6103,
      "step": 4450
    },
    {
      "epoch": 0.7370378021070026,
      "grad_norm": 18.04844856262207,
      "learning_rate": 3.6853412659064615e-05,
      "loss": 1.5527,
      "step": 4460
    },
    {
      "epoch": 0.7386903532328032,
      "grad_norm": 16.644474029541016,
      "learning_rate": 3.6936043629152205e-05,
      "loss": 1.7021,
      "step": 4470
    },
    {
      "epoch": 0.7403429043586036,
      "grad_norm": 12.994518280029297,
      "learning_rate": 3.7018674599239795e-05,
      "loss": 1.779,
      "step": 4480
    },
    {
      "epoch": 0.741995455484404,
      "grad_norm": 8.77868938446045,
      "learning_rate": 3.7101305569327385e-05,
      "loss": 1.6658,
      "step": 4490
    },
    {
      "epoch": 0.7436480066102045,
      "grad_norm": 16.25444793701172,
      "learning_rate": 3.7183936539414975e-05,
      "loss": 1.7969,
      "step": 4500
    },
    {
      "epoch": 0.745300557736005,
      "grad_norm": 15.957270622253418,
      "learning_rate": 3.7266567509502566e-05,
      "loss": 1.6981,
      "step": 4510
    },
    {
      "epoch": 0.7469531088618054,
      "grad_norm": 8.807640075683594,
      "learning_rate": 3.7349198479590156e-05,
      "loss": 1.8519,
      "step": 4520
    },
    {
      "epoch": 0.7486056599876059,
      "grad_norm": 13.07224178314209,
      "learning_rate": 3.743182944967774e-05,
      "loss": 1.8313,
      "step": 4530
    },
    {
      "epoch": 0.7502582111134063,
      "grad_norm": 14.359870910644531,
      "learning_rate": 3.751446041976533e-05,
      "loss": 1.7593,
      "step": 4540
    },
    {
      "epoch": 0.7519107622392067,
      "grad_norm": 10.050626754760742,
      "learning_rate": 3.759709138985292e-05,
      "loss": 1.7225,
      "step": 4550
    },
    {
      "epoch": 0.7535633133650073,
      "grad_norm": 20.963890075683594,
      "learning_rate": 3.76797223599405e-05,
      "loss": 1.7461,
      "step": 4560
    },
    {
      "epoch": 0.7552158644908077,
      "grad_norm": 14.734147071838379,
      "learning_rate": 3.776235333002809e-05,
      "loss": 1.7066,
      "step": 4570
    },
    {
      "epoch": 0.7568684156166081,
      "grad_norm": 12.787677764892578,
      "learning_rate": 3.784498430011569e-05,
      "loss": 1.7834,
      "step": 4580
    },
    {
      "epoch": 0.7585209667424085,
      "grad_norm": 11.134603500366211,
      "learning_rate": 3.792761527020327e-05,
      "loss": 1.6588,
      "step": 4590
    },
    {
      "epoch": 0.7601735178682091,
      "grad_norm": 7.746461391448975,
      "learning_rate": 3.8010246240290863e-05,
      "loss": 1.5581,
      "step": 4600
    },
    {
      "epoch": 0.7618260689940095,
      "grad_norm": 8.252470016479492,
      "learning_rate": 3.8092877210378454e-05,
      "loss": 1.7961,
      "step": 4610
    },
    {
      "epoch": 0.7634786201198099,
      "grad_norm": 18.933685302734375,
      "learning_rate": 3.8175508180466044e-05,
      "loss": 1.5873,
      "step": 4620
    },
    {
      "epoch": 0.7651311712456104,
      "grad_norm": 8.816574096679688,
      "learning_rate": 3.825813915055363e-05,
      "loss": 1.6589,
      "step": 4630
    },
    {
      "epoch": 0.7667837223714109,
      "grad_norm": 36.53535842895508,
      "learning_rate": 3.834077012064122e-05,
      "loss": 1.6683,
      "step": 4640
    },
    {
      "epoch": 0.7684362734972113,
      "grad_norm": 18.802400588989258,
      "learning_rate": 3.842340109072881e-05,
      "loss": 1.5741,
      "step": 4650
    },
    {
      "epoch": 0.7700888246230118,
      "grad_norm": 15.650837898254395,
      "learning_rate": 3.850603206081639e-05,
      "loss": 1.7178,
      "step": 4660
    },
    {
      "epoch": 0.7717413757488122,
      "grad_norm": 9.750154495239258,
      "learning_rate": 3.858866303090398e-05,
      "loss": 1.5336,
      "step": 4670
    },
    {
      "epoch": 0.7733939268746127,
      "grad_norm": 10.917224884033203,
      "learning_rate": 3.867129400099158e-05,
      "loss": 1.6023,
      "step": 4680
    },
    {
      "epoch": 0.7750464780004132,
      "grad_norm": 15.657227516174316,
      "learning_rate": 3.875392497107916e-05,
      "loss": 1.7192,
      "step": 4690
    },
    {
      "epoch": 0.7766990291262136,
      "grad_norm": 8.965598106384277,
      "learning_rate": 3.883655594116675e-05,
      "loss": 1.7072,
      "step": 4700
    },
    {
      "epoch": 0.778351580252014,
      "grad_norm": 5.406772613525391,
      "learning_rate": 3.891918691125434e-05,
      "loss": 1.617,
      "step": 4710
    },
    {
      "epoch": 0.7800041313778145,
      "grad_norm": 40.23686599731445,
      "learning_rate": 3.900181788134193e-05,
      "loss": 1.5586,
      "step": 4720
    },
    {
      "epoch": 0.781656682503615,
      "grad_norm": 6.773040294647217,
      "learning_rate": 3.9084448851429515e-05,
      "loss": 1.6325,
      "step": 4730
    },
    {
      "epoch": 0.7833092336294154,
      "grad_norm": 15.4874906539917,
      "learning_rate": 3.9167079821517105e-05,
      "loss": 1.8084,
      "step": 4740
    },
    {
      "epoch": 0.7849617847552158,
      "grad_norm": 9.330727577209473,
      "learning_rate": 3.9249710791604695e-05,
      "loss": 1.6506,
      "step": 4750
    },
    {
      "epoch": 0.7866143358810164,
      "grad_norm": 8.181884765625,
      "learning_rate": 3.933234176169228e-05,
      "loss": 1.74,
      "step": 4760
    },
    {
      "epoch": 0.7882668870068168,
      "grad_norm": 18.493993759155273,
      "learning_rate": 3.9414972731779876e-05,
      "loss": 1.5708,
      "step": 4770
    },
    {
      "epoch": 0.7899194381326172,
      "grad_norm": 8.724126815795898,
      "learning_rate": 3.9497603701867466e-05,
      "loss": 1.6966,
      "step": 4780
    },
    {
      "epoch": 0.7915719892584177,
      "grad_norm": 11.860855102539062,
      "learning_rate": 3.958023467195505e-05,
      "loss": 1.6671,
      "step": 4790
    },
    {
      "epoch": 0.7932245403842182,
      "grad_norm": 15.39936351776123,
      "learning_rate": 3.966286564204264e-05,
      "loss": 1.6334,
      "step": 4800
    },
    {
      "epoch": 0.7948770915100186,
      "grad_norm": 13.233650207519531,
      "learning_rate": 3.974549661213023e-05,
      "loss": 1.7003,
      "step": 4810
    },
    {
      "epoch": 0.796529642635819,
      "grad_norm": 13.286338806152344,
      "learning_rate": 3.982812758221782e-05,
      "loss": 1.6586,
      "step": 4820
    },
    {
      "epoch": 0.7981821937616195,
      "grad_norm": 70.44236755371094,
      "learning_rate": 3.99107585523054e-05,
      "loss": 1.7162,
      "step": 4830
    },
    {
      "epoch": 0.7998347448874199,
      "grad_norm": 38.03678894042969,
      "learning_rate": 3.999338952239299e-05,
      "loss": 1.6638,
      "step": 4840
    },
    {
      "epoch": 0.8014872960132204,
      "grad_norm": 9.280917167663574,
      "learning_rate": 4.0076020492480584e-05,
      "loss": 1.5643,
      "step": 4850
    },
    {
      "epoch": 0.8031398471390209,
      "grad_norm": 46.079708099365234,
      "learning_rate": 4.015865146256817e-05,
      "loss": 1.7017,
      "step": 4860
    },
    {
      "epoch": 0.8047923982648213,
      "grad_norm": 7.869788646697998,
      "learning_rate": 4.0241282432655764e-05,
      "loss": 1.6566,
      "step": 4870
    },
    {
      "epoch": 0.8064449493906217,
      "grad_norm": 13.424615859985352,
      "learning_rate": 4.0323913402743354e-05,
      "loss": 1.7328,
      "step": 4880
    },
    {
      "epoch": 0.8080975005164223,
      "grad_norm": 16.283950805664062,
      "learning_rate": 4.040654437283094e-05,
      "loss": 1.6783,
      "step": 4890
    },
    {
      "epoch": 0.8097500516422227,
      "grad_norm": 53.05949020385742,
      "learning_rate": 4.048917534291853e-05,
      "loss": 1.6176,
      "step": 4900
    },
    {
      "epoch": 0.8114026027680231,
      "grad_norm": 11.25322151184082,
      "learning_rate": 4.057180631300612e-05,
      "loss": 1.6248,
      "step": 4910
    },
    {
      "epoch": 0.8130551538938235,
      "grad_norm": 30.585168838500977,
      "learning_rate": 4.065443728309371e-05,
      "loss": 1.6711,
      "step": 4920
    },
    {
      "epoch": 0.8147077050196241,
      "grad_norm": 8.419675827026367,
      "learning_rate": 4.073706825318129e-05,
      "loss": 1.6889,
      "step": 4930
    },
    {
      "epoch": 0.8163602561454245,
      "grad_norm": 32.17693328857422,
      "learning_rate": 4.081969922326888e-05,
      "loss": 1.7163,
      "step": 4940
    },
    {
      "epoch": 0.8180128072712249,
      "grad_norm": 11.359280586242676,
      "learning_rate": 4.090233019335647e-05,
      "loss": 1.6454,
      "step": 4950
    },
    {
      "epoch": 0.8196653583970254,
      "grad_norm": 15.660289764404297,
      "learning_rate": 4.098496116344406e-05,
      "loss": 1.7507,
      "step": 4960
    },
    {
      "epoch": 0.8213179095228259,
      "grad_norm": 10.861181259155273,
      "learning_rate": 4.106759213353165e-05,
      "loss": 1.648,
      "step": 4970
    },
    {
      "epoch": 0.8229704606486263,
      "grad_norm": 8.45966625213623,
      "learning_rate": 4.115022310361924e-05,
      "loss": 1.6779,
      "step": 4980
    },
    {
      "epoch": 0.8246230117744268,
      "grad_norm": 8.247032165527344,
      "learning_rate": 4.1232854073706825e-05,
      "loss": 1.7897,
      "step": 4990
    },
    {
      "epoch": 0.8262755629002272,
      "grad_norm": 10.746397018432617,
      "learning_rate": 4.1315485043794416e-05,
      "loss": 1.6776,
      "step": 5000
    },
    {
      "epoch": 0.8279281140260277,
      "grad_norm": 11.24117660522461,
      "learning_rate": 4.1398116013882006e-05,
      "loss": 1.6349,
      "step": 5010
    },
    {
      "epoch": 0.8295806651518282,
      "grad_norm": 13.31236457824707,
      "learning_rate": 4.1480746983969596e-05,
      "loss": 1.5985,
      "step": 5020
    },
    {
      "epoch": 0.8312332162776286,
      "grad_norm": 32.01264572143555,
      "learning_rate": 4.156337795405718e-05,
      "loss": 1.7555,
      "step": 5030
    },
    {
      "epoch": 0.832885767403429,
      "grad_norm": 9.02853775024414,
      "learning_rate": 4.164600892414477e-05,
      "loss": 1.5571,
      "step": 5040
    },
    {
      "epoch": 0.8345383185292296,
      "grad_norm": 20.635509490966797,
      "learning_rate": 4.172863989423236e-05,
      "loss": 1.7462,
      "step": 5050
    },
    {
      "epoch": 0.83619086965503,
      "grad_norm": 11.947609901428223,
      "learning_rate": 4.181127086431995e-05,
      "loss": 1.6536,
      "step": 5060
    },
    {
      "epoch": 0.8378434207808304,
      "grad_norm": 7.472456932067871,
      "learning_rate": 4.189390183440754e-05,
      "loss": 1.8139,
      "step": 5070
    },
    {
      "epoch": 0.8394959719066308,
      "grad_norm": 11.594853401184082,
      "learning_rate": 4.197653280449513e-05,
      "loss": 1.5398,
      "step": 5080
    },
    {
      "epoch": 0.8411485230324313,
      "grad_norm": 22.656166076660156,
      "learning_rate": 4.2059163774582713e-05,
      "loss": 1.7391,
      "step": 5090
    },
    {
      "epoch": 0.8428010741582318,
      "grad_norm": 15.495370864868164,
      "learning_rate": 4.2141794744670304e-05,
      "loss": 1.8129,
      "step": 5100
    },
    {
      "epoch": 0.8444536252840322,
      "grad_norm": 6.661768436431885,
      "learning_rate": 4.2224425714757894e-05,
      "loss": 1.6834,
      "step": 5110
    },
    {
      "epoch": 0.8461061764098327,
      "grad_norm": 7.477358818054199,
      "learning_rate": 4.2307056684845484e-05,
      "loss": 1.5815,
      "step": 5120
    },
    {
      "epoch": 0.8477587275356331,
      "grad_norm": 11.148330688476562,
      "learning_rate": 4.238968765493307e-05,
      "loss": 1.6634,
      "step": 5130
    },
    {
      "epoch": 0.8494112786614336,
      "grad_norm": 9.076679229736328,
      "learning_rate": 4.247231862502066e-05,
      "loss": 1.6155,
      "step": 5140
    },
    {
      "epoch": 0.851063829787234,
      "grad_norm": 10.360162734985352,
      "learning_rate": 4.2554949595108254e-05,
      "loss": 1.5972,
      "step": 5150
    },
    {
      "epoch": 0.8527163809130345,
      "grad_norm": 10.570571899414062,
      "learning_rate": 4.263758056519584e-05,
      "loss": 1.7933,
      "step": 5160
    },
    {
      "epoch": 0.8543689320388349,
      "grad_norm": 7.64302396774292,
      "learning_rate": 4.272021153528343e-05,
      "loss": 1.7115,
      "step": 5170
    },
    {
      "epoch": 0.8560214831646354,
      "grad_norm": 11.577961921691895,
      "learning_rate": 4.280284250537102e-05,
      "loss": 1.7791,
      "step": 5180
    },
    {
      "epoch": 0.8576740342904359,
      "grad_norm": 9.828375816345215,
      "learning_rate": 4.28854734754586e-05,
      "loss": 1.7543,
      "step": 5190
    },
    {
      "epoch": 0.8593265854162363,
      "grad_norm": 10.680025100708008,
      "learning_rate": 4.296810444554619e-05,
      "loss": 1.569,
      "step": 5200
    },
    {
      "epoch": 0.8609791365420367,
      "grad_norm": 10.61168384552002,
      "learning_rate": 4.305073541563378e-05,
      "loss": 1.6397,
      "step": 5210
    },
    {
      "epoch": 0.8626316876678373,
      "grad_norm": 13.09295654296875,
      "learning_rate": 4.313336638572137e-05,
      "loss": 1.663,
      "step": 5220
    },
    {
      "epoch": 0.8642842387936377,
      "grad_norm": 10.244070053100586,
      "learning_rate": 4.3215997355808955e-05,
      "loss": 1.8205,
      "step": 5230
    },
    {
      "epoch": 0.8659367899194381,
      "grad_norm": 14.051331520080566,
      "learning_rate": 4.329862832589655e-05,
      "loss": 1.7455,
      "step": 5240
    },
    {
      "epoch": 0.8675893410452385,
      "grad_norm": 14.486235618591309,
      "learning_rate": 4.338125929598414e-05,
      "loss": 1.6745,
      "step": 5250
    },
    {
      "epoch": 0.8692418921710391,
      "grad_norm": 10.930896759033203,
      "learning_rate": 4.3463890266071726e-05,
      "loss": 1.6661,
      "step": 5260
    },
    {
      "epoch": 0.8708944432968395,
      "grad_norm": 23.3206729888916,
      "learning_rate": 4.3546521236159316e-05,
      "loss": 1.6653,
      "step": 5270
    },
    {
      "epoch": 0.8725469944226399,
      "grad_norm": 22.273244857788086,
      "learning_rate": 4.3629152206246906e-05,
      "loss": 1.7524,
      "step": 5280
    },
    {
      "epoch": 0.8741995455484404,
      "grad_norm": 11.886351585388184,
      "learning_rate": 4.371178317633449e-05,
      "loss": 1.7256,
      "step": 5290
    },
    {
      "epoch": 0.8758520966742409,
      "grad_norm": 8.98186206817627,
      "learning_rate": 4.379441414642208e-05,
      "loss": 1.7292,
      "step": 5300
    },
    {
      "epoch": 0.8775046478000413,
      "grad_norm": 10.950017929077148,
      "learning_rate": 4.387704511650967e-05,
      "loss": 1.7558,
      "step": 5310
    },
    {
      "epoch": 0.8791571989258418,
      "grad_norm": 10.48299789428711,
      "learning_rate": 4.395967608659726e-05,
      "loss": 1.7535,
      "step": 5320
    },
    {
      "epoch": 0.8808097500516422,
      "grad_norm": 22.365188598632812,
      "learning_rate": 4.4042307056684843e-05,
      "loss": 1.7458,
      "step": 5330
    },
    {
      "epoch": 0.8824623011774427,
      "grad_norm": 9.840352058410645,
      "learning_rate": 4.412493802677244e-05,
      "loss": 1.668,
      "step": 5340
    },
    {
      "epoch": 0.8841148523032432,
      "grad_norm": 7.298482894897461,
      "learning_rate": 4.4207568996860024e-05,
      "loss": 1.672,
      "step": 5350
    },
    {
      "epoch": 0.8857674034290436,
      "grad_norm": 26.250947952270508,
      "learning_rate": 4.4290199966947614e-05,
      "loss": 1.7415,
      "step": 5360
    },
    {
      "epoch": 0.887419954554844,
      "grad_norm": 15.844700813293457,
      "learning_rate": 4.4372830937035204e-05,
      "loss": 1.6139,
      "step": 5370
    },
    {
      "epoch": 0.8890725056806444,
      "grad_norm": 7.872781276702881,
      "learning_rate": 4.4455461907122794e-05,
      "loss": 1.6652,
      "step": 5380
    },
    {
      "epoch": 0.890725056806445,
      "grad_norm": 11.659095764160156,
      "learning_rate": 4.453809287721038e-05,
      "loss": 1.6276,
      "step": 5390
    },
    {
      "epoch": 0.8923776079322454,
      "grad_norm": 25.442895889282227,
      "learning_rate": 4.462072384729797e-05,
      "loss": 1.7076,
      "step": 5400
    },
    {
      "epoch": 0.8940301590580458,
      "grad_norm": 40.754371643066406,
      "learning_rate": 4.470335481738556e-05,
      "loss": 1.7066,
      "step": 5410
    },
    {
      "epoch": 0.8956827101838463,
      "grad_norm": 23.09174156188965,
      "learning_rate": 4.478598578747315e-05,
      "loss": 1.7866,
      "step": 5420
    },
    {
      "epoch": 0.8973352613096468,
      "grad_norm": 24.503602981567383,
      "learning_rate": 4.486861675756074e-05,
      "loss": 1.5919,
      "step": 5430
    },
    {
      "epoch": 0.8989878124354472,
      "grad_norm": 23.102312088012695,
      "learning_rate": 4.495124772764833e-05,
      "loss": 1.7015,
      "step": 5440
    },
    {
      "epoch": 0.9006403635612477,
      "grad_norm": 8.59858226776123,
      "learning_rate": 4.503387869773591e-05,
      "loss": 1.7018,
      "step": 5450
    },
    {
      "epoch": 0.9022929146870481,
      "grad_norm": 23.99196434020996,
      "learning_rate": 4.51165096678235e-05,
      "loss": 1.627,
      "step": 5460
    },
    {
      "epoch": 0.9039454658128486,
      "grad_norm": 14.380967140197754,
      "learning_rate": 4.519914063791109e-05,
      "loss": 1.5606,
      "step": 5470
    },
    {
      "epoch": 0.905598016938649,
      "grad_norm": 18.584856033325195,
      "learning_rate": 4.528177160799868e-05,
      "loss": 1.8168,
      "step": 5480
    },
    {
      "epoch": 0.9072505680644495,
      "grad_norm": 11.22840690612793,
      "learning_rate": 4.5364402578086266e-05,
      "loss": 1.612,
      "step": 5490
    },
    {
      "epoch": 0.9089031191902499,
      "grad_norm": 20.63224983215332,
      "learning_rate": 4.5447033548173856e-05,
      "loss": 1.6359,
      "step": 5500
    },
    {
      "epoch": 0.9105556703160504,
      "grad_norm": 9.386175155639648,
      "learning_rate": 4.5529664518261446e-05,
      "loss": 1.7047,
      "step": 5510
    },
    {
      "epoch": 0.9122082214418509,
      "grad_norm": 12.939002990722656,
      "learning_rate": 4.5612295488349036e-05,
      "loss": 1.6124,
      "step": 5520
    },
    {
      "epoch": 0.9138607725676513,
      "grad_norm": 10.644198417663574,
      "learning_rate": 4.5694926458436626e-05,
      "loss": 1.6824,
      "step": 5530
    },
    {
      "epoch": 0.9155133236934517,
      "grad_norm": 15.673925399780273,
      "learning_rate": 4.5777557428524216e-05,
      "loss": 1.7603,
      "step": 5540
    },
    {
      "epoch": 0.9171658748192523,
      "grad_norm": 10.065011024475098,
      "learning_rate": 4.58601883986118e-05,
      "loss": 1.804,
      "step": 5550
    },
    {
      "epoch": 0.9188184259450527,
      "grad_norm": 9.452924728393555,
      "learning_rate": 4.594281936869939e-05,
      "loss": 1.7851,
      "step": 5560
    },
    {
      "epoch": 0.9204709770708531,
      "grad_norm": 10.779854774475098,
      "learning_rate": 4.602545033878698e-05,
      "loss": 1.7237,
      "step": 5570
    },
    {
      "epoch": 0.9221235281966536,
      "grad_norm": 10.967293739318848,
      "learning_rate": 4.610808130887457e-05,
      "loss": 1.6173,
      "step": 5580
    },
    {
      "epoch": 0.9237760793224541,
      "grad_norm": 91.86517333984375,
      "learning_rate": 4.6190712278962154e-05,
      "loss": 1.6469,
      "step": 5590
    },
    {
      "epoch": 0.9254286304482545,
      "grad_norm": 4.923385143280029,
      "learning_rate": 4.6273343249049744e-05,
      "loss": 1.5892,
      "step": 5600
    },
    {
      "epoch": 0.927081181574055,
      "grad_norm": 19.811214447021484,
      "learning_rate": 4.6355974219137334e-05,
      "loss": 1.5998,
      "step": 5610
    },
    {
      "epoch": 0.9287337326998554,
      "grad_norm": 9.930047035217285,
      "learning_rate": 4.6438605189224924e-05,
      "loss": 1.6312,
      "step": 5620
    },
    {
      "epoch": 0.9303862838256558,
      "grad_norm": 20.98347282409668,
      "learning_rate": 4.6521236159312514e-05,
      "loss": 1.7513,
      "step": 5630
    },
    {
      "epoch": 0.9320388349514563,
      "grad_norm": 11.083711624145508,
      "learning_rate": 4.6603867129400104e-05,
      "loss": 1.7647,
      "step": 5640
    },
    {
      "epoch": 0.9336913860772568,
      "grad_norm": 15.540210723876953,
      "learning_rate": 4.668649809948769e-05,
      "loss": 1.606,
      "step": 5650
    },
    {
      "epoch": 0.9353439372030572,
      "grad_norm": 40.91162872314453,
      "learning_rate": 4.676912906957528e-05,
      "loss": 1.5821,
      "step": 5660
    },
    {
      "epoch": 0.9369964883288576,
      "grad_norm": 10.069815635681152,
      "learning_rate": 4.685176003966287e-05,
      "loss": 1.5963,
      "step": 5670
    },
    {
      "epoch": 0.9386490394546582,
      "grad_norm": 7.199189186096191,
      "learning_rate": 4.693439100975046e-05,
      "loss": 1.6151,
      "step": 5680
    },
    {
      "epoch": 0.9403015905804586,
      "grad_norm": 14.112994194030762,
      "learning_rate": 4.701702197983804e-05,
      "loss": 1.8498,
      "step": 5690
    },
    {
      "epoch": 0.941954141706259,
      "grad_norm": 17.025121688842773,
      "learning_rate": 4.709965294992563e-05,
      "loss": 1.5699,
      "step": 5700
    },
    {
      "epoch": 0.9436066928320594,
      "grad_norm": 22.465831756591797,
      "learning_rate": 4.718228392001322e-05,
      "loss": 1.7636,
      "step": 5710
    },
    {
      "epoch": 0.94525924395786,
      "grad_norm": 16.813859939575195,
      "learning_rate": 4.726491489010081e-05,
      "loss": 1.6191,
      "step": 5720
    },
    {
      "epoch": 0.9469117950836604,
      "grad_norm": 12.327750205993652,
      "learning_rate": 4.73475458601884e-05,
      "loss": 1.7167,
      "step": 5730
    },
    {
      "epoch": 0.9485643462094608,
      "grad_norm": 12.070796966552734,
      "learning_rate": 4.743017683027599e-05,
      "loss": 1.7759,
      "step": 5740
    },
    {
      "epoch": 0.9502168973352613,
      "grad_norm": 23.6120548248291,
      "learning_rate": 4.7512807800363576e-05,
      "loss": 1.781,
      "step": 5750
    },
    {
      "epoch": 0.9518694484610618,
      "grad_norm": 12.13965892791748,
      "learning_rate": 4.7595438770451166e-05,
      "loss": 1.4901,
      "step": 5760
    },
    {
      "epoch": 0.9535219995868622,
      "grad_norm": 10.353153228759766,
      "learning_rate": 4.7678069740538756e-05,
      "loss": 1.7373,
      "step": 5770
    },
    {
      "epoch": 0.9551745507126627,
      "grad_norm": 13.95853328704834,
      "learning_rate": 4.7760700710626346e-05,
      "loss": 1.771,
      "step": 5780
    },
    {
      "epoch": 0.9568271018384631,
      "grad_norm": 9.67155647277832,
      "learning_rate": 4.784333168071393e-05,
      "loss": 1.6026,
      "step": 5790
    },
    {
      "epoch": 0.9584796529642636,
      "grad_norm": 20.621294021606445,
      "learning_rate": 4.792596265080152e-05,
      "loss": 1.6695,
      "step": 5800
    },
    {
      "epoch": 0.9601322040900641,
      "grad_norm": 24.754667282104492,
      "learning_rate": 4.800859362088912e-05,
      "loss": 1.8208,
      "step": 5810
    },
    {
      "epoch": 0.9617847552158645,
      "grad_norm": 8.666321754455566,
      "learning_rate": 4.80912245909767e-05,
      "loss": 1.7137,
      "step": 5820
    },
    {
      "epoch": 0.9634373063416649,
      "grad_norm": 43.4861946105957,
      "learning_rate": 4.817385556106429e-05,
      "loss": 1.6497,
      "step": 5830
    },
    {
      "epoch": 0.9650898574674655,
      "grad_norm": 30.73326873779297,
      "learning_rate": 4.825648653115188e-05,
      "loss": 1.7121,
      "step": 5840
    },
    {
      "epoch": 0.9667424085932659,
      "grad_norm": 33.42872619628906,
      "learning_rate": 4.8339117501239464e-05,
      "loss": 1.8122,
      "step": 5850
    },
    {
      "epoch": 0.9683949597190663,
      "grad_norm": 27.492687225341797,
      "learning_rate": 4.8421748471327054e-05,
      "loss": 1.6811,
      "step": 5860
    },
    {
      "epoch": 0.9700475108448667,
      "grad_norm": 6.809346675872803,
      "learning_rate": 4.8504379441414644e-05,
      "loss": 1.6784,
      "step": 5870
    },
    {
      "epoch": 0.9717000619706673,
      "grad_norm": 6.992137432098389,
      "learning_rate": 4.8587010411502234e-05,
      "loss": 1.5981,
      "step": 5880
    },
    {
      "epoch": 0.9733526130964677,
      "grad_norm": 9.034411430358887,
      "learning_rate": 4.866964138158982e-05,
      "loss": 1.7404,
      "step": 5890
    },
    {
      "epoch": 0.9750051642222681,
      "grad_norm": 7.147427082061768,
      "learning_rate": 4.8752272351677415e-05,
      "loss": 1.719,
      "step": 5900
    },
    {
      "epoch": 0.9766577153480686,
      "grad_norm": 6.848790168762207,
      "learning_rate": 4.8834903321765005e-05,
      "loss": 1.6843,
      "step": 5910
    },
    {
      "epoch": 0.978310266473869,
      "grad_norm": 13.439620971679688,
      "learning_rate": 4.891753429185259e-05,
      "loss": 1.7116,
      "step": 5920
    },
    {
      "epoch": 0.9799628175996695,
      "grad_norm": 6.594333648681641,
      "learning_rate": 4.900016526194018e-05,
      "loss": 1.8479,
      "step": 5930
    },
    {
      "epoch": 0.98161536872547,
      "grad_norm": 7.724569320678711,
      "learning_rate": 4.908279623202777e-05,
      "loss": 1.6023,
      "step": 5940
    },
    {
      "epoch": 0.9832679198512704,
      "grad_norm": 41.57175064086914,
      "learning_rate": 4.916542720211535e-05,
      "loss": 1.7183,
      "step": 5950
    },
    {
      "epoch": 0.9849204709770708,
      "grad_norm": 16.64209747314453,
      "learning_rate": 4.924805817220294e-05,
      "loss": 1.6207,
      "step": 5960
    },
    {
      "epoch": 0.9865730221028713,
      "grad_norm": 13.659209251403809,
      "learning_rate": 4.933068914229053e-05,
      "loss": 1.6747,
      "step": 5970
    },
    {
      "epoch": 0.9882255732286718,
      "grad_norm": 7.966761589050293,
      "learning_rate": 4.941332011237812e-05,
      "loss": 1.6441,
      "step": 5980
    },
    {
      "epoch": 0.9898781243544722,
      "grad_norm": 5.628538131713867,
      "learning_rate": 4.9495951082465706e-05,
      "loss": 1.7499,
      "step": 5990
    },
    {
      "epoch": 0.9915306754802726,
      "grad_norm": 8.069196701049805,
      "learning_rate": 4.95785820525533e-05,
      "loss": 1.88,
      "step": 6000
    },
    {
      "epoch": 0.9931832266060732,
      "grad_norm": 51.40681457519531,
      "learning_rate": 4.966121302264089e-05,
      "loss": 1.6778,
      "step": 6010
    },
    {
      "epoch": 0.9948357777318736,
      "grad_norm": 6.667893886566162,
      "learning_rate": 4.9743843992728476e-05,
      "loss": 1.6331,
      "step": 6020
    },
    {
      "epoch": 0.996488328857674,
      "grad_norm": 8.195232391357422,
      "learning_rate": 4.9826474962816066e-05,
      "loss": 1.6024,
      "step": 6030
    },
    {
      "epoch": 0.9981408799834744,
      "grad_norm": 14.904173851013184,
      "learning_rate": 4.9909105932903657e-05,
      "loss": 1.7938,
      "step": 6040
    },
    {
      "epoch": 0.999793431109275,
      "grad_norm": 9.155719757080078,
      "learning_rate": 4.999173690299124e-05,
      "loss": 1.5715,
      "step": 6050
    },
    {
      "epoch": 0.999958686221855,
      "eval_accuracy": 0.2757837991204426,
      "eval_loss": 2.0155398845672607,
      "eval_runtime": 854.7614,
      "eval_samples_per_second": 32.987,
      "eval_steps_per_second": 8.247,
      "step": 6051
    },
    {
      "epoch": 1.0014459822350754,
      "grad_norm": 9.607635498046875,
      "learning_rate": 4.999173690299124e-05,
      "loss": 1.6654,
      "step": 6060
    },
    {
      "epoch": 1.003098533360876,
      "grad_norm": 5.138330459594727,
      "learning_rate": 4.998255568409262e-05,
      "loss": 1.5768,
      "step": 6070
    },
    {
      "epoch": 1.0047510844866763,
      "grad_norm": 28.794178009033203,
      "learning_rate": 4.9973374465194e-05,
      "loss": 1.6652,
      "step": 6080
    },
    {
      "epoch": 1.0064036356124768,
      "grad_norm": 14.260908126831055,
      "learning_rate": 4.996419324629538e-05,
      "loss": 1.5967,
      "step": 6090
    },
    {
      "epoch": 1.0080561867382771,
      "grad_norm": 11.221604347229004,
      "learning_rate": 4.995501202739676e-05,
      "loss": 1.6488,
      "step": 6100
    },
    {
      "epoch": 1.0097087378640777,
      "grad_norm": 8.829093933105469,
      "learning_rate": 4.994583080849814e-05,
      "loss": 1.6887,
      "step": 6110
    },
    {
      "epoch": 1.0113612889898782,
      "grad_norm": 10.191542625427246,
      "learning_rate": 4.9936649589599515e-05,
      "loss": 1.7274,
      "step": 6120
    },
    {
      "epoch": 1.0130138401156785,
      "grad_norm": 9.99532413482666,
      "learning_rate": 4.99274683707009e-05,
      "loss": 1.6879,
      "step": 6130
    },
    {
      "epoch": 1.014666391241479,
      "grad_norm": 26.136659622192383,
      "learning_rate": 4.9918287151802274e-05,
      "loss": 1.8429,
      "step": 6140
    },
    {
      "epoch": 1.0163189423672794,
      "grad_norm": 7.408060073852539,
      "learning_rate": 4.9909105932903657e-05,
      "loss": 1.6909,
      "step": 6150
    },
    {
      "epoch": 1.01797149349308,
      "grad_norm": 15.327078819274902,
      "learning_rate": 4.989992471400504e-05,
      "loss": 1.7461,
      "step": 6160
    },
    {
      "epoch": 1.0196240446188805,
      "grad_norm": 22.485273361206055,
      "learning_rate": 4.9890743495106415e-05,
      "loss": 1.767,
      "step": 6170
    },
    {
      "epoch": 1.0212765957446808,
      "grad_norm": 21.41472816467285,
      "learning_rate": 4.988156227620779e-05,
      "loss": 1.6902,
      "step": 6180
    },
    {
      "epoch": 1.0229291468704813,
      "grad_norm": 8.6235933303833,
      "learning_rate": 4.987238105730917e-05,
      "loss": 1.6762,
      "step": 6190
    },
    {
      "epoch": 1.0245816979962818,
      "grad_norm": 13.099306106567383,
      "learning_rate": 4.986319983841055e-05,
      "loss": 1.7881,
      "step": 6200
    },
    {
      "epoch": 1.0262342491220822,
      "grad_norm": 6.052161693572998,
      "learning_rate": 4.9854018619511925e-05,
      "loss": 1.5323,
      "step": 6210
    },
    {
      "epoch": 1.0278868002478827,
      "grad_norm": 9.501975059509277,
      "learning_rate": 4.984483740061331e-05,
      "loss": 1.6617,
      "step": 6220
    },
    {
      "epoch": 1.029539351373683,
      "grad_norm": 68.59492492675781,
      "learning_rate": 4.9835656181714684e-05,
      "loss": 1.655,
      "step": 6230
    },
    {
      "epoch": 1.0311919024994836,
      "grad_norm": 33.55989456176758,
      "learning_rate": 4.9826474962816066e-05,
      "loss": 1.7361,
      "step": 6240
    },
    {
      "epoch": 1.032844453625284,
      "grad_norm": 33.48898696899414,
      "learning_rate": 4.981729374391744e-05,
      "loss": 1.667,
      "step": 6250
    },
    {
      "epoch": 1.0344970047510844,
      "grad_norm": 21.978069305419922,
      "learning_rate": 4.9808112525018825e-05,
      "loss": 1.6726,
      "step": 6260
    },
    {
      "epoch": 1.036149555876885,
      "grad_norm": 27.184450149536133,
      "learning_rate": 4.979893130612021e-05,
      "loss": 1.6208,
      "step": 6270
    },
    {
      "epoch": 1.0378021070026855,
      "grad_norm": 12.480914115905762,
      "learning_rate": 4.9789750087221583e-05,
      "loss": 1.707,
      "step": 6280
    },
    {
      "epoch": 1.0394546581284858,
      "grad_norm": 9.19139575958252,
      "learning_rate": 4.9780568868322966e-05,
      "loss": 1.6405,
      "step": 6290
    },
    {
      "epoch": 1.0411072092542863,
      "grad_norm": 92.05270385742188,
      "learning_rate": 4.977138764942434e-05,
      "loss": 1.5893,
      "step": 6300
    },
    {
      "epoch": 1.0427597603800867,
      "grad_norm": 17.508121490478516,
      "learning_rate": 4.976220643052572e-05,
      "loss": 1.6037,
      "step": 6310
    },
    {
      "epoch": 1.0444123115058872,
      "grad_norm": 46.28389358520508,
      "learning_rate": 4.9753025211627094e-05,
      "loss": 1.5507,
      "step": 6320
    },
    {
      "epoch": 1.0460648626316877,
      "grad_norm": 26.94911003112793,
      "learning_rate": 4.9743843992728476e-05,
      "loss": 1.8003,
      "step": 6330
    },
    {
      "epoch": 1.047717413757488,
      "grad_norm": 7.172421932220459,
      "learning_rate": 4.973466277382985e-05,
      "loss": 1.7725,
      "step": 6340
    },
    {
      "epoch": 1.0493699648832886,
      "grad_norm": 13.676774978637695,
      "learning_rate": 4.9725481554931235e-05,
      "loss": 1.5975,
      "step": 6350
    },
    {
      "epoch": 1.0510225160090891,
      "grad_norm": 9.980374336242676,
      "learning_rate": 4.971630033603261e-05,
      "loss": 1.5381,
      "step": 6360
    },
    {
      "epoch": 1.0526750671348895,
      "grad_norm": 11.59304428100586,
      "learning_rate": 4.970711911713399e-05,
      "loss": 1.7605,
      "step": 6370
    },
    {
      "epoch": 1.05432761826069,
      "grad_norm": 5.923514366149902,
      "learning_rate": 4.969793789823537e-05,
      "loss": 1.6649,
      "step": 6380
    },
    {
      "epoch": 1.0559801693864903,
      "grad_norm": 10.02888298034668,
      "learning_rate": 4.968875667933675e-05,
      "loss": 1.7624,
      "step": 6390
    },
    {
      "epoch": 1.0576327205122908,
      "grad_norm": 24.110645294189453,
      "learning_rate": 4.9679575460438134e-05,
      "loss": 1.7263,
      "step": 6400
    },
    {
      "epoch": 1.0592852716380914,
      "grad_norm": 29.78762435913086,
      "learning_rate": 4.967039424153951e-05,
      "loss": 1.5858,
      "step": 6410
    },
    {
      "epoch": 1.0609378227638917,
      "grad_norm": 10.339710235595703,
      "learning_rate": 4.966121302264089e-05,
      "loss": 1.5688,
      "step": 6420
    },
    {
      "epoch": 1.0625903738896922,
      "grad_norm": 12.489291191101074,
      "learning_rate": 4.965203180374227e-05,
      "loss": 1.7094,
      "step": 6430
    },
    {
      "epoch": 1.0642429250154928,
      "grad_norm": 7.957202434539795,
      "learning_rate": 4.9642850584843645e-05,
      "loss": 1.6154,
      "step": 6440
    },
    {
      "epoch": 1.065895476141293,
      "grad_norm": 8.301098823547363,
      "learning_rate": 4.963366936594502e-05,
      "loss": 1.7572,
      "step": 6450
    },
    {
      "epoch": 1.0675480272670936,
      "grad_norm": 16.680442810058594,
      "learning_rate": 4.96244881470464e-05,
      "loss": 1.7111,
      "step": 6460
    },
    {
      "epoch": 1.069200578392894,
      "grad_norm": 8.970675468444824,
      "learning_rate": 4.961530692814778e-05,
      "loss": 1.7194,
      "step": 6470
    },
    {
      "epoch": 1.0708531295186945,
      "grad_norm": 16.222583770751953,
      "learning_rate": 4.960612570924916e-05,
      "loss": 1.6507,
      "step": 6480
    },
    {
      "epoch": 1.072505680644495,
      "grad_norm": 7.895381450653076,
      "learning_rate": 4.959694449035054e-05,
      "loss": 1.6932,
      "step": 6490
    },
    {
      "epoch": 1.0741582317702953,
      "grad_norm": 19.77764129638672,
      "learning_rate": 4.958776327145192e-05,
      "loss": 1.8433,
      "step": 6500
    },
    {
      "epoch": 1.0758107828960959,
      "grad_norm": 14.908355712890625,
      "learning_rate": 4.95785820525533e-05,
      "loss": 1.5765,
      "step": 6510
    },
    {
      "epoch": 1.0774633340218962,
      "grad_norm": 10.35196304321289,
      "learning_rate": 4.956940083365468e-05,
      "loss": 1.8107,
      "step": 6520
    },
    {
      "epoch": 1.0791158851476967,
      "grad_norm": 9.052248001098633,
      "learning_rate": 4.956021961475606e-05,
      "loss": 1.7856,
      "step": 6530
    },
    {
      "epoch": 1.0807684362734973,
      "grad_norm": 7.395838737487793,
      "learning_rate": 4.955103839585744e-05,
      "loss": 1.691,
      "step": 6540
    },
    {
      "epoch": 1.0824209873992976,
      "grad_norm": 36.31234359741211,
      "learning_rate": 4.954185717695882e-05,
      "loss": 1.6575,
      "step": 6550
    },
    {
      "epoch": 1.0840735385250981,
      "grad_norm": 19.741281509399414,
      "learning_rate": 4.9532675958060196e-05,
      "loss": 1.7273,
      "step": 6560
    },
    {
      "epoch": 1.0857260896508987,
      "grad_norm": 44.102500915527344,
      "learning_rate": 4.952349473916157e-05,
      "loss": 1.7415,
      "step": 6570
    },
    {
      "epoch": 1.087378640776699,
      "grad_norm": 5.051490783691406,
      "learning_rate": 4.951431352026295e-05,
      "loss": 1.6828,
      "step": 6580
    },
    {
      "epoch": 1.0890311919024995,
      "grad_norm": 6.60805606842041,
      "learning_rate": 4.950513230136433e-05,
      "loss": 1.6364,
      "step": 6590
    },
    {
      "epoch": 1.0906837430282998,
      "grad_norm": 18.499725341796875,
      "learning_rate": 4.9495951082465706e-05,
      "loss": 1.6844,
      "step": 6600
    },
    {
      "epoch": 1.0923362941541004,
      "grad_norm": 18.149259567260742,
      "learning_rate": 4.948676986356709e-05,
      "loss": 1.5962,
      "step": 6610
    },
    {
      "epoch": 1.093988845279901,
      "grad_norm": 17.02660369873047,
      "learning_rate": 4.947758864466847e-05,
      "loss": 1.6441,
      "step": 6620
    },
    {
      "epoch": 1.0956413964057012,
      "grad_norm": 13.559640884399414,
      "learning_rate": 4.946840742576985e-05,
      "loss": 1.5714,
      "step": 6630
    },
    {
      "epoch": 1.0972939475315018,
      "grad_norm": 12.437646865844727,
      "learning_rate": 4.945922620687123e-05,
      "loss": 1.7886,
      "step": 6640
    },
    {
      "epoch": 1.0989464986573023,
      "grad_norm": 10.138182640075684,
      "learning_rate": 4.9450044987972605e-05,
      "loss": 1.6831,
      "step": 6650
    },
    {
      "epoch": 1.1005990497831026,
      "grad_norm": 18.947307586669922,
      "learning_rate": 4.944086376907399e-05,
      "loss": 1.634,
      "step": 6660
    },
    {
      "epoch": 1.1022516009089032,
      "grad_norm": 7.246707439422607,
      "learning_rate": 4.9431682550175364e-05,
      "loss": 1.6895,
      "step": 6670
    },
    {
      "epoch": 1.1039041520347035,
      "grad_norm": 8.21375846862793,
      "learning_rate": 4.9422501331276747e-05,
      "loss": 1.6634,
      "step": 6680
    },
    {
      "epoch": 1.105556703160504,
      "grad_norm": 14.554459571838379,
      "learning_rate": 4.941332011237812e-05,
      "loss": 1.7784,
      "step": 6690
    },
    {
      "epoch": 1.1072092542863046,
      "grad_norm": 16.16337776184082,
      "learning_rate": 4.94041388934795e-05,
      "loss": 1.6209,
      "step": 6700
    },
    {
      "epoch": 1.1088618054121049,
      "grad_norm": 25.227270126342773,
      "learning_rate": 4.9394957674580874e-05,
      "loss": 1.5893,
      "step": 6710
    },
    {
      "epoch": 1.1105143565379054,
      "grad_norm": 11.184518814086914,
      "learning_rate": 4.938577645568226e-05,
      "loss": 1.7018,
      "step": 6720
    },
    {
      "epoch": 1.1121669076637057,
      "grad_norm": 7.978889465332031,
      "learning_rate": 4.937659523678364e-05,
      "loss": 1.7783,
      "step": 6730
    },
    {
      "epoch": 1.1138194587895063,
      "grad_norm": 11.834269523620605,
      "learning_rate": 4.9367414017885015e-05,
      "loss": 1.6588,
      "step": 6740
    },
    {
      "epoch": 1.1154720099153068,
      "grad_norm": 7.372729778289795,
      "learning_rate": 4.93582327989864e-05,
      "loss": 1.7051,
      "step": 6750
    },
    {
      "epoch": 1.1171245610411071,
      "grad_norm": 15.451234817504883,
      "learning_rate": 4.9349051580087774e-05,
      "loss": 1.5759,
      "step": 6760
    },
    {
      "epoch": 1.1187771121669077,
      "grad_norm": 7.165411949157715,
      "learning_rate": 4.9339870361189156e-05,
      "loss": 1.4484,
      "step": 6770
    },
    {
      "epoch": 1.1204296632927082,
      "grad_norm": 9.450922012329102,
      "learning_rate": 4.933068914229053e-05,
      "loss": 1.6673,
      "step": 6780
    },
    {
      "epoch": 1.1220822144185085,
      "grad_norm": 7.094033718109131,
      "learning_rate": 4.9321507923391915e-05,
      "loss": 1.7646,
      "step": 6790
    },
    {
      "epoch": 1.123734765544309,
      "grad_norm": 19.683202743530273,
      "learning_rate": 4.931232670449329e-05,
      "loss": 1.576,
      "step": 6800
    },
    {
      "epoch": 1.1253873166701096,
      "grad_norm": 7.054886341094971,
      "learning_rate": 4.9303145485594673e-05,
      "loss": 1.6599,
      "step": 6810
    },
    {
      "epoch": 1.12703986779591,
      "grad_norm": 11.178540229797363,
      "learning_rate": 4.929396426669605e-05,
      "loss": 1.7556,
      "step": 6820
    },
    {
      "epoch": 1.1286924189217105,
      "grad_norm": 7.649514198303223,
      "learning_rate": 4.9284783047797425e-05,
      "loss": 1.6763,
      "step": 6830
    },
    {
      "epoch": 1.1303449700475108,
      "grad_norm": 27.767852783203125,
      "learning_rate": 4.927560182889881e-05,
      "loss": 1.6954,
      "step": 6840
    },
    {
      "epoch": 1.1319975211733113,
      "grad_norm": 46.87465286254883,
      "learning_rate": 4.9266420610000184e-05,
      "loss": 1.6347,
      "step": 6850
    },
    {
      "epoch": 1.1336500722991119,
      "grad_norm": 11.488672256469727,
      "learning_rate": 4.9257239391101566e-05,
      "loss": 1.6059,
      "step": 6860
    },
    {
      "epoch": 1.1353026234249122,
      "grad_norm": 21.95306396484375,
      "learning_rate": 4.924805817220294e-05,
      "loss": 1.7236,
      "step": 6870
    },
    {
      "epoch": 1.1369551745507127,
      "grad_norm": 7.683815956115723,
      "learning_rate": 4.9238876953304325e-05,
      "loss": 1.7474,
      "step": 6880
    },
    {
      "epoch": 1.138607725676513,
      "grad_norm": 41.58155822753906,
      "learning_rate": 4.92296957344057e-05,
      "loss": 1.6803,
      "step": 6890
    },
    {
      "epoch": 1.1402602768023136,
      "grad_norm": 6.919406890869141,
      "learning_rate": 4.922051451550708e-05,
      "loss": 1.5835,
      "step": 6900
    },
    {
      "epoch": 1.141912827928114,
      "grad_norm": 11.925729751586914,
      "learning_rate": 4.921133329660846e-05,
      "loss": 1.52,
      "step": 6910
    },
    {
      "epoch": 1.1435653790539144,
      "grad_norm": 7.959477424621582,
      "learning_rate": 4.920215207770984e-05,
      "loss": 1.5885,
      "step": 6920
    },
    {
      "epoch": 1.145217930179715,
      "grad_norm": 8.630328178405762,
      "learning_rate": 4.919297085881122e-05,
      "loss": 1.616,
      "step": 6930
    },
    {
      "epoch": 1.1468704813055153,
      "grad_norm": 15.663494110107422,
      "learning_rate": 4.91837896399126e-05,
      "loss": 1.6099,
      "step": 6940
    },
    {
      "epoch": 1.1485230324313158,
      "grad_norm": 9.951146125793457,
      "learning_rate": 4.9174608421013976e-05,
      "loss": 1.6716,
      "step": 6950
    },
    {
      "epoch": 1.1501755835571164,
      "grad_norm": 7.788767337799072,
      "learning_rate": 4.916542720211535e-05,
      "loss": 1.7533,
      "step": 6960
    },
    {
      "epoch": 1.1518281346829167,
      "grad_norm": 18.3308048248291,
      "learning_rate": 4.9156245983216735e-05,
      "loss": 1.8288,
      "step": 6970
    },
    {
      "epoch": 1.1534806858087172,
      "grad_norm": 26.051170349121094,
      "learning_rate": 4.914706476431811e-05,
      "loss": 1.6644,
      "step": 6980
    },
    {
      "epoch": 1.1551332369345177,
      "grad_norm": 12.27104663848877,
      "learning_rate": 4.913788354541949e-05,
      "loss": 1.6927,
      "step": 6990
    },
    {
      "epoch": 1.156785788060318,
      "grad_norm": 18.62868881225586,
      "learning_rate": 4.912870232652087e-05,
      "loss": 1.6167,
      "step": 7000
    },
    {
      "epoch": 1.1584383391861186,
      "grad_norm": 5.9332122802734375,
      "learning_rate": 4.911952110762225e-05,
      "loss": 1.7339,
      "step": 7010
    },
    {
      "epoch": 1.1600908903119191,
      "grad_norm": 12.154903411865234,
      "learning_rate": 4.911033988872363e-05,
      "loss": 1.7497,
      "step": 7020
    },
    {
      "epoch": 1.1617434414377195,
      "grad_norm": 19.24437713623047,
      "learning_rate": 4.910115866982501e-05,
      "loss": 1.7414,
      "step": 7030
    },
    {
      "epoch": 1.16339599256352,
      "grad_norm": 26.495933532714844,
      "learning_rate": 4.9091977450926386e-05,
      "loss": 1.6141,
      "step": 7040
    },
    {
      "epoch": 1.1650485436893203,
      "grad_norm": 6.528138160705566,
      "learning_rate": 4.908279623202777e-05,
      "loss": 1.5709,
      "step": 7050
    },
    {
      "epoch": 1.1667010948151209,
      "grad_norm": 9.650068283081055,
      "learning_rate": 4.9073615013129144e-05,
      "loss": 1.5795,
      "step": 7060
    },
    {
      "epoch": 1.1683536459409214,
      "grad_norm": 8.132938385009766,
      "learning_rate": 4.906443379423053e-05,
      "loss": 1.6091,
      "step": 7070
    },
    {
      "epoch": 1.1700061970667217,
      "grad_norm": 6.303673267364502,
      "learning_rate": 4.90552525753319e-05,
      "loss": 1.5958,
      "step": 7080
    },
    {
      "epoch": 1.1716587481925222,
      "grad_norm": 8.630463600158691,
      "learning_rate": 4.904607135643328e-05,
      "loss": 1.5698,
      "step": 7090
    },
    {
      "epoch": 1.1733112993183226,
      "grad_norm": 12.576354026794434,
      "learning_rate": 4.903689013753466e-05,
      "loss": 1.7056,
      "step": 7100
    },
    {
      "epoch": 1.174963850444123,
      "grad_norm": 56.43805694580078,
      "learning_rate": 4.902770891863604e-05,
      "loss": 1.7336,
      "step": 7110
    },
    {
      "epoch": 1.1766164015699236,
      "grad_norm": 9.944549560546875,
      "learning_rate": 4.901852769973742e-05,
      "loss": 1.6921,
      "step": 7120
    },
    {
      "epoch": 1.178268952695724,
      "grad_norm": 7.986307621002197,
      "learning_rate": 4.9009346480838796e-05,
      "loss": 1.544,
      "step": 7130
    },
    {
      "epoch": 1.1799215038215245,
      "grad_norm": 6.309409141540527,
      "learning_rate": 4.900016526194018e-05,
      "loss": 1.6083,
      "step": 7140
    },
    {
      "epoch": 1.1815740549473248,
      "grad_norm": 36.086181640625,
      "learning_rate": 4.8990984043041554e-05,
      "loss": 1.6664,
      "step": 7150
    },
    {
      "epoch": 1.1832266060731254,
      "grad_norm": 8.67955493927002,
      "learning_rate": 4.898180282414294e-05,
      "loss": 1.574,
      "step": 7160
    },
    {
      "epoch": 1.184879157198926,
      "grad_norm": 7.240811347961426,
      "learning_rate": 4.897262160524431e-05,
      "loss": 1.6852,
      "step": 7170
    },
    {
      "epoch": 1.1865317083247262,
      "grad_norm": 15.414640426635742,
      "learning_rate": 4.8963440386345695e-05,
      "loss": 1.5493,
      "step": 7180
    },
    {
      "epoch": 1.1881842594505267,
      "grad_norm": 9.224071502685547,
      "learning_rate": 4.895425916744708e-05,
      "loss": 1.6202,
      "step": 7190
    },
    {
      "epoch": 1.1898368105763273,
      "grad_norm": 14.074639320373535,
      "learning_rate": 4.8945077948548454e-05,
      "loss": 1.5506,
      "step": 7200
    },
    {
      "epoch": 1.1914893617021276,
      "grad_norm": 11.120677947998047,
      "learning_rate": 4.893589672964983e-05,
      "loss": 1.6851,
      "step": 7210
    },
    {
      "epoch": 1.1931419128279281,
      "grad_norm": 10.850142478942871,
      "learning_rate": 4.8926715510751206e-05,
      "loss": 1.4931,
      "step": 7220
    },
    {
      "epoch": 1.1947944639537287,
      "grad_norm": 18.69784164428711,
      "learning_rate": 4.891753429185259e-05,
      "loss": 1.7118,
      "step": 7230
    },
    {
      "epoch": 1.196447015079529,
      "grad_norm": 8.451671600341797,
      "learning_rate": 4.8908353072953964e-05,
      "loss": 1.6661,
      "step": 7240
    },
    {
      "epoch": 1.1980995662053295,
      "grad_norm": 28.36526107788086,
      "learning_rate": 4.889917185405535e-05,
      "loss": 1.6517,
      "step": 7250
    },
    {
      "epoch": 1.1997521173311299,
      "grad_norm": 8.399593353271484,
      "learning_rate": 4.888999063515672e-05,
      "loss": 1.5306,
      "step": 7260
    },
    {
      "epoch": 1.2014046684569304,
      "grad_norm": 15.712769508361816,
      "learning_rate": 4.8880809416258105e-05,
      "loss": 1.6245,
      "step": 7270
    },
    {
      "epoch": 1.203057219582731,
      "grad_norm": 8.373371124267578,
      "learning_rate": 4.887162819735948e-05,
      "loss": 1.6816,
      "step": 7280
    },
    {
      "epoch": 1.2047097707085312,
      "grad_norm": 11.655881881713867,
      "learning_rate": 4.8862446978460864e-05,
      "loss": 1.76,
      "step": 7290
    },
    {
      "epoch": 1.2063623218343318,
      "grad_norm": 8.07646369934082,
      "learning_rate": 4.8853265759562246e-05,
      "loss": 1.7914,
      "step": 7300
    },
    {
      "epoch": 1.208014872960132,
      "grad_norm": 24.092042922973633,
      "learning_rate": 4.884408454066362e-05,
      "loss": 1.6793,
      "step": 7310
    },
    {
      "epoch": 1.2096674240859326,
      "grad_norm": 10.025934219360352,
      "learning_rate": 4.8834903321765005e-05,
      "loss": 1.7057,
      "step": 7320
    },
    {
      "epoch": 1.2113199752117332,
      "grad_norm": 9.85206413269043,
      "learning_rate": 4.882572210286638e-05,
      "loss": 1.5728,
      "step": 7330
    },
    {
      "epoch": 1.2129725263375335,
      "grad_norm": 13.544818878173828,
      "learning_rate": 4.8816540883967757e-05,
      "loss": 1.783,
      "step": 7340
    },
    {
      "epoch": 1.214625077463334,
      "grad_norm": 17.217933654785156,
      "learning_rate": 4.880735966506913e-05,
      "loss": 1.8045,
      "step": 7350
    },
    {
      "epoch": 1.2162776285891344,
      "grad_norm": 24.209354400634766,
      "learning_rate": 4.8798178446170515e-05,
      "loss": 1.6106,
      "step": 7360
    },
    {
      "epoch": 1.217930179714935,
      "grad_norm": 24.636571884155273,
      "learning_rate": 4.878899722727189e-05,
      "loss": 1.6975,
      "step": 7370
    },
    {
      "epoch": 1.2195827308407354,
      "grad_norm": 16.614473342895508,
      "learning_rate": 4.8779816008373274e-05,
      "loss": 1.7127,
      "step": 7380
    },
    {
      "epoch": 1.2212352819665357,
      "grad_norm": 57.88620376586914,
      "learning_rate": 4.877063478947465e-05,
      "loss": 1.7516,
      "step": 7390
    },
    {
      "epoch": 1.2228878330923363,
      "grad_norm": 28.026735305786133,
      "learning_rate": 4.876145357057603e-05,
      "loss": 1.7203,
      "step": 7400
    },
    {
      "epoch": 1.2245403842181368,
      "grad_norm": 8.524139404296875,
      "learning_rate": 4.8752272351677415e-05,
      "loss": 1.5154,
      "step": 7410
    },
    {
      "epoch": 1.2261929353439371,
      "grad_norm": 11.738126754760742,
      "learning_rate": 4.874309113277879e-05,
      "loss": 1.6059,
      "step": 7420
    },
    {
      "epoch": 1.2278454864697377,
      "grad_norm": 35.693504333496094,
      "learning_rate": 4.873390991388017e-05,
      "loss": 1.716,
      "step": 7430
    },
    {
      "epoch": 1.2294980375955382,
      "grad_norm": 16.629175186157227,
      "learning_rate": 4.872472869498155e-05,
      "loss": 1.7924,
      "step": 7440
    },
    {
      "epoch": 1.2311505887213385,
      "grad_norm": 9.903701782226562,
      "learning_rate": 4.871554747608293e-05,
      "loss": 1.8271,
      "step": 7450
    },
    {
      "epoch": 1.232803139847139,
      "grad_norm": 10.66246509552002,
      "learning_rate": 4.870636625718431e-05,
      "loss": 1.7185,
      "step": 7460
    },
    {
      "epoch": 1.2344556909729394,
      "grad_norm": 8.404884338378906,
      "learning_rate": 4.8697185038285683e-05,
      "loss": 1.7562,
      "step": 7470
    },
    {
      "epoch": 1.23610824209874,
      "grad_norm": 14.62071418762207,
      "learning_rate": 4.868800381938706e-05,
      "loss": 1.7265,
      "step": 7480
    },
    {
      "epoch": 1.2377607932245405,
      "grad_norm": 21.035802841186523,
      "learning_rate": 4.867882260048844e-05,
      "loss": 1.6935,
      "step": 7490
    },
    {
      "epoch": 1.2394133443503408,
      "grad_norm": 13.605401039123535,
      "learning_rate": 4.866964138158982e-05,
      "loss": 1.6896,
      "step": 7500
    },
    {
      "epoch": 1.2410658954761413,
      "grad_norm": 31.078174591064453,
      "learning_rate": 4.86604601626912e-05,
      "loss": 1.6084,
      "step": 7510
    },
    {
      "epoch": 1.2427184466019416,
      "grad_norm": 11.396821975708008,
      "learning_rate": 4.8651278943792576e-05,
      "loss": 1.6015,
      "step": 7520
    },
    {
      "epoch": 1.2443709977277422,
      "grad_norm": 10.65982723236084,
      "learning_rate": 4.864209772489396e-05,
      "loss": 1.6715,
      "step": 7530
    },
    {
      "epoch": 1.2460235488535427,
      "grad_norm": 34.76524353027344,
      "learning_rate": 4.863291650599534e-05,
      "loss": 1.7029,
      "step": 7540
    },
    {
      "epoch": 1.247676099979343,
      "grad_norm": 21.453048706054688,
      "learning_rate": 4.862373528709672e-05,
      "loss": 1.6779,
      "step": 7550
    },
    {
      "epoch": 1.2493286511051436,
      "grad_norm": 9.208992958068848,
      "learning_rate": 4.86145540681981e-05,
      "loss": 1.7251,
      "step": 7560
    },
    {
      "epoch": 1.250981202230944,
      "grad_norm": 9.565735816955566,
      "learning_rate": 4.8605372849299476e-05,
      "loss": 1.6829,
      "step": 7570
    },
    {
      "epoch": 1.2526337533567444,
      "grad_norm": 8.265256881713867,
      "learning_rate": 4.859619163040086e-05,
      "loss": 1.6577,
      "step": 7580
    },
    {
      "epoch": 1.254286304482545,
      "grad_norm": 14.750138282775879,
      "learning_rate": 4.8587010411502234e-05,
      "loss": 1.6659,
      "step": 7590
    },
    {
      "epoch": 1.2559388556083455,
      "grad_norm": 30.3494930267334,
      "learning_rate": 4.857782919260361e-05,
      "loss": 1.5829,
      "step": 7600
    },
    {
      "epoch": 1.2575914067341458,
      "grad_norm": 13.287981033325195,
      "learning_rate": 4.8568647973704986e-05,
      "loss": 1.6724,
      "step": 7610
    },
    {
      "epoch": 1.2592439578599464,
      "grad_norm": 29.33921241760254,
      "learning_rate": 4.855946675480637e-05,
      "loss": 1.5848,
      "step": 7620
    },
    {
      "epoch": 1.2608965089857467,
      "grad_norm": 7.407712459564209,
      "learning_rate": 4.8550285535907745e-05,
      "loss": 1.6882,
      "step": 7630
    },
    {
      "epoch": 1.2625490601115472,
      "grad_norm": 6.948355674743652,
      "learning_rate": 4.854110431700913e-05,
      "loss": 1.6093,
      "step": 7640
    },
    {
      "epoch": 1.2642016112373478,
      "grad_norm": 13.627409934997559,
      "learning_rate": 4.853192309811051e-05,
      "loss": 1.6143,
      "step": 7650
    },
    {
      "epoch": 1.265854162363148,
      "grad_norm": 10.397058486938477,
      "learning_rate": 4.8522741879211886e-05,
      "loss": 1.6205,
      "step": 7660
    },
    {
      "epoch": 1.2675067134889486,
      "grad_norm": 12.780577659606934,
      "learning_rate": 4.851356066031327e-05,
      "loss": 1.6489,
      "step": 7670
    },
    {
      "epoch": 1.269159264614749,
      "grad_norm": 36.843563079833984,
      "learning_rate": 4.8504379441414644e-05,
      "loss": 1.5179,
      "step": 7680
    },
    {
      "epoch": 1.2708118157405495,
      "grad_norm": 8.480043411254883,
      "learning_rate": 4.849519822251603e-05,
      "loss": 1.9982,
      "step": 7690
    },
    {
      "epoch": 1.27246436686635,
      "grad_norm": 8.302135467529297,
      "learning_rate": 4.84860170036174e-05,
      "loss": 1.7339,
      "step": 7700
    },
    {
      "epoch": 1.2741169179921503,
      "grad_norm": 7.312380790710449,
      "learning_rate": 4.8476835784718785e-05,
      "loss": 1.7085,
      "step": 7710
    },
    {
      "epoch": 1.2757694691179509,
      "grad_norm": 9.16114330291748,
      "learning_rate": 4.846765456582016e-05,
      "loss": 1.6013,
      "step": 7720
    },
    {
      "epoch": 1.2774220202437512,
      "grad_norm": 10.258267402648926,
      "learning_rate": 4.845847334692154e-05,
      "loss": 1.5444,
      "step": 7730
    },
    {
      "epoch": 1.2790745713695517,
      "grad_norm": 41.98063278198242,
      "learning_rate": 4.844929212802291e-05,
      "loss": 1.7147,
      "step": 7740
    },
    {
      "epoch": 1.2807271224953523,
      "grad_norm": 5.9125237464904785,
      "learning_rate": 4.8440110909124296e-05,
      "loss": 1.5708,
      "step": 7750
    },
    {
      "epoch": 1.2823796736211526,
      "grad_norm": 10.1311616897583,
      "learning_rate": 4.843092969022568e-05,
      "loss": 1.6326,
      "step": 7760
    },
    {
      "epoch": 1.284032224746953,
      "grad_norm": 9.376750946044922,
      "learning_rate": 4.8421748471327054e-05,
      "loss": 1.7641,
      "step": 7770
    },
    {
      "epoch": 1.2856847758727534,
      "grad_norm": 8.097874641418457,
      "learning_rate": 4.841256725242844e-05,
      "loss": 1.5963,
      "step": 7780
    },
    {
      "epoch": 1.287337326998554,
      "grad_norm": 8.448856353759766,
      "learning_rate": 4.840338603352981e-05,
      "loss": 1.5056,
      "step": 7790
    },
    {
      "epoch": 1.2889898781243545,
      "grad_norm": 10.571368217468262,
      "learning_rate": 4.8394204814631195e-05,
      "loss": 1.7242,
      "step": 7800
    },
    {
      "epoch": 1.290642429250155,
      "grad_norm": 28.546253204345703,
      "learning_rate": 4.838502359573257e-05,
      "loss": 1.7191,
      "step": 7810
    },
    {
      "epoch": 1.2922949803759554,
      "grad_norm": 20.1251220703125,
      "learning_rate": 4.8375842376833954e-05,
      "loss": 1.6916,
      "step": 7820
    },
    {
      "epoch": 1.293947531501756,
      "grad_norm": 10.026655197143555,
      "learning_rate": 4.836666115793533e-05,
      "loss": 1.6829,
      "step": 7830
    },
    {
      "epoch": 1.2956000826275562,
      "grad_norm": 21.374874114990234,
      "learning_rate": 4.835747993903671e-05,
      "loss": 1.5028,
      "step": 7840
    },
    {
      "epoch": 1.2972526337533568,
      "grad_norm": 42.24235916137695,
      "learning_rate": 4.834829872013809e-05,
      "loss": 1.631,
      "step": 7850
    },
    {
      "epoch": 1.2989051848791573,
      "grad_norm": 6.917150974273682,
      "learning_rate": 4.8339117501239464e-05,
      "loss": 1.7458,
      "step": 7860
    },
    {
      "epoch": 1.3005577360049576,
      "grad_norm": 26.261568069458008,
      "learning_rate": 4.8329936282340847e-05,
      "loss": 1.7711,
      "step": 7870
    },
    {
      "epoch": 1.3022102871307581,
      "grad_norm": 5.60324239730835,
      "learning_rate": 4.832075506344222e-05,
      "loss": 1.726,
      "step": 7880
    },
    {
      "epoch": 1.3038628382565585,
      "grad_norm": 12.937496185302734,
      "learning_rate": 4.8311573844543605e-05,
      "loss": 1.7153,
      "step": 7890
    },
    {
      "epoch": 1.305515389382359,
      "grad_norm": 14.33709716796875,
      "learning_rate": 4.830239262564498e-05,
      "loss": 1.713,
      "step": 7900
    },
    {
      "epoch": 1.3071679405081595,
      "grad_norm": 8.725417137145996,
      "learning_rate": 4.8293211406746364e-05,
      "loss": 1.7221,
      "step": 7910
    },
    {
      "epoch": 1.3088204916339599,
      "grad_norm": 13.017943382263184,
      "learning_rate": 4.828403018784774e-05,
      "loss": 1.5677,
      "step": 7920
    },
    {
      "epoch": 1.3104730427597604,
      "grad_norm": 10.267081260681152,
      "learning_rate": 4.827484896894912e-05,
      "loss": 1.6575,
      "step": 7930
    },
    {
      "epoch": 1.3121255938855607,
      "grad_norm": 4.439043045043945,
      "learning_rate": 4.82656677500505e-05,
      "loss": 1.6943,
      "step": 7940
    },
    {
      "epoch": 1.3137781450113613,
      "grad_norm": 9.938862800598145,
      "learning_rate": 4.825648653115188e-05,
      "loss": 1.6691,
      "step": 7950
    },
    {
      "epoch": 1.3154306961371618,
      "grad_norm": 7.4380669593811035,
      "learning_rate": 4.8247305312253256e-05,
      "loss": 1.6049,
      "step": 7960
    },
    {
      "epoch": 1.3170832472629623,
      "grad_norm": 9.574049949645996,
      "learning_rate": 4.823812409335464e-05,
      "loss": 1.5485,
      "step": 7970
    },
    {
      "epoch": 1.3187357983887626,
      "grad_norm": 6.69705057144165,
      "learning_rate": 4.8228942874456015e-05,
      "loss": 1.7664,
      "step": 7980
    },
    {
      "epoch": 1.3203883495145632,
      "grad_norm": 51.60087203979492,
      "learning_rate": 4.821976165555739e-05,
      "loss": 1.7427,
      "step": 7990
    },
    {
      "epoch": 1.3220409006403635,
      "grad_norm": 8.675333023071289,
      "learning_rate": 4.821058043665877e-05,
      "loss": 1.5976,
      "step": 8000
    },
    {
      "epoch": 1.323693451766164,
      "grad_norm": 11.984296798706055,
      "learning_rate": 4.820139921776015e-05,
      "loss": 1.655,
      "step": 8010
    },
    {
      "epoch": 1.3253460028919646,
      "grad_norm": 7.3102617263793945,
      "learning_rate": 4.819221799886153e-05,
      "loss": 1.6596,
      "step": 8020
    },
    {
      "epoch": 1.326998554017765,
      "grad_norm": 5.259336948394775,
      "learning_rate": 4.818303677996291e-05,
      "loss": 1.6986,
      "step": 8030
    },
    {
      "epoch": 1.3286511051435654,
      "grad_norm": 8.154091835021973,
      "learning_rate": 4.817385556106429e-05,
      "loss": 1.7009,
      "step": 8040
    },
    {
      "epoch": 1.3303036562693658,
      "grad_norm": 6.396356582641602,
      "learning_rate": 4.8164674342165666e-05,
      "loss": 1.5735,
      "step": 8050
    },
    {
      "epoch": 1.3319562073951663,
      "grad_norm": 7.4831647872924805,
      "learning_rate": 4.815549312326705e-05,
      "loss": 1.684,
      "step": 8060
    },
    {
      "epoch": 1.3336087585209668,
      "grad_norm": 9.999634742736816,
      "learning_rate": 4.8146311904368425e-05,
      "loss": 1.7047,
      "step": 8070
    },
    {
      "epoch": 1.3352613096467671,
      "grad_norm": 9.42754077911377,
      "learning_rate": 4.813713068546981e-05,
      "loss": 1.7498,
      "step": 8080
    },
    {
      "epoch": 1.3369138607725677,
      "grad_norm": 14.416214942932129,
      "learning_rate": 4.812794946657118e-05,
      "loss": 1.6447,
      "step": 8090
    },
    {
      "epoch": 1.338566411898368,
      "grad_norm": 6.572526454925537,
      "learning_rate": 4.8118768247672566e-05,
      "loss": 1.6883,
      "step": 8100
    },
    {
      "epoch": 1.3402189630241685,
      "grad_norm": 27.258047103881836,
      "learning_rate": 4.810958702877394e-05,
      "loss": 1.606,
      "step": 8110
    },
    {
      "epoch": 1.341871514149969,
      "grad_norm": 6.700786590576172,
      "learning_rate": 4.810040580987532e-05,
      "loss": 1.6751,
      "step": 8120
    },
    {
      "epoch": 1.3435240652757694,
      "grad_norm": 8.797626495361328,
      "learning_rate": 4.80912245909767e-05,
      "loss": 1.6453,
      "step": 8130
    },
    {
      "epoch": 1.34517661640157,
      "grad_norm": 14.59374713897705,
      "learning_rate": 4.8082043372078076e-05,
      "loss": 1.6201,
      "step": 8140
    },
    {
      "epoch": 1.3468291675273703,
      "grad_norm": 12.410117149353027,
      "learning_rate": 4.807286215317946e-05,
      "loss": 1.6923,
      "step": 8150
    },
    {
      "epoch": 1.3484817186531708,
      "grad_norm": 5.754236221313477,
      "learning_rate": 4.8063680934280835e-05,
      "loss": 1.5902,
      "step": 8160
    },
    {
      "epoch": 1.3501342697789713,
      "grad_norm": 21.738666534423828,
      "learning_rate": 4.805449971538222e-05,
      "loss": 1.7867,
      "step": 8170
    },
    {
      "epoch": 1.3517868209047719,
      "grad_norm": 11.151928901672363,
      "learning_rate": 4.804531849648359e-05,
      "loss": 1.6477,
      "step": 8180
    },
    {
      "epoch": 1.3534393720305722,
      "grad_norm": 7.115407466888428,
      "learning_rate": 4.8036137277584976e-05,
      "loss": 1.6847,
      "step": 8190
    },
    {
      "epoch": 1.3550919231563727,
      "grad_norm": 9.613741874694824,
      "learning_rate": 4.802695605868635e-05,
      "loss": 1.7105,
      "step": 8200
    },
    {
      "epoch": 1.356744474282173,
      "grad_norm": 9.931926727294922,
      "learning_rate": 4.8017774839787734e-05,
      "loss": 1.6286,
      "step": 8210
    },
    {
      "epoch": 1.3583970254079736,
      "grad_norm": 46.38539505004883,
      "learning_rate": 4.800859362088912e-05,
      "loss": 1.7295,
      "step": 8220
    },
    {
      "epoch": 1.3600495765337741,
      "grad_norm": 11.234518051147461,
      "learning_rate": 4.799941240199049e-05,
      "loss": 1.6839,
      "step": 8230
    },
    {
      "epoch": 1.3617021276595744,
      "grad_norm": 29.555355072021484,
      "learning_rate": 4.799023118309187e-05,
      "loss": 1.6353,
      "step": 8240
    },
    {
      "epoch": 1.363354678785375,
      "grad_norm": 13.247050285339355,
      "learning_rate": 4.7981049964193244e-05,
      "loss": 1.7446,
      "step": 8250
    },
    {
      "epoch": 1.3650072299111753,
      "grad_norm": 31.702011108398438,
      "learning_rate": 4.797186874529463e-05,
      "loss": 1.6522,
      "step": 8260
    },
    {
      "epoch": 1.3666597810369758,
      "grad_norm": 19.7655029296875,
      "learning_rate": 4.7962687526396e-05,
      "loss": 1.7344,
      "step": 8270
    },
    {
      "epoch": 1.3683123321627764,
      "grad_norm": 31.14563751220703,
      "learning_rate": 4.7953506307497386e-05,
      "loss": 1.6795,
      "step": 8280
    },
    {
      "epoch": 1.3699648832885767,
      "grad_norm": 10.664438247680664,
      "learning_rate": 4.794432508859876e-05,
      "loss": 1.7746,
      "step": 8290
    },
    {
      "epoch": 1.3716174344143772,
      "grad_norm": 14.965887069702148,
      "learning_rate": 4.7935143869700144e-05,
      "loss": 1.6984,
      "step": 8300
    },
    {
      "epoch": 1.3732699855401775,
      "grad_norm": 18.51651954650879,
      "learning_rate": 4.792596265080152e-05,
      "loss": 1.6421,
      "step": 8310
    },
    {
      "epoch": 1.374922536665978,
      "grad_norm": 8.034571647644043,
      "learning_rate": 4.79167814319029e-05,
      "loss": 1.6664,
      "step": 8320
    },
    {
      "epoch": 1.3765750877917786,
      "grad_norm": 11.700427055358887,
      "learning_rate": 4.7907600213004285e-05,
      "loss": 1.6122,
      "step": 8330
    },
    {
      "epoch": 1.378227638917579,
      "grad_norm": 7.673017501831055,
      "learning_rate": 4.789841899410566e-05,
      "loss": 1.6424,
      "step": 8340
    },
    {
      "epoch": 1.3798801900433795,
      "grad_norm": 4.827954292297363,
      "learning_rate": 4.7889237775207044e-05,
      "loss": 1.6598,
      "step": 8350
    },
    {
      "epoch": 1.3815327411691798,
      "grad_norm": 11.37001895904541,
      "learning_rate": 4.788005655630842e-05,
      "loss": 1.6502,
      "step": 8360
    },
    {
      "epoch": 1.3831852922949803,
      "grad_norm": 7.629409313201904,
      "learning_rate": 4.7870875337409795e-05,
      "loss": 1.7068,
      "step": 8370
    },
    {
      "epoch": 1.3848378434207809,
      "grad_norm": 21.615652084350586,
      "learning_rate": 4.786169411851117e-05,
      "loss": 1.5153,
      "step": 8380
    },
    {
      "epoch": 1.3864903945465814,
      "grad_norm": 9.277376174926758,
      "learning_rate": 4.7852512899612554e-05,
      "loss": 1.6594,
      "step": 8390
    },
    {
      "epoch": 1.3881429456723817,
      "grad_norm": 11.108504295349121,
      "learning_rate": 4.784333168071393e-05,
      "loss": 1.6291,
      "step": 8400
    },
    {
      "epoch": 1.3897954967981823,
      "grad_norm": 9.173980712890625,
      "learning_rate": 4.783415046181531e-05,
      "loss": 1.6998,
      "step": 8410
    },
    {
      "epoch": 1.3914480479239826,
      "grad_norm": 8.537729263305664,
      "learning_rate": 4.782496924291669e-05,
      "loss": 1.5601,
      "step": 8420
    },
    {
      "epoch": 1.3931005990497831,
      "grad_norm": 10.740826606750488,
      "learning_rate": 4.781578802401807e-05,
      "loss": 1.4999,
      "step": 8430
    },
    {
      "epoch": 1.3947531501755837,
      "grad_norm": 9.543370246887207,
      "learning_rate": 4.7806606805119453e-05,
      "loss": 1.6375,
      "step": 8440
    },
    {
      "epoch": 1.396405701301384,
      "grad_norm": 9.984679222106934,
      "learning_rate": 4.779742558622083e-05,
      "loss": 1.5013,
      "step": 8450
    },
    {
      "epoch": 1.3980582524271845,
      "grad_norm": 9.201363563537598,
      "learning_rate": 4.778824436732221e-05,
      "loss": 1.5612,
      "step": 8460
    },
    {
      "epoch": 1.3997108035529848,
      "grad_norm": 17.85106086730957,
      "learning_rate": 4.777906314842359e-05,
      "loss": 1.7215,
      "step": 8470
    },
    {
      "epoch": 1.4013633546787854,
      "grad_norm": 8.731996536254883,
      "learning_rate": 4.776988192952497e-05,
      "loss": 1.4592,
      "step": 8480
    },
    {
      "epoch": 1.403015905804586,
      "grad_norm": 12.443288803100586,
      "learning_rate": 4.7760700710626346e-05,
      "loss": 1.6517,
      "step": 8490
    },
    {
      "epoch": 1.4046684569303862,
      "grad_norm": 4.665842056274414,
      "learning_rate": 4.775151949172772e-05,
      "loss": 1.5992,
      "step": 8500
    },
    {
      "epoch": 1.4063210080561868,
      "grad_norm": 8.88791275024414,
      "learning_rate": 4.77423382728291e-05,
      "loss": 1.7198,
      "step": 8510
    },
    {
      "epoch": 1.407973559181987,
      "grad_norm": 7.840359687805176,
      "learning_rate": 4.773315705393048e-05,
      "loss": 1.633,
      "step": 8520
    },
    {
      "epoch": 1.4096261103077876,
      "grad_norm": 42.78237533569336,
      "learning_rate": 4.7723975835031857e-05,
      "loss": 1.7015,
      "step": 8530
    },
    {
      "epoch": 1.4112786614335882,
      "grad_norm": 31.31909942626953,
      "learning_rate": 4.771479461613324e-05,
      "loss": 1.8256,
      "step": 8540
    },
    {
      "epoch": 1.4129312125593887,
      "grad_norm": 9.952142715454102,
      "learning_rate": 4.7705613397234615e-05,
      "loss": 1.7117,
      "step": 8550
    },
    {
      "epoch": 1.414583763685189,
      "grad_norm": 10.509651184082031,
      "learning_rate": 4.7696432178336e-05,
      "loss": 1.6899,
      "step": 8560
    },
    {
      "epoch": 1.4162363148109893,
      "grad_norm": 13.65649700164795,
      "learning_rate": 4.768725095943738e-05,
      "loss": 1.6753,
      "step": 8570
    },
    {
      "epoch": 1.4178888659367899,
      "grad_norm": 9.891709327697754,
      "learning_rate": 4.7678069740538756e-05,
      "loss": 1.6893,
      "step": 8580
    },
    {
      "epoch": 1.4195414170625904,
      "grad_norm": 6.675931453704834,
      "learning_rate": 4.766888852164014e-05,
      "loss": 1.6134,
      "step": 8590
    },
    {
      "epoch": 1.421193968188391,
      "grad_norm": 8.769001007080078,
      "learning_rate": 4.7659707302741515e-05,
      "loss": 1.6775,
      "step": 8600
    },
    {
      "epoch": 1.4228465193141913,
      "grad_norm": 9.261868476867676,
      "learning_rate": 4.76505260838429e-05,
      "loss": 1.7286,
      "step": 8610
    },
    {
      "epoch": 1.4244990704399918,
      "grad_norm": 9.514866828918457,
      "learning_rate": 4.764134486494427e-05,
      "loss": 1.5825,
      "step": 8620
    },
    {
      "epoch": 1.4261516215657921,
      "grad_norm": 11.191073417663574,
      "learning_rate": 4.763216364604565e-05,
      "loss": 1.699,
      "step": 8630
    },
    {
      "epoch": 1.4278041726915927,
      "grad_norm": 7.584383487701416,
      "learning_rate": 4.7622982427147025e-05,
      "loss": 1.7905,
      "step": 8640
    },
    {
      "epoch": 1.4294567238173932,
      "grad_norm": 7.346413612365723,
      "learning_rate": 4.761380120824841e-05,
      "loss": 1.7111,
      "step": 8650
    },
    {
      "epoch": 1.4311092749431935,
      "grad_norm": 12.299200057983398,
      "learning_rate": 4.7604619989349783e-05,
      "loss": 1.6624,
      "step": 8660
    },
    {
      "epoch": 1.432761826068994,
      "grad_norm": 28.52010154724121,
      "learning_rate": 4.7595438770451166e-05,
      "loss": 1.7085,
      "step": 8670
    },
    {
      "epoch": 1.4344143771947944,
      "grad_norm": 20.789058685302734,
      "learning_rate": 4.758625755155255e-05,
      "loss": 1.7056,
      "step": 8680
    },
    {
      "epoch": 1.436066928320595,
      "grad_norm": 13.599698066711426,
      "learning_rate": 4.7577076332653925e-05,
      "loss": 1.5663,
      "step": 8690
    },
    {
      "epoch": 1.4377194794463954,
      "grad_norm": 15.90054702758789,
      "learning_rate": 4.756789511375531e-05,
      "loss": 1.7743,
      "step": 8700
    },
    {
      "epoch": 1.4393720305721958,
      "grad_norm": 21.3345890045166,
      "learning_rate": 4.755871389485668e-05,
      "loss": 1.7051,
      "step": 8710
    },
    {
      "epoch": 1.4410245816979963,
      "grad_norm": 10.17492389678955,
      "learning_rate": 4.7549532675958066e-05,
      "loss": 1.6872,
      "step": 8720
    },
    {
      "epoch": 1.4426771328237966,
      "grad_norm": 5.552209377288818,
      "learning_rate": 4.754035145705944e-05,
      "loss": 1.6925,
      "step": 8730
    },
    {
      "epoch": 1.4443296839495972,
      "grad_norm": 17.34671401977539,
      "learning_rate": 4.7531170238160824e-05,
      "loss": 1.5588,
      "step": 8740
    },
    {
      "epoch": 1.4459822350753977,
      "grad_norm": 7.345114231109619,
      "learning_rate": 4.75219890192622e-05,
      "loss": 1.6705,
      "step": 8750
    },
    {
      "epoch": 1.4476347862011982,
      "grad_norm": 7.885573387145996,
      "learning_rate": 4.7512807800363576e-05,
      "loss": 1.5831,
      "step": 8760
    },
    {
      "epoch": 1.4492873373269985,
      "grad_norm": 26.88873291015625,
      "learning_rate": 4.750362658146495e-05,
      "loss": 1.6182,
      "step": 8770
    },
    {
      "epoch": 1.450939888452799,
      "grad_norm": 6.879226207733154,
      "learning_rate": 4.7494445362566334e-05,
      "loss": 1.8231,
      "step": 8780
    },
    {
      "epoch": 1.4525924395785994,
      "grad_norm": 15.858061790466309,
      "learning_rate": 4.748526414366772e-05,
      "loss": 1.5887,
      "step": 8790
    },
    {
      "epoch": 1.4542449907044,
      "grad_norm": 6.855731010437012,
      "learning_rate": 4.747608292476909e-05,
      "loss": 1.7003,
      "step": 8800
    },
    {
      "epoch": 1.4558975418302005,
      "grad_norm": 32.11435317993164,
      "learning_rate": 4.7466901705870475e-05,
      "loss": 1.682,
      "step": 8810
    },
    {
      "epoch": 1.4575500929560008,
      "grad_norm": 10.053679466247559,
      "learning_rate": 4.745772048697185e-05,
      "loss": 1.6652,
      "step": 8820
    },
    {
      "epoch": 1.4592026440818013,
      "grad_norm": 9.131372451782227,
      "learning_rate": 4.7448539268073234e-05,
      "loss": 1.592,
      "step": 8830
    },
    {
      "epoch": 1.4608551952076017,
      "grad_norm": 7.688348293304443,
      "learning_rate": 4.743935804917461e-05,
      "loss": 1.6068,
      "step": 8840
    },
    {
      "epoch": 1.4625077463334022,
      "grad_norm": 6.706598281860352,
      "learning_rate": 4.743017683027599e-05,
      "loss": 1.5156,
      "step": 8850
    },
    {
      "epoch": 1.4641602974592027,
      "grad_norm": 10.917819023132324,
      "learning_rate": 4.742099561137737e-05,
      "loss": 1.5903,
      "step": 8860
    },
    {
      "epoch": 1.465812848585003,
      "grad_norm": 21.588254928588867,
      "learning_rate": 4.741181439247875e-05,
      "loss": 1.6745,
      "step": 8870
    },
    {
      "epoch": 1.4674653997108036,
      "grad_norm": 20.283750534057617,
      "learning_rate": 4.740263317358013e-05,
      "loss": 1.5973,
      "step": 8880
    },
    {
      "epoch": 1.469117950836604,
      "grad_norm": 21.922521591186523,
      "learning_rate": 4.73934519546815e-05,
      "loss": 1.5737,
      "step": 8890
    },
    {
      "epoch": 1.4707705019624044,
      "grad_norm": 8.129682540893555,
      "learning_rate": 4.7384270735782885e-05,
      "loss": 1.7585,
      "step": 8900
    },
    {
      "epoch": 1.472423053088205,
      "grad_norm": 24.920692443847656,
      "learning_rate": 4.737508951688426e-05,
      "loss": 1.6209,
      "step": 8910
    },
    {
      "epoch": 1.4740756042140053,
      "grad_norm": 10.06619930267334,
      "learning_rate": 4.7365908297985644e-05,
      "loss": 1.686,
      "step": 8920
    },
    {
      "epoch": 1.4757281553398058,
      "grad_norm": 13.576436042785645,
      "learning_rate": 4.735672707908702e-05,
      "loss": 1.6421,
      "step": 8930
    },
    {
      "epoch": 1.4773807064656062,
      "grad_norm": 13.374571800231934,
      "learning_rate": 4.73475458601884e-05,
      "loss": 1.7228,
      "step": 8940
    },
    {
      "epoch": 1.4790332575914067,
      "grad_norm": 7.154562473297119,
      "learning_rate": 4.733836464128978e-05,
      "loss": 1.6549,
      "step": 8950
    },
    {
      "epoch": 1.4806858087172072,
      "grad_norm": 15.508869171142578,
      "learning_rate": 4.732918342239116e-05,
      "loss": 1.675,
      "step": 8960
    },
    {
      "epoch": 1.4823383598430078,
      "grad_norm": 10.355504989624023,
      "learning_rate": 4.732000220349254e-05,
      "loss": 1.6165,
      "step": 8970
    },
    {
      "epoch": 1.483990910968808,
      "grad_norm": 9.313375473022461,
      "learning_rate": 4.731082098459392e-05,
      "loss": 1.6555,
      "step": 8980
    },
    {
      "epoch": 1.4856434620946086,
      "grad_norm": 11.76880168914795,
      "learning_rate": 4.7301639765695295e-05,
      "loss": 1.7011,
      "step": 8990
    },
    {
      "epoch": 1.487296013220409,
      "grad_norm": 9.017674446105957,
      "learning_rate": 4.729245854679668e-05,
      "loss": 1.696,
      "step": 9000
    },
    {
      "epoch": 1.4889485643462095,
      "grad_norm": 30.168582916259766,
      "learning_rate": 4.7283277327898054e-05,
      "loss": 1.6945,
      "step": 9010
    },
    {
      "epoch": 1.49060111547201,
      "grad_norm": 28.239177703857422,
      "learning_rate": 4.727409610899943e-05,
      "loss": 1.5782,
      "step": 9020
    },
    {
      "epoch": 1.4922536665978103,
      "grad_norm": 7.943751335144043,
      "learning_rate": 4.726491489010081e-05,
      "loss": 1.6587,
      "step": 9030
    },
    {
      "epoch": 1.4939062177236109,
      "grad_norm": 19.905715942382812,
      "learning_rate": 4.725573367120219e-05,
      "loss": 1.5575,
      "step": 9040
    },
    {
      "epoch": 1.4955587688494112,
      "grad_norm": 10.705276489257812,
      "learning_rate": 4.724655245230357e-05,
      "loss": 1.59,
      "step": 9050
    },
    {
      "epoch": 1.4972113199752117,
      "grad_norm": 5.611824035644531,
      "learning_rate": 4.7237371233404947e-05,
      "loss": 1.63,
      "step": 9060
    },
    {
      "epoch": 1.4988638711010123,
      "grad_norm": 14.214753150939941,
      "learning_rate": 4.722819001450633e-05,
      "loss": 1.6358,
      "step": 9070
    },
    {
      "epoch": 1.5005164222268128,
      "grad_norm": 12.516484260559082,
      "learning_rate": 4.7219008795607705e-05,
      "loss": 1.7702,
      "step": 9080
    },
    {
      "epoch": 1.5021689733526131,
      "grad_norm": 7.916662693023682,
      "learning_rate": 4.720982757670909e-05,
      "loss": 1.677,
      "step": 9090
    },
    {
      "epoch": 1.5038215244784134,
      "grad_norm": 8.450326919555664,
      "learning_rate": 4.7200646357810464e-05,
      "loss": 1.6053,
      "step": 9100
    },
    {
      "epoch": 1.505474075604214,
      "grad_norm": 15.917101860046387,
      "learning_rate": 4.7191465138911846e-05,
      "loss": 1.6568,
      "step": 9110
    },
    {
      "epoch": 1.5071266267300145,
      "grad_norm": 10.047342300415039,
      "learning_rate": 4.718228392001322e-05,
      "loss": 1.7409,
      "step": 9120
    },
    {
      "epoch": 1.508779177855815,
      "grad_norm": 8.033047676086426,
      "learning_rate": 4.7173102701114605e-05,
      "loss": 1.6499,
      "step": 9130
    },
    {
      "epoch": 1.5104317289816154,
      "grad_norm": 8.630041122436523,
      "learning_rate": 4.716392148221598e-05,
      "loss": 1.5798,
      "step": 9140
    },
    {
      "epoch": 1.5120842801074157,
      "grad_norm": 18.576356887817383,
      "learning_rate": 4.7154740263317356e-05,
      "loss": 1.7651,
      "step": 9150
    },
    {
      "epoch": 1.5137368312332162,
      "grad_norm": 11.440652847290039,
      "learning_rate": 4.714555904441874e-05,
      "loss": 1.6616,
      "step": 9160
    },
    {
      "epoch": 1.5153893823590168,
      "grad_norm": 11.976500511169434,
      "learning_rate": 4.7136377825520115e-05,
      "loss": 1.673,
      "step": 9170
    },
    {
      "epoch": 1.5170419334848173,
      "grad_norm": 7.650939464569092,
      "learning_rate": 4.71271966066215e-05,
      "loss": 1.5774,
      "step": 9180
    },
    {
      "epoch": 1.5186944846106176,
      "grad_norm": 8.787154197692871,
      "learning_rate": 4.711801538772287e-05,
      "loss": 1.581,
      "step": 9190
    },
    {
      "epoch": 1.520347035736418,
      "grad_norm": 11.962265014648438,
      "learning_rate": 4.7108834168824256e-05,
      "loss": 1.5368,
      "step": 9200
    },
    {
      "epoch": 1.5219995868622185,
      "grad_norm": 8.907034873962402,
      "learning_rate": 4.709965294992563e-05,
      "loss": 1.5875,
      "step": 9210
    },
    {
      "epoch": 1.523652137988019,
      "grad_norm": 13.364984512329102,
      "learning_rate": 4.7090471731027014e-05,
      "loss": 1.6646,
      "step": 9220
    },
    {
      "epoch": 1.5253046891138196,
      "grad_norm": 7.773966312408447,
      "learning_rate": 4.708129051212839e-05,
      "loss": 1.5193,
      "step": 9230
    },
    {
      "epoch": 1.5269572402396199,
      "grad_norm": 6.378410339355469,
      "learning_rate": 4.707210929322977e-05,
      "loss": 1.7244,
      "step": 9240
    },
    {
      "epoch": 1.5286097913654204,
      "grad_norm": 15.127907752990723,
      "learning_rate": 4.7062928074331156e-05,
      "loss": 1.6726,
      "step": 9250
    },
    {
      "epoch": 1.5302623424912207,
      "grad_norm": 16.456928253173828,
      "learning_rate": 4.705374685543253e-05,
      "loss": 1.7078,
      "step": 9260
    },
    {
      "epoch": 1.5319148936170213,
      "grad_norm": 11.239667892456055,
      "learning_rate": 4.704456563653391e-05,
      "loss": 1.6224,
      "step": 9270
    },
    {
      "epoch": 1.5335674447428218,
      "grad_norm": 26.4625186920166,
      "learning_rate": 4.703538441763528e-05,
      "loss": 1.6131,
      "step": 9280
    },
    {
      "epoch": 1.5352199958686223,
      "grad_norm": 7.001266956329346,
      "learning_rate": 4.7026203198736666e-05,
      "loss": 1.6851,
      "step": 9290
    },
    {
      "epoch": 1.5368725469944227,
      "grad_norm": 4.010128974914551,
      "learning_rate": 4.701702197983804e-05,
      "loss": 1.5748,
      "step": 9300
    },
    {
      "epoch": 1.538525098120223,
      "grad_norm": 5.630429267883301,
      "learning_rate": 4.7007840760939424e-05,
      "loss": 1.7095,
      "step": 9310
    },
    {
      "epoch": 1.5401776492460235,
      "grad_norm": 10.235610961914062,
      "learning_rate": 4.69986595420408e-05,
      "loss": 1.7108,
      "step": 9320
    },
    {
      "epoch": 1.541830200371824,
      "grad_norm": 27.188196182250977,
      "learning_rate": 4.698947832314218e-05,
      "loss": 1.6722,
      "step": 9330
    },
    {
      "epoch": 1.5434827514976246,
      "grad_norm": 9.516741752624512,
      "learning_rate": 4.698029710424356e-05,
      "loss": 1.5869,
      "step": 9340
    },
    {
      "epoch": 1.545135302623425,
      "grad_norm": 16.91329574584961,
      "learning_rate": 4.697111588534494e-05,
      "loss": 1.8084,
      "step": 9350
    },
    {
      "epoch": 1.5467878537492252,
      "grad_norm": 26.68416976928711,
      "learning_rate": 4.6961934666446324e-05,
      "loss": 1.5328,
      "step": 9360
    },
    {
      "epoch": 1.5484404048750258,
      "grad_norm": 7.9282073974609375,
      "learning_rate": 4.69527534475477e-05,
      "loss": 1.6512,
      "step": 9370
    },
    {
      "epoch": 1.5500929560008263,
      "grad_norm": 10.000015258789062,
      "learning_rate": 4.694357222864908e-05,
      "loss": 1.6211,
      "step": 9380
    },
    {
      "epoch": 1.5517455071266268,
      "grad_norm": 51.1877555847168,
      "learning_rate": 4.693439100975046e-05,
      "loss": 1.5976,
      "step": 9390
    },
    {
      "epoch": 1.5533980582524272,
      "grad_norm": 14.719481468200684,
      "learning_rate": 4.6925209790851834e-05,
      "loss": 1.5779,
      "step": 9400
    },
    {
      "epoch": 1.5550506093782275,
      "grad_norm": 9.398426055908203,
      "learning_rate": 4.691602857195321e-05,
      "loss": 1.5547,
      "step": 9410
    },
    {
      "epoch": 1.556703160504028,
      "grad_norm": 13.369629859924316,
      "learning_rate": 4.690684735305459e-05,
      "loss": 1.625,
      "step": 9420
    },
    {
      "epoch": 1.5583557116298286,
      "grad_norm": 7.2386674880981445,
      "learning_rate": 4.689766613415597e-05,
      "loss": 1.6962,
      "step": 9430
    },
    {
      "epoch": 1.560008262755629,
      "grad_norm": 6.752237319946289,
      "learning_rate": 4.688848491525735e-05,
      "loss": 1.6528,
      "step": 9440
    },
    {
      "epoch": 1.5616608138814294,
      "grad_norm": 16.95864486694336,
      "learning_rate": 4.687930369635873e-05,
      "loss": 1.6891,
      "step": 9450
    },
    {
      "epoch": 1.56331336500723,
      "grad_norm": 29.162368774414062,
      "learning_rate": 4.687012247746011e-05,
      "loss": 1.6036,
      "step": 9460
    },
    {
      "epoch": 1.5649659161330303,
      "grad_norm": 8.65707778930664,
      "learning_rate": 4.686094125856149e-05,
      "loss": 1.749,
      "step": 9470
    },
    {
      "epoch": 1.5666184672588308,
      "grad_norm": 8.36571979522705,
      "learning_rate": 4.685176003966287e-05,
      "loss": 1.597,
      "step": 9480
    },
    {
      "epoch": 1.5682710183846313,
      "grad_norm": 10.10719108581543,
      "learning_rate": 4.684257882076425e-05,
      "loss": 1.7746,
      "step": 9490
    },
    {
      "epoch": 1.5699235695104319,
      "grad_norm": 7.152252197265625,
      "learning_rate": 4.683339760186563e-05,
      "loss": 1.604,
      "step": 9500
    },
    {
      "epoch": 1.5715761206362322,
      "grad_norm": 7.251582145690918,
      "learning_rate": 4.682421638296701e-05,
      "loss": 1.599,
      "step": 9510
    },
    {
      "epoch": 1.5732286717620325,
      "grad_norm": 14.877774238586426,
      "learning_rate": 4.6815035164068385e-05,
      "loss": 1.5955,
      "step": 9520
    },
    {
      "epoch": 1.574881222887833,
      "grad_norm": 7.264001846313477,
      "learning_rate": 4.680585394516976e-05,
      "loss": 1.5446,
      "step": 9530
    },
    {
      "epoch": 1.5765337740136336,
      "grad_norm": 31.467344284057617,
      "learning_rate": 4.679667272627114e-05,
      "loss": 1.6975,
      "step": 9540
    },
    {
      "epoch": 1.5781863251394341,
      "grad_norm": 6.771696090698242,
      "learning_rate": 4.678749150737252e-05,
      "loss": 1.5352,
      "step": 9550
    },
    {
      "epoch": 1.5798388762652344,
      "grad_norm": 16.742000579833984,
      "learning_rate": 4.6778310288473895e-05,
      "loss": 1.7136,
      "step": 9560
    },
    {
      "epoch": 1.5814914273910348,
      "grad_norm": 29.50790023803711,
      "learning_rate": 4.676912906957528e-05,
      "loss": 1.6819,
      "step": 9570
    },
    {
      "epoch": 1.5831439785168353,
      "grad_norm": 15.069239616394043,
      "learning_rate": 4.675994785067666e-05,
      "loss": 1.5652,
      "step": 9580
    },
    {
      "epoch": 1.5847965296426358,
      "grad_norm": 10.359492301940918,
      "learning_rate": 4.6750766631778036e-05,
      "loss": 1.7212,
      "step": 9590
    },
    {
      "epoch": 1.5864490807684364,
      "grad_norm": 14.912342071533203,
      "learning_rate": 4.674158541287942e-05,
      "loss": 1.7613,
      "step": 9600
    },
    {
      "epoch": 1.5881016318942367,
      "grad_norm": 15.676534652709961,
      "learning_rate": 4.6732404193980795e-05,
      "loss": 1.5133,
      "step": 9610
    },
    {
      "epoch": 1.5897541830200372,
      "grad_norm": 6.48195743560791,
      "learning_rate": 4.672322297508218e-05,
      "loss": 1.6921,
      "step": 9620
    },
    {
      "epoch": 1.5914067341458376,
      "grad_norm": 10.377370834350586,
      "learning_rate": 4.6714041756183553e-05,
      "loss": 1.6365,
      "step": 9630
    },
    {
      "epoch": 1.593059285271638,
      "grad_norm": 6.733672142028809,
      "learning_rate": 4.6704860537284936e-05,
      "loss": 1.6149,
      "step": 9640
    },
    {
      "epoch": 1.5947118363974386,
      "grad_norm": 13.145439147949219,
      "learning_rate": 4.669567931838631e-05,
      "loss": 1.7273,
      "step": 9650
    },
    {
      "epoch": 1.596364387523239,
      "grad_norm": 23.76000213623047,
      "learning_rate": 4.668649809948769e-05,
      "loss": 1.5651,
      "step": 9660
    },
    {
      "epoch": 1.5980169386490395,
      "grad_norm": 10.361004829406738,
      "learning_rate": 4.6677316880589064e-05,
      "loss": 1.679,
      "step": 9670
    },
    {
      "epoch": 1.5996694897748398,
      "grad_norm": 7.273348808288574,
      "learning_rate": 4.6668135661690446e-05,
      "loss": 1.6333,
      "step": 9680
    },
    {
      "epoch": 1.6013220409006403,
      "grad_norm": 5.444534778594971,
      "learning_rate": 4.665895444279182e-05,
      "loss": 1.6189,
      "step": 9690
    },
    {
      "epoch": 1.6029745920264409,
      "grad_norm": 11.040206909179688,
      "learning_rate": 4.6649773223893205e-05,
      "loss": 1.5579,
      "step": 9700
    },
    {
      "epoch": 1.6046271431522414,
      "grad_norm": 8.984979629516602,
      "learning_rate": 4.664059200499459e-05,
      "loss": 1.5681,
      "step": 9710
    },
    {
      "epoch": 1.6062796942780417,
      "grad_norm": 8.338875770568848,
      "learning_rate": 4.663141078609596e-05,
      "loss": 1.6869,
      "step": 9720
    },
    {
      "epoch": 1.607932245403842,
      "grad_norm": 8.329351425170898,
      "learning_rate": 4.6622229567197346e-05,
      "loss": 1.4787,
      "step": 9730
    },
    {
      "epoch": 1.6095847965296426,
      "grad_norm": 9.657584190368652,
      "learning_rate": 4.661304834829872e-05,
      "loss": 1.7737,
      "step": 9740
    },
    {
      "epoch": 1.6112373476554431,
      "grad_norm": 9.302536964416504,
      "learning_rate": 4.6603867129400104e-05,
      "loss": 1.6792,
      "step": 9750
    },
    {
      "epoch": 1.6128898987812437,
      "grad_norm": 7.729322910308838,
      "learning_rate": 4.659468591050148e-05,
      "loss": 1.6573,
      "step": 9760
    },
    {
      "epoch": 1.614542449907044,
      "grad_norm": 8.147910118103027,
      "learning_rate": 4.658550469160286e-05,
      "loss": 1.5995,
      "step": 9770
    },
    {
      "epoch": 1.6161950010328443,
      "grad_norm": 7.201321601867676,
      "learning_rate": 4.657632347270424e-05,
      "loss": 1.6919,
      "step": 9780
    },
    {
      "epoch": 1.6178475521586448,
      "grad_norm": 8.503691673278809,
      "learning_rate": 4.6567142253805615e-05,
      "loss": 1.6173,
      "step": 9790
    },
    {
      "epoch": 1.6195001032844454,
      "grad_norm": 21.700334548950195,
      "learning_rate": 4.655796103490699e-05,
      "loss": 1.5879,
      "step": 9800
    },
    {
      "epoch": 1.621152654410246,
      "grad_norm": 18.57465171813965,
      "learning_rate": 4.654877981600837e-05,
      "loss": 1.6012,
      "step": 9810
    },
    {
      "epoch": 1.6228052055360462,
      "grad_norm": 6.389706611633301,
      "learning_rate": 4.6539598597109756e-05,
      "loss": 1.6895,
      "step": 9820
    },
    {
      "epoch": 1.6244577566618468,
      "grad_norm": 6.799575328826904,
      "learning_rate": 4.653041737821113e-05,
      "loss": 1.6661,
      "step": 9830
    },
    {
      "epoch": 1.626110307787647,
      "grad_norm": 6.362144947052002,
      "learning_rate": 4.6521236159312514e-05,
      "loss": 1.6193,
      "step": 9840
    },
    {
      "epoch": 1.6277628589134476,
      "grad_norm": 18.13498306274414,
      "learning_rate": 4.651205494041389e-05,
      "loss": 1.6063,
      "step": 9850
    },
    {
      "epoch": 1.6294154100392482,
      "grad_norm": 15.989859580993652,
      "learning_rate": 4.650287372151527e-05,
      "loss": 1.6043,
      "step": 9860
    },
    {
      "epoch": 1.6310679611650487,
      "grad_norm": 30.983396530151367,
      "learning_rate": 4.649369250261665e-05,
      "loss": 1.8636,
      "step": 9870
    },
    {
      "epoch": 1.632720512290849,
      "grad_norm": 9.010120391845703,
      "learning_rate": 4.648451128371803e-05,
      "loss": 1.6487,
      "step": 9880
    },
    {
      "epoch": 1.6343730634166493,
      "grad_norm": 6.68414831161499,
      "learning_rate": 4.647533006481941e-05,
      "loss": 1.6081,
      "step": 9890
    },
    {
      "epoch": 1.6360256145424499,
      "grad_norm": 9.62691593170166,
      "learning_rate": 4.646614884592079e-05,
      "loss": 1.5234,
      "step": 9900
    },
    {
      "epoch": 1.6376781656682504,
      "grad_norm": 7.007033348083496,
      "learning_rate": 4.6456967627022166e-05,
      "loss": 1.584,
      "step": 9910
    },
    {
      "epoch": 1.639330716794051,
      "grad_norm": 17.63177490234375,
      "learning_rate": 4.644778640812354e-05,
      "loss": 1.4388,
      "step": 9920
    },
    {
      "epoch": 1.6409832679198513,
      "grad_norm": 19.87029266357422,
      "learning_rate": 4.6438605189224924e-05,
      "loss": 1.5444,
      "step": 9930
    },
    {
      "epoch": 1.6426358190456516,
      "grad_norm": 28.15524673461914,
      "learning_rate": 4.64294239703263e-05,
      "loss": 1.6218,
      "step": 9940
    },
    {
      "epoch": 1.6442883701714521,
      "grad_norm": 22.747636795043945,
      "learning_rate": 4.642024275142768e-05,
      "loss": 1.5842,
      "step": 9950
    },
    {
      "epoch": 1.6459409212972527,
      "grad_norm": 12.167838096618652,
      "learning_rate": 4.641106153252906e-05,
      "loss": 1.5029,
      "step": 9960
    },
    {
      "epoch": 1.6475934724230532,
      "grad_norm": 9.144806861877441,
      "learning_rate": 4.640188031363044e-05,
      "loss": 1.6125,
      "step": 9970
    },
    {
      "epoch": 1.6492460235488535,
      "grad_norm": 11.655991554260254,
      "learning_rate": 4.639269909473182e-05,
      "loss": 1.5915,
      "step": 9980
    },
    {
      "epoch": 1.6508985746746538,
      "grad_norm": 29.67504119873047,
      "learning_rate": 4.63835178758332e-05,
      "loss": 1.6178,
      "step": 9990
    },
    {
      "epoch": 1.6525511258004544,
      "grad_norm": 18.424442291259766,
      "learning_rate": 4.6374336656934575e-05,
      "loss": 1.46,
      "step": 10000
    },
    {
      "epoch": 1.654203676926255,
      "grad_norm": 13.235391616821289,
      "learning_rate": 4.636515543803596e-05,
      "loss": 1.5958,
      "step": 10010
    },
    {
      "epoch": 1.6558562280520555,
      "grad_norm": 10.103015899658203,
      "learning_rate": 4.6355974219137334e-05,
      "loss": 1.7109,
      "step": 10020
    },
    {
      "epoch": 1.6575087791778558,
      "grad_norm": 10.416471481323242,
      "learning_rate": 4.6346793000238717e-05,
      "loss": 1.7097,
      "step": 10030
    },
    {
      "epoch": 1.6591613303036563,
      "grad_norm": 12.568252563476562,
      "learning_rate": 4.633761178134009e-05,
      "loss": 1.6269,
      "step": 10040
    },
    {
      "epoch": 1.6608138814294566,
      "grad_norm": 65.68162536621094,
      "learning_rate": 4.632843056244147e-05,
      "loss": 1.5952,
      "step": 10050
    },
    {
      "epoch": 1.6624664325552572,
      "grad_norm": 29.50044822692871,
      "learning_rate": 4.631924934354285e-05,
      "loss": 1.529,
      "step": 10060
    },
    {
      "epoch": 1.6641189836810577,
      "grad_norm": 33.595272064208984,
      "learning_rate": 4.631006812464423e-05,
      "loss": 1.5998,
      "step": 10070
    },
    {
      "epoch": 1.6657715348068582,
      "grad_norm": 8.986908912658691,
      "learning_rate": 4.630088690574561e-05,
      "loss": 1.7082,
      "step": 10080
    },
    {
      "epoch": 1.6674240859326586,
      "grad_norm": 12.171062469482422,
      "learning_rate": 4.6291705686846985e-05,
      "loss": 1.6,
      "step": 10090
    },
    {
      "epoch": 1.6690766370584589,
      "grad_norm": 28.181230545043945,
      "learning_rate": 4.628252446794837e-05,
      "loss": 1.6835,
      "step": 10100
    },
    {
      "epoch": 1.6707291881842594,
      "grad_norm": 13.5730562210083,
      "learning_rate": 4.6273343249049744e-05,
      "loss": 1.7877,
      "step": 10110
    },
    {
      "epoch": 1.67238173931006,
      "grad_norm": 13.940990447998047,
      "learning_rate": 4.6264162030151126e-05,
      "loss": 1.7061,
      "step": 10120
    },
    {
      "epoch": 1.6740342904358605,
      "grad_norm": 10.991562843322754,
      "learning_rate": 4.62549808112525e-05,
      "loss": 1.6114,
      "step": 10130
    },
    {
      "epoch": 1.6756868415616608,
      "grad_norm": 24.29729461669922,
      "learning_rate": 4.6245799592353885e-05,
      "loss": 1.6884,
      "step": 10140
    },
    {
      "epoch": 1.6773393926874611,
      "grad_norm": 11.747648239135742,
      "learning_rate": 4.623661837345526e-05,
      "loss": 1.6264,
      "step": 10150
    },
    {
      "epoch": 1.6789919438132617,
      "grad_norm": 10.77138614654541,
      "learning_rate": 4.6227437154556643e-05,
      "loss": 1.6901,
      "step": 10160
    },
    {
      "epoch": 1.6806444949390622,
      "grad_norm": 13.280838012695312,
      "learning_rate": 4.621825593565802e-05,
      "loss": 1.7122,
      "step": 10170
    },
    {
      "epoch": 1.6822970460648627,
      "grad_norm": 12.801493644714355,
      "learning_rate": 4.6209074716759395e-05,
      "loss": 1.5463,
      "step": 10180
    },
    {
      "epoch": 1.683949597190663,
      "grad_norm": 11.150125503540039,
      "learning_rate": 4.619989349786078e-05,
      "loss": 1.6348,
      "step": 10190
    },
    {
      "epoch": 1.6856021483164634,
      "grad_norm": 8.492911338806152,
      "learning_rate": 4.6190712278962154e-05,
      "loss": 1.545,
      "step": 10200
    },
    {
      "epoch": 1.687254699442264,
      "grad_norm": 7.204408168792725,
      "learning_rate": 4.6181531060063536e-05,
      "loss": 1.6983,
      "step": 10210
    },
    {
      "epoch": 1.6889072505680645,
      "grad_norm": 53.46291732788086,
      "learning_rate": 4.617234984116491e-05,
      "loss": 1.6098,
      "step": 10220
    },
    {
      "epoch": 1.690559801693865,
      "grad_norm": 5.803036212921143,
      "learning_rate": 4.6163168622266295e-05,
      "loss": 1.5576,
      "step": 10230
    },
    {
      "epoch": 1.6922123528196653,
      "grad_norm": 20.07267951965332,
      "learning_rate": 4.615398740336767e-05,
      "loss": 1.7001,
      "step": 10240
    },
    {
      "epoch": 1.6938649039454658,
      "grad_norm": 9.970727920532227,
      "learning_rate": 4.614480618446905e-05,
      "loss": 1.6336,
      "step": 10250
    },
    {
      "epoch": 1.6955174550712662,
      "grad_norm": 9.943961143493652,
      "learning_rate": 4.613562496557043e-05,
      "loss": 1.7672,
      "step": 10260
    },
    {
      "epoch": 1.6971700061970667,
      "grad_norm": 13.809309959411621,
      "learning_rate": 4.612644374667181e-05,
      "loss": 1.6802,
      "step": 10270
    },
    {
      "epoch": 1.6988225573228672,
      "grad_norm": 5.0867462158203125,
      "learning_rate": 4.6117262527773194e-05,
      "loss": 1.6411,
      "step": 10280
    },
    {
      "epoch": 1.7004751084486678,
      "grad_norm": 15.1895751953125,
      "learning_rate": 4.610808130887457e-05,
      "loss": 1.6637,
      "step": 10290
    },
    {
      "epoch": 1.702127659574468,
      "grad_norm": 5.628765106201172,
      "learning_rate": 4.6098900089975946e-05,
      "loss": 1.6516,
      "step": 10300
    },
    {
      "epoch": 1.7037802107002684,
      "grad_norm": 9.596723556518555,
      "learning_rate": 4.608971887107732e-05,
      "loss": 1.5923,
      "step": 10310
    },
    {
      "epoch": 1.705432761826069,
      "grad_norm": 6.9686455726623535,
      "learning_rate": 4.6080537652178705e-05,
      "loss": 1.5587,
      "step": 10320
    },
    {
      "epoch": 1.7070853129518695,
      "grad_norm": 12.421316146850586,
      "learning_rate": 4.607135643328008e-05,
      "loss": 1.6681,
      "step": 10330
    },
    {
      "epoch": 1.70873786407767,
      "grad_norm": 21.520915985107422,
      "learning_rate": 4.606217521438146e-05,
      "loss": 1.6138,
      "step": 10340
    },
    {
      "epoch": 1.7103904152034703,
      "grad_norm": 8.50918960571289,
      "learning_rate": 4.605299399548284e-05,
      "loss": 1.7301,
      "step": 10350
    },
    {
      "epoch": 1.7120429663292707,
      "grad_norm": 10.39163875579834,
      "learning_rate": 4.604381277658422e-05,
      "loss": 1.6752,
      "step": 10360
    },
    {
      "epoch": 1.7136955174550712,
      "grad_norm": 6.190021514892578,
      "learning_rate": 4.60346315576856e-05,
      "loss": 1.5676,
      "step": 10370
    },
    {
      "epoch": 1.7153480685808717,
      "grad_norm": 9.422684669494629,
      "learning_rate": 4.602545033878698e-05,
      "loss": 1.5699,
      "step": 10380
    },
    {
      "epoch": 1.7170006197066723,
      "grad_norm": 7.96755838394165,
      "learning_rate": 4.601626911988836e-05,
      "loss": 1.6369,
      "step": 10390
    },
    {
      "epoch": 1.7186531708324726,
      "grad_norm": 6.596837997436523,
      "learning_rate": 4.600708790098974e-05,
      "loss": 1.6232,
      "step": 10400
    },
    {
      "epoch": 1.7203057219582731,
      "grad_norm": 18.394948959350586,
      "learning_rate": 4.599790668209112e-05,
      "loss": 1.6457,
      "step": 10410
    },
    {
      "epoch": 1.7219582730840735,
      "grad_norm": 9.514167785644531,
      "learning_rate": 4.59887254631925e-05,
      "loss": 1.5002,
      "step": 10420
    },
    {
      "epoch": 1.723610824209874,
      "grad_norm": 15.90142822265625,
      "learning_rate": 4.597954424429387e-05,
      "loss": 1.6668,
      "step": 10430
    },
    {
      "epoch": 1.7252633753356745,
      "grad_norm": 14.313610076904297,
      "learning_rate": 4.597036302539525e-05,
      "loss": 1.6253,
      "step": 10440
    },
    {
      "epoch": 1.7269159264614748,
      "grad_norm": 5.3916144371032715,
      "learning_rate": 4.596118180649663e-05,
      "loss": 1.5947,
      "step": 10450
    },
    {
      "epoch": 1.7285684775872754,
      "grad_norm": 9.352823257446289,
      "learning_rate": 4.595200058759801e-05,
      "loss": 1.502,
      "step": 10460
    },
    {
      "epoch": 1.7302210287130757,
      "grad_norm": 27.035520553588867,
      "learning_rate": 4.594281936869939e-05,
      "loss": 1.7585,
      "step": 10470
    },
    {
      "epoch": 1.7318735798388762,
      "grad_norm": 5.941956520080566,
      "learning_rate": 4.5933638149800766e-05,
      "loss": 1.8005,
      "step": 10480
    },
    {
      "epoch": 1.7335261309646768,
      "grad_norm": 7.316133499145508,
      "learning_rate": 4.592445693090215e-05,
      "loss": 1.5243,
      "step": 10490
    },
    {
      "epoch": 1.7351786820904773,
      "grad_norm": 7.501877784729004,
      "learning_rate": 4.591527571200353e-05,
      "loss": 1.5457,
      "step": 10500
    },
    {
      "epoch": 1.7368312332162776,
      "grad_norm": 8.191024780273438,
      "learning_rate": 4.590609449310491e-05,
      "loss": 1.6192,
      "step": 10510
    },
    {
      "epoch": 1.738483784342078,
      "grad_norm": 7.398270130157471,
      "learning_rate": 4.589691327420629e-05,
      "loss": 1.713,
      "step": 10520
    },
    {
      "epoch": 1.7401363354678785,
      "grad_norm": 7.048802375793457,
      "learning_rate": 4.5887732055307665e-05,
      "loss": 1.7066,
      "step": 10530
    },
    {
      "epoch": 1.741788886593679,
      "grad_norm": 6.611818313598633,
      "learning_rate": 4.587855083640905e-05,
      "loss": 1.5422,
      "step": 10540
    },
    {
      "epoch": 1.7434414377194796,
      "grad_norm": 11.106283187866211,
      "learning_rate": 4.5869369617510424e-05,
      "loss": 1.6228,
      "step": 10550
    },
    {
      "epoch": 1.7450939888452799,
      "grad_norm": 17.441743850708008,
      "learning_rate": 4.58601883986118e-05,
      "loss": 1.6286,
      "step": 10560
    },
    {
      "epoch": 1.7467465399710802,
      "grad_norm": 17.710969924926758,
      "learning_rate": 4.585100717971318e-05,
      "loss": 1.5696,
      "step": 10570
    },
    {
      "epoch": 1.7483990910968807,
      "grad_norm": 12.676824569702148,
      "learning_rate": 4.584182596081456e-05,
      "loss": 1.6077,
      "step": 10580
    },
    {
      "epoch": 1.7500516422226813,
      "grad_norm": 8.125213623046875,
      "learning_rate": 4.5832644741915934e-05,
      "loss": 1.7088,
      "step": 10590
    },
    {
      "epoch": 1.7517041933484818,
      "grad_norm": 7.755391597747803,
      "learning_rate": 4.582346352301732e-05,
      "loss": 1.638,
      "step": 10600
    },
    {
      "epoch": 1.7533567444742821,
      "grad_norm": 19.214263916015625,
      "learning_rate": 4.58142823041187e-05,
      "loss": 1.5969,
      "step": 10610
    },
    {
      "epoch": 1.7550092956000827,
      "grad_norm": 32.08985900878906,
      "learning_rate": 4.5805101085220075e-05,
      "loss": 1.7992,
      "step": 10620
    },
    {
      "epoch": 1.756661846725883,
      "grad_norm": 6.143647193908691,
      "learning_rate": 4.579591986632146e-05,
      "loss": 1.6176,
      "step": 10630
    },
    {
      "epoch": 1.7583143978516835,
      "grad_norm": 12.872699737548828,
      "learning_rate": 4.5786738647422834e-05,
      "loss": 1.5866,
      "step": 10640
    },
    {
      "epoch": 1.759966948977484,
      "grad_norm": 7.934300899505615,
      "learning_rate": 4.5777557428524216e-05,
      "loss": 1.5976,
      "step": 10650
    },
    {
      "epoch": 1.7616195001032846,
      "grad_norm": 6.824174404144287,
      "learning_rate": 4.576837620962559e-05,
      "loss": 1.5911,
      "step": 10660
    },
    {
      "epoch": 1.763272051229085,
      "grad_norm": 7.65330696105957,
      "learning_rate": 4.5759194990726975e-05,
      "loss": 1.5306,
      "step": 10670
    },
    {
      "epoch": 1.7649246023548852,
      "grad_norm": 4.150641918182373,
      "learning_rate": 4.575001377182835e-05,
      "loss": 1.6023,
      "step": 10680
    },
    {
      "epoch": 1.7665771534806858,
      "grad_norm": 15.337532997131348,
      "learning_rate": 4.574083255292973e-05,
      "loss": 1.5932,
      "step": 10690
    },
    {
      "epoch": 1.7682297046064863,
      "grad_norm": 10.681439399719238,
      "learning_rate": 4.573165133403111e-05,
      "loss": 1.7345,
      "step": 10700
    },
    {
      "epoch": 1.7698822557322869,
      "grad_norm": 21.317352294921875,
      "learning_rate": 4.5722470115132485e-05,
      "loss": 1.6743,
      "step": 10710
    },
    {
      "epoch": 1.7715348068580872,
      "grad_norm": 25.704092025756836,
      "learning_rate": 4.571328889623386e-05,
      "loss": 1.3808,
      "step": 10720
    },
    {
      "epoch": 1.7731873579838875,
      "grad_norm": 7.380794525146484,
      "learning_rate": 4.5704107677335244e-05,
      "loss": 1.6079,
      "step": 10730
    },
    {
      "epoch": 1.774839909109688,
      "grad_norm": 12.925481796264648,
      "learning_rate": 4.5694926458436626e-05,
      "loss": 1.656,
      "step": 10740
    },
    {
      "epoch": 1.7764924602354886,
      "grad_norm": 62.927032470703125,
      "learning_rate": 4.5685745239538e-05,
      "loss": 1.7919,
      "step": 10750
    },
    {
      "epoch": 1.778145011361289,
      "grad_norm": 13.639461517333984,
      "learning_rate": 4.5676564020639385e-05,
      "loss": 1.6534,
      "step": 10760
    },
    {
      "epoch": 1.7797975624870894,
      "grad_norm": 7.046122074127197,
      "learning_rate": 4.566738280174076e-05,
      "loss": 1.549,
      "step": 10770
    },
    {
      "epoch": 1.7814501136128897,
      "grad_norm": 12.599806785583496,
      "learning_rate": 4.565820158284214e-05,
      "loss": 1.6526,
      "step": 10780
    },
    {
      "epoch": 1.7831026647386903,
      "grad_norm": 8.485870361328125,
      "learning_rate": 4.564902036394352e-05,
      "loss": 1.7176,
      "step": 10790
    },
    {
      "epoch": 1.7847552158644908,
      "grad_norm": 16.592004776000977,
      "learning_rate": 4.56398391450449e-05,
      "loss": 1.5627,
      "step": 10800
    },
    {
      "epoch": 1.7864077669902914,
      "grad_norm": 11.336044311523438,
      "learning_rate": 4.563065792614628e-05,
      "loss": 1.6818,
      "step": 10810
    },
    {
      "epoch": 1.7880603181160917,
      "grad_norm": 8.369160652160645,
      "learning_rate": 4.5621476707247653e-05,
      "loss": 1.6238,
      "step": 10820
    },
    {
      "epoch": 1.7897128692418922,
      "grad_norm": 9.106146812438965,
      "learning_rate": 4.5612295488349036e-05,
      "loss": 1.4657,
      "step": 10830
    },
    {
      "epoch": 1.7913654203676925,
      "grad_norm": 18.96662712097168,
      "learning_rate": 4.560311426945041e-05,
      "loss": 1.6315,
      "step": 10840
    },
    {
      "epoch": 1.793017971493493,
      "grad_norm": 9.854429244995117,
      "learning_rate": 4.5593933050551795e-05,
      "loss": 1.685,
      "step": 10850
    },
    {
      "epoch": 1.7946705226192936,
      "grad_norm": 20.518041610717773,
      "learning_rate": 4.558475183165317e-05,
      "loss": 1.6613,
      "step": 10860
    },
    {
      "epoch": 1.7963230737450941,
      "grad_norm": 16.10544204711914,
      "learning_rate": 4.557557061275455e-05,
      "loss": 1.6003,
      "step": 10870
    },
    {
      "epoch": 1.7979756248708945,
      "grad_norm": 28.675254821777344,
      "learning_rate": 4.556638939385593e-05,
      "loss": 1.5901,
      "step": 10880
    },
    {
      "epoch": 1.7996281759966948,
      "grad_norm": 27.472984313964844,
      "learning_rate": 4.555720817495731e-05,
      "loss": 1.6684,
      "step": 10890
    },
    {
      "epoch": 1.8012807271224953,
      "grad_norm": 7.602771759033203,
      "learning_rate": 4.554802695605869e-05,
      "loss": 1.8076,
      "step": 10900
    },
    {
      "epoch": 1.8029332782482959,
      "grad_norm": 8.444791793823242,
      "learning_rate": 4.553884573716007e-05,
      "loss": 1.7259,
      "step": 10910
    },
    {
      "epoch": 1.8045858293740964,
      "grad_norm": 8.157547950744629,
      "learning_rate": 4.5529664518261446e-05,
      "loss": 1.7186,
      "step": 10920
    },
    {
      "epoch": 1.8062383804998967,
      "grad_norm": 7.917896270751953,
      "learning_rate": 4.552048329936283e-05,
      "loss": 1.6898,
      "step": 10930
    },
    {
      "epoch": 1.807890931625697,
      "grad_norm": 12.761302947998047,
      "learning_rate": 4.5511302080464204e-05,
      "loss": 1.6634,
      "step": 10940
    },
    {
      "epoch": 1.8095434827514976,
      "grad_norm": 5.908732891082764,
      "learning_rate": 4.550212086156558e-05,
      "loss": 1.6357,
      "step": 10950
    },
    {
      "epoch": 1.811196033877298,
      "grad_norm": 6.7102766036987305,
      "learning_rate": 4.549293964266696e-05,
      "loss": 1.6942,
      "step": 10960
    },
    {
      "epoch": 1.8128485850030986,
      "grad_norm": 117.56930541992188,
      "learning_rate": 4.548375842376834e-05,
      "loss": 1.6321,
      "step": 10970
    },
    {
      "epoch": 1.814501136128899,
      "grad_norm": 9.186972618103027,
      "learning_rate": 4.547457720486972e-05,
      "loss": 1.6335,
      "step": 10980
    },
    {
      "epoch": 1.8161536872546993,
      "grad_norm": 12.602619171142578,
      "learning_rate": 4.54653959859711e-05,
      "loss": 1.6513,
      "step": 10990
    },
    {
      "epoch": 1.8178062383804998,
      "grad_norm": 6.2811279296875,
      "learning_rate": 4.545621476707248e-05,
      "loss": 1.5188,
      "step": 11000
    },
    {
      "epoch": 1.8194587895063004,
      "grad_norm": 15.940876960754395,
      "learning_rate": 4.5447033548173856e-05,
      "loss": 1.7214,
      "step": 11010
    },
    {
      "epoch": 1.821111340632101,
      "grad_norm": 8.485485076904297,
      "learning_rate": 4.543785232927524e-05,
      "loss": 1.7004,
      "step": 11020
    },
    {
      "epoch": 1.8227638917579012,
      "grad_norm": 37.532020568847656,
      "learning_rate": 4.5428671110376614e-05,
      "loss": 1.5909,
      "step": 11030
    },
    {
      "epoch": 1.8244164428837017,
      "grad_norm": 10.41656494140625,
      "learning_rate": 4.5419489891478e-05,
      "loss": 1.5895,
      "step": 11040
    },
    {
      "epoch": 1.826068994009502,
      "grad_norm": 5.756435394287109,
      "learning_rate": 4.541030867257937e-05,
      "loss": 1.6121,
      "step": 11050
    },
    {
      "epoch": 1.8277215451353026,
      "grad_norm": 16.178802490234375,
      "learning_rate": 4.5401127453680755e-05,
      "loss": 1.5903,
      "step": 11060
    },
    {
      "epoch": 1.8293740962611031,
      "grad_norm": 10.916646957397461,
      "learning_rate": 4.539194623478213e-05,
      "loss": 1.7007,
      "step": 11070
    },
    {
      "epoch": 1.8310266473869037,
      "grad_norm": 6.1422224044799805,
      "learning_rate": 4.538276501588351e-05,
      "loss": 1.6766,
      "step": 11080
    },
    {
      "epoch": 1.832679198512704,
      "grad_norm": 15.099671363830566,
      "learning_rate": 4.537358379698489e-05,
      "loss": 1.6593,
      "step": 11090
    },
    {
      "epoch": 1.8343317496385043,
      "grad_norm": 19.930587768554688,
      "learning_rate": 4.5364402578086266e-05,
      "loss": 1.6625,
      "step": 11100
    },
    {
      "epoch": 1.8359843007643049,
      "grad_norm": 9.389049530029297,
      "learning_rate": 4.535522135918765e-05,
      "loss": 1.6434,
      "step": 11110
    },
    {
      "epoch": 1.8376368518901054,
      "grad_norm": 10.457796096801758,
      "learning_rate": 4.5346040140289024e-05,
      "loss": 1.6743,
      "step": 11120
    },
    {
      "epoch": 1.839289403015906,
      "grad_norm": 9.54992961883545,
      "learning_rate": 4.533685892139041e-05,
      "loss": 1.6617,
      "step": 11130
    },
    {
      "epoch": 1.8409419541417062,
      "grad_norm": 19.115575790405273,
      "learning_rate": 4.532767770249178e-05,
      "loss": 1.6583,
      "step": 11140
    },
    {
      "epoch": 1.8425945052675066,
      "grad_norm": 9.527848243713379,
      "learning_rate": 4.5318496483593165e-05,
      "loss": 1.5821,
      "step": 11150
    },
    {
      "epoch": 1.844247056393307,
      "grad_norm": 9.40245246887207,
      "learning_rate": 4.530931526469454e-05,
      "loss": 1.6554,
      "step": 11160
    },
    {
      "epoch": 1.8458996075191076,
      "grad_norm": 18.67144012451172,
      "learning_rate": 4.5300134045795924e-05,
      "loss": 1.7357,
      "step": 11170
    },
    {
      "epoch": 1.8475521586449082,
      "grad_norm": 25.492530822753906,
      "learning_rate": 4.5290952826897306e-05,
      "loss": 1.6508,
      "step": 11180
    },
    {
      "epoch": 1.8492047097707085,
      "grad_norm": 13.621424674987793,
      "learning_rate": 4.528177160799868e-05,
      "loss": 1.7588,
      "step": 11190
    },
    {
      "epoch": 1.850857260896509,
      "grad_norm": 12.195111274719238,
      "learning_rate": 4.527259038910006e-05,
      "loss": 1.6934,
      "step": 11200
    },
    {
      "epoch": 1.8525098120223094,
      "grad_norm": 9.302632331848145,
      "learning_rate": 4.526340917020144e-05,
      "loss": 1.5907,
      "step": 11210
    },
    {
      "epoch": 1.85416236314811,
      "grad_norm": 6.693589687347412,
      "learning_rate": 4.5254227951302817e-05,
      "loss": 1.7577,
      "step": 11220
    },
    {
      "epoch": 1.8558149142739104,
      "grad_norm": 7.7038116455078125,
      "learning_rate": 4.524504673240419e-05,
      "loss": 1.574,
      "step": 11230
    },
    {
      "epoch": 1.8574674653997107,
      "grad_norm": 10.348893165588379,
      "learning_rate": 4.5235865513505575e-05,
      "loss": 1.5739,
      "step": 11240
    },
    {
      "epoch": 1.8591200165255113,
      "grad_norm": 12.050825119018555,
      "learning_rate": 4.522668429460695e-05,
      "loss": 1.658,
      "step": 11250
    },
    {
      "epoch": 1.8607725676513116,
      "grad_norm": 47.17829132080078,
      "learning_rate": 4.5217503075708334e-05,
      "loss": 1.4564,
      "step": 11260
    },
    {
      "epoch": 1.8624251187771121,
      "grad_norm": 8.023752212524414,
      "learning_rate": 4.520832185680971e-05,
      "loss": 1.5252,
      "step": 11270
    },
    {
      "epoch": 1.8640776699029127,
      "grad_norm": 11.899310111999512,
      "learning_rate": 4.519914063791109e-05,
      "loss": 1.7046,
      "step": 11280
    },
    {
      "epoch": 1.8657302210287132,
      "grad_norm": 7.167046546936035,
      "learning_rate": 4.518995941901247e-05,
      "loss": 1.6141,
      "step": 11290
    },
    {
      "epoch": 1.8673827721545135,
      "grad_norm": 42.9312858581543,
      "learning_rate": 4.518077820011385e-05,
      "loss": 1.6363,
      "step": 11300
    },
    {
      "epoch": 1.8690353232803139,
      "grad_norm": 12.356457710266113,
      "learning_rate": 4.517159698121523e-05,
      "loss": 1.7502,
      "step": 11310
    },
    {
      "epoch": 1.8706878744061144,
      "grad_norm": 5.740696430206299,
      "learning_rate": 4.516241576231661e-05,
      "loss": 1.6006,
      "step": 11320
    },
    {
      "epoch": 1.872340425531915,
      "grad_norm": 5.727828502655029,
      "learning_rate": 4.5153234543417985e-05,
      "loss": 1.6271,
      "step": 11330
    },
    {
      "epoch": 1.8739929766577155,
      "grad_norm": 10.743067741394043,
      "learning_rate": 4.514405332451937e-05,
      "loss": 1.5755,
      "step": 11340
    },
    {
      "epoch": 1.8756455277835158,
      "grad_norm": 14.965692520141602,
      "learning_rate": 4.5134872105620743e-05,
      "loss": 1.53,
      "step": 11350
    },
    {
      "epoch": 1.877298078909316,
      "grad_norm": 11.617084503173828,
      "learning_rate": 4.512569088672212e-05,
      "loss": 1.6813,
      "step": 11360
    },
    {
      "epoch": 1.8789506300351166,
      "grad_norm": 253.05593872070312,
      "learning_rate": 4.51165096678235e-05,
      "loss": 1.5946,
      "step": 11370
    },
    {
      "epoch": 1.8806031811609172,
      "grad_norm": 9.766687393188477,
      "learning_rate": 4.510732844892488e-05,
      "loss": 1.5899,
      "step": 11380
    },
    {
      "epoch": 1.8822557322867177,
      "grad_norm": 22.282573699951172,
      "learning_rate": 4.509814723002626e-05,
      "loss": 1.6214,
      "step": 11390
    },
    {
      "epoch": 1.883908283412518,
      "grad_norm": 7.993402481079102,
      "learning_rate": 4.5088966011127636e-05,
      "loss": 1.4566,
      "step": 11400
    },
    {
      "epoch": 1.8855608345383186,
      "grad_norm": 8.523614883422852,
      "learning_rate": 4.507978479222902e-05,
      "loss": 1.6375,
      "step": 11410
    },
    {
      "epoch": 1.887213385664119,
      "grad_norm": 32.4261474609375,
      "learning_rate": 4.50706035733304e-05,
      "loss": 1.6616,
      "step": 11420
    },
    {
      "epoch": 1.8888659367899194,
      "grad_norm": 9.417035102844238,
      "learning_rate": 4.506142235443178e-05,
      "loss": 1.5917,
      "step": 11430
    },
    {
      "epoch": 1.89051848791572,
      "grad_norm": 46.04423904418945,
      "learning_rate": 4.505224113553316e-05,
      "loss": 1.6917,
      "step": 11440
    },
    {
      "epoch": 1.8921710390415205,
      "grad_norm": 14.668797492980957,
      "learning_rate": 4.5043059916634536e-05,
      "loss": 1.5452,
      "step": 11450
    },
    {
      "epoch": 1.8938235901673208,
      "grad_norm": 10.886049270629883,
      "learning_rate": 4.503387869773591e-05,
      "loss": 1.5244,
      "step": 11460
    },
    {
      "epoch": 1.8954761412931211,
      "grad_norm": 10.956747055053711,
      "learning_rate": 4.5024697478837294e-05,
      "loss": 1.5967,
      "step": 11470
    },
    {
      "epoch": 1.8971286924189217,
      "grad_norm": 15.368989944458008,
      "learning_rate": 4.501551625993867e-05,
      "loss": 1.5721,
      "step": 11480
    },
    {
      "epoch": 1.8987812435447222,
      "grad_norm": 8.719017028808594,
      "learning_rate": 4.5006335041040046e-05,
      "loss": 1.6545,
      "step": 11490
    },
    {
      "epoch": 1.9004337946705228,
      "grad_norm": 10.52745246887207,
      "learning_rate": 4.499715382214143e-05,
      "loss": 1.6317,
      "step": 11500
    },
    {
      "epoch": 1.902086345796323,
      "grad_norm": 27.014604568481445,
      "learning_rate": 4.4987972603242805e-05,
      "loss": 1.4449,
      "step": 11510
    },
    {
      "epoch": 1.9037388969221234,
      "grad_norm": 14.527423858642578,
      "learning_rate": 4.497879138434419e-05,
      "loss": 1.6464,
      "step": 11520
    },
    {
      "epoch": 1.905391448047924,
      "grad_norm": 28.174190521240234,
      "learning_rate": 4.496961016544557e-05,
      "loss": 1.5219,
      "step": 11530
    },
    {
      "epoch": 1.9070439991737245,
      "grad_norm": 10.394495964050293,
      "learning_rate": 4.4960428946546946e-05,
      "loss": 1.6479,
      "step": 11540
    },
    {
      "epoch": 1.908696550299525,
      "grad_norm": 29.0770263671875,
      "learning_rate": 4.495124772764833e-05,
      "loss": 1.5932,
      "step": 11550
    },
    {
      "epoch": 1.9103491014253253,
      "grad_norm": 6.238829612731934,
      "learning_rate": 4.4942066508749704e-05,
      "loss": 1.5789,
      "step": 11560
    },
    {
      "epoch": 1.9120016525511256,
      "grad_norm": 7.41868257522583,
      "learning_rate": 4.493288528985109e-05,
      "loss": 1.5565,
      "step": 11570
    },
    {
      "epoch": 1.9136542036769262,
      "grad_norm": 18.375661849975586,
      "learning_rate": 4.492370407095246e-05,
      "loss": 1.6516,
      "step": 11580
    },
    {
      "epoch": 1.9153067548027267,
      "grad_norm": 11.252774238586426,
      "learning_rate": 4.491452285205384e-05,
      "loss": 1.4842,
      "step": 11590
    },
    {
      "epoch": 1.9169593059285273,
      "grad_norm": 13.860923767089844,
      "learning_rate": 4.490534163315522e-05,
      "loss": 1.6212,
      "step": 11600
    },
    {
      "epoch": 1.9186118570543276,
      "grad_norm": 6.298261642456055,
      "learning_rate": 4.48961604142566e-05,
      "loss": 1.5285,
      "step": 11610
    },
    {
      "epoch": 1.9202644081801281,
      "grad_norm": 19.169706344604492,
      "learning_rate": 4.488697919535797e-05,
      "loss": 1.6163,
      "step": 11620
    },
    {
      "epoch": 1.9219169593059284,
      "grad_norm": 15.898575782775879,
      "learning_rate": 4.4877797976459356e-05,
      "loss": 1.609,
      "step": 11630
    },
    {
      "epoch": 1.923569510431729,
      "grad_norm": 16.889341354370117,
      "learning_rate": 4.486861675756074e-05,
      "loss": 1.5827,
      "step": 11640
    },
    {
      "epoch": 1.9252220615575295,
      "grad_norm": 8.887215614318848,
      "learning_rate": 4.4859435538662114e-05,
      "loss": 1.581,
      "step": 11650
    },
    {
      "epoch": 1.92687461268333,
      "grad_norm": 27.01335906982422,
      "learning_rate": 4.48502543197635e-05,
      "loss": 1.7247,
      "step": 11660
    },
    {
      "epoch": 1.9285271638091304,
      "grad_norm": 7.634554862976074,
      "learning_rate": 4.484107310086487e-05,
      "loss": 1.5232,
      "step": 11670
    },
    {
      "epoch": 1.9301797149349307,
      "grad_norm": 17.882055282592773,
      "learning_rate": 4.4831891881966255e-05,
      "loss": 1.5038,
      "step": 11680
    },
    {
      "epoch": 1.9318322660607312,
      "grad_norm": 24.722803115844727,
      "learning_rate": 4.482271066306763e-05,
      "loss": 1.4617,
      "step": 11690
    },
    {
      "epoch": 1.9334848171865318,
      "grad_norm": 62.85787582397461,
      "learning_rate": 4.4813529444169014e-05,
      "loss": 1.6762,
      "step": 11700
    },
    {
      "epoch": 1.9351373683123323,
      "grad_norm": 20.34491729736328,
      "learning_rate": 4.480434822527039e-05,
      "loss": 1.7011,
      "step": 11710
    },
    {
      "epoch": 1.9367899194381326,
      "grad_norm": 8.389205932617188,
      "learning_rate": 4.4795167006371765e-05,
      "loss": 1.6886,
      "step": 11720
    },
    {
      "epoch": 1.938442470563933,
      "grad_norm": 7.912249565124512,
      "learning_rate": 4.478598578747315e-05,
      "loss": 1.6878,
      "step": 11730
    },
    {
      "epoch": 1.9400950216897335,
      "grad_norm": 26.233190536499023,
      "learning_rate": 4.4776804568574524e-05,
      "loss": 1.7822,
      "step": 11740
    },
    {
      "epoch": 1.941747572815534,
      "grad_norm": 24.81481170654297,
      "learning_rate": 4.4767623349675907e-05,
      "loss": 1.7393,
      "step": 11750
    },
    {
      "epoch": 1.9434001239413345,
      "grad_norm": 19.601308822631836,
      "learning_rate": 4.475844213077728e-05,
      "loss": 1.5535,
      "step": 11760
    },
    {
      "epoch": 1.9450526750671349,
      "grad_norm": 9.827634811401367,
      "learning_rate": 4.4749260911878665e-05,
      "loss": 1.5759,
      "step": 11770
    },
    {
      "epoch": 1.9467052261929352,
      "grad_norm": 10.674678802490234,
      "learning_rate": 4.474007969298004e-05,
      "loss": 1.6419,
      "step": 11780
    },
    {
      "epoch": 1.9483577773187357,
      "grad_norm": 13.497589111328125,
      "learning_rate": 4.4730898474081424e-05,
      "loss": 1.6282,
      "step": 11790
    },
    {
      "epoch": 1.9500103284445363,
      "grad_norm": 9.30100154876709,
      "learning_rate": 4.47217172551828e-05,
      "loss": 1.6183,
      "step": 11800
    },
    {
      "epoch": 1.9516628795703368,
      "grad_norm": 9.254987716674805,
      "learning_rate": 4.471253603628418e-05,
      "loss": 1.6552,
      "step": 11810
    },
    {
      "epoch": 1.9533154306961371,
      "grad_norm": 14.543704986572266,
      "learning_rate": 4.470335481738556e-05,
      "loss": 1.5962,
      "step": 11820
    },
    {
      "epoch": 1.9549679818219377,
      "grad_norm": 17.067127227783203,
      "learning_rate": 4.469417359848694e-05,
      "loss": 1.6292,
      "step": 11830
    },
    {
      "epoch": 1.956620532947738,
      "grad_norm": 49.91341018676758,
      "learning_rate": 4.4684992379588316e-05,
      "loss": 1.6171,
      "step": 11840
    },
    {
      "epoch": 1.9582730840735385,
      "grad_norm": 11.141582489013672,
      "learning_rate": 4.467581116068969e-05,
      "loss": 1.6286,
      "step": 11850
    },
    {
      "epoch": 1.959925635199339,
      "grad_norm": 148.42442321777344,
      "learning_rate": 4.4666629941791075e-05,
      "loss": 1.598,
      "step": 11860
    },
    {
      "epoch": 1.9615781863251396,
      "grad_norm": 7.553905010223389,
      "learning_rate": 4.465744872289245e-05,
      "loss": 1.7413,
      "step": 11870
    },
    {
      "epoch": 1.96323073745094,
      "grad_norm": 8.26811408996582,
      "learning_rate": 4.464826750399383e-05,
      "loss": 1.7399,
      "step": 11880
    },
    {
      "epoch": 1.9648832885767402,
      "grad_norm": 12.610078811645508,
      "learning_rate": 4.463908628509521e-05,
      "loss": 1.6173,
      "step": 11890
    },
    {
      "epoch": 1.9665358397025408,
      "grad_norm": 12.24221134185791,
      "learning_rate": 4.462990506619659e-05,
      "loss": 1.5597,
      "step": 11900
    },
    {
      "epoch": 1.9681883908283413,
      "grad_norm": 32.20631790161133,
      "learning_rate": 4.462072384729797e-05,
      "loss": 1.6357,
      "step": 11910
    },
    {
      "epoch": 1.9698409419541418,
      "grad_norm": 13.021040916442871,
      "learning_rate": 4.461154262839935e-05,
      "loss": 1.5987,
      "step": 11920
    },
    {
      "epoch": 1.9714934930799421,
      "grad_norm": 9.593809127807617,
      "learning_rate": 4.4602361409500726e-05,
      "loss": 1.5853,
      "step": 11930
    },
    {
      "epoch": 1.9731460442057425,
      "grad_norm": 6.821667194366455,
      "learning_rate": 4.459318019060211e-05,
      "loss": 1.5064,
      "step": 11940
    },
    {
      "epoch": 1.974798595331543,
      "grad_norm": 15.331339836120605,
      "learning_rate": 4.4583998971703485e-05,
      "loss": 1.6008,
      "step": 11950
    },
    {
      "epoch": 1.9764511464573435,
      "grad_norm": 8.000182151794434,
      "learning_rate": 4.457481775280487e-05,
      "loss": 1.655,
      "step": 11960
    },
    {
      "epoch": 1.978103697583144,
      "grad_norm": 8.434917449951172,
      "learning_rate": 4.456563653390624e-05,
      "loss": 1.753,
      "step": 11970
    },
    {
      "epoch": 1.9797562487089444,
      "grad_norm": 48.595760345458984,
      "learning_rate": 4.4556455315007626e-05,
      "loss": 1.685,
      "step": 11980
    },
    {
      "epoch": 1.981408799834745,
      "grad_norm": 9.696410179138184,
      "learning_rate": 4.4547274096109e-05,
      "loss": 1.6055,
      "step": 11990
    },
    {
      "epoch": 1.9830613509605453,
      "grad_norm": 9.626373291015625,
      "learning_rate": 4.453809287721038e-05,
      "loss": 1.7424,
      "step": 12000
    },
    {
      "epoch": 1.9847139020863458,
      "grad_norm": 62.46161651611328,
      "learning_rate": 4.452891165831176e-05,
      "loss": 1.596,
      "step": 12010
    },
    {
      "epoch": 1.9863664532121463,
      "grad_norm": 25.460296630859375,
      "learning_rate": 4.4519730439413136e-05,
      "loss": 1.7073,
      "step": 12020
    },
    {
      "epoch": 1.9880190043379469,
      "grad_norm": 6.691403865814209,
      "learning_rate": 4.451054922051452e-05,
      "loss": 1.61,
      "step": 12030
    },
    {
      "epoch": 1.9896715554637472,
      "grad_norm": 9.661092758178711,
      "learning_rate": 4.4501368001615895e-05,
      "loss": 1.6125,
      "step": 12040
    },
    {
      "epoch": 1.9913241065895475,
      "grad_norm": 26.178430557250977,
      "learning_rate": 4.449218678271728e-05,
      "loss": 1.5553,
      "step": 12050
    },
    {
      "epoch": 1.992976657715348,
      "grad_norm": 96.2996826171875,
      "learning_rate": 4.448300556381865e-05,
      "loss": 1.7229,
      "step": 12060
    },
    {
      "epoch": 1.9946292088411486,
      "grad_norm": 18.224470138549805,
      "learning_rate": 4.4473824344920036e-05,
      "loss": 1.6561,
      "step": 12070
    },
    {
      "epoch": 1.9962817599669491,
      "grad_norm": 12.857390403747559,
      "learning_rate": 4.446464312602141e-05,
      "loss": 1.5519,
      "step": 12080
    },
    {
      "epoch": 1.9979343110927494,
      "grad_norm": 65.5601577758789,
      "learning_rate": 4.4455461907122794e-05,
      "loss": 1.5376,
      "step": 12090
    },
    {
      "epoch": 1.9995868622185498,
      "grad_norm": 31.238407135009766,
      "learning_rate": 4.444628068822417e-05,
      "loss": 1.6433,
      "step": 12100
    },
    {
      "epoch": 1.99991737244371,
      "eval_accuracy": 0.2733721095190807,
      "eval_loss": 2.2641355991363525,
      "eval_runtime": 819.3221,
      "eval_samples_per_second": 34.414,
      "eval_steps_per_second": 8.603,
      "step": 12102
    },
    {
      "epoch": 2.0012394133443503,
      "grad_norm": 44.95764923095703,
      "learning_rate": 4.443709946932555e-05,
      "loss": 1.6572,
      "step": 12110
    },
    {
      "epoch": 2.002891964470151,
      "grad_norm": 28.433679580688477,
      "learning_rate": 4.442791825042693e-05,
      "loss": 1.5288,
      "step": 12120
    },
    {
      "epoch": 2.0045445155959514,
      "grad_norm": 20.113248825073242,
      "learning_rate": 4.4418737031528304e-05,
      "loss": 1.5625,
      "step": 12130
    },
    {
      "epoch": 2.006197066721752,
      "grad_norm": 18.262046813964844,
      "learning_rate": 4.440955581262969e-05,
      "loss": 1.6292,
      "step": 12140
    },
    {
      "epoch": 2.007849617847552,
      "grad_norm": 9.9489164352417,
      "learning_rate": 4.440037459373106e-05,
      "loss": 1.7362,
      "step": 12150
    },
    {
      "epoch": 2.0095021689733525,
      "grad_norm": 71.15792846679688,
      "learning_rate": 4.4391193374832446e-05,
      "loss": 1.7299,
      "step": 12160
    },
    {
      "epoch": 2.011154720099153,
      "grad_norm": 12.936304092407227,
      "learning_rate": 4.438201215593382e-05,
      "loss": 1.5459,
      "step": 12170
    },
    {
      "epoch": 2.0128072712249536,
      "grad_norm": 17.437440872192383,
      "learning_rate": 4.4372830937035204e-05,
      "loss": 1.604,
      "step": 12180
    },
    {
      "epoch": 2.014459822350754,
      "grad_norm": 12.949777603149414,
      "learning_rate": 4.436364971813658e-05,
      "loss": 1.5845,
      "step": 12190
    },
    {
      "epoch": 2.0161123734765543,
      "grad_norm": 34.75144958496094,
      "learning_rate": 4.435446849923796e-05,
      "loss": 1.7017,
      "step": 12200
    },
    {
      "epoch": 2.017764924602355,
      "grad_norm": 25.475833892822266,
      "learning_rate": 4.4345287280339345e-05,
      "loss": 1.6947,
      "step": 12210
    },
    {
      "epoch": 2.0194174757281553,
      "grad_norm": 10.156431198120117,
      "learning_rate": 4.433610606144072e-05,
      "loss": 1.6284,
      "step": 12220
    },
    {
      "epoch": 2.021070026853956,
      "grad_norm": 19.941617965698242,
      "learning_rate": 4.43269248425421e-05,
      "loss": 1.6042,
      "step": 12230
    },
    {
      "epoch": 2.0227225779797564,
      "grad_norm": 9.263725280761719,
      "learning_rate": 4.431774362364348e-05,
      "loss": 1.6048,
      "step": 12240
    },
    {
      "epoch": 2.0243751291055565,
      "grad_norm": 7.461373805999756,
      "learning_rate": 4.4308562404744855e-05,
      "loss": 1.6171,
      "step": 12250
    },
    {
      "epoch": 2.026027680231357,
      "grad_norm": 28.221065521240234,
      "learning_rate": 4.429938118584623e-05,
      "loss": 1.6176,
      "step": 12260
    },
    {
      "epoch": 2.0276802313571576,
      "grad_norm": 6.061464309692383,
      "learning_rate": 4.4290199966947614e-05,
      "loss": 1.5904,
      "step": 12270
    },
    {
      "epoch": 2.029332782482958,
      "grad_norm": 7.372241973876953,
      "learning_rate": 4.428101874804899e-05,
      "loss": 1.5756,
      "step": 12280
    },
    {
      "epoch": 2.0309853336087587,
      "grad_norm": 5.46327543258667,
      "learning_rate": 4.427183752915037e-05,
      "loss": 1.6985,
      "step": 12290
    },
    {
      "epoch": 2.0326378847345588,
      "grad_norm": 8.150008201599121,
      "learning_rate": 4.426265631025175e-05,
      "loss": 1.6157,
      "step": 12300
    },
    {
      "epoch": 2.0342904358603593,
      "grad_norm": 7.626075267791748,
      "learning_rate": 4.425347509135313e-05,
      "loss": 1.5679,
      "step": 12310
    },
    {
      "epoch": 2.03594298698616,
      "grad_norm": 14.908061981201172,
      "learning_rate": 4.424429387245451e-05,
      "loss": 1.548,
      "step": 12320
    },
    {
      "epoch": 2.0375955381119604,
      "grad_norm": 7.012806415557861,
      "learning_rate": 4.423511265355589e-05,
      "loss": 1.7013,
      "step": 12330
    },
    {
      "epoch": 2.039248089237761,
      "grad_norm": 5.985696792602539,
      "learning_rate": 4.422593143465727e-05,
      "loss": 1.543,
      "step": 12340
    },
    {
      "epoch": 2.0409006403635614,
      "grad_norm": 20.78647232055664,
      "learning_rate": 4.421675021575865e-05,
      "loss": 1.6405,
      "step": 12350
    },
    {
      "epoch": 2.0425531914893615,
      "grad_norm": 11.975393295288086,
      "learning_rate": 4.4207568996860024e-05,
      "loss": 1.6962,
      "step": 12360
    },
    {
      "epoch": 2.044205742615162,
      "grad_norm": 8.53466796875,
      "learning_rate": 4.4198387777961406e-05,
      "loss": 1.6913,
      "step": 12370
    },
    {
      "epoch": 2.0458582937409626,
      "grad_norm": 9.030325889587402,
      "learning_rate": 4.418920655906278e-05,
      "loss": 1.578,
      "step": 12380
    },
    {
      "epoch": 2.047510844866763,
      "grad_norm": 21.684886932373047,
      "learning_rate": 4.418002534016416e-05,
      "loss": 1.5179,
      "step": 12390
    },
    {
      "epoch": 2.0491633959925637,
      "grad_norm": 13.426318168640137,
      "learning_rate": 4.417084412126554e-05,
      "loss": 1.637,
      "step": 12400
    },
    {
      "epoch": 2.050815947118364,
      "grad_norm": 6.491002082824707,
      "learning_rate": 4.4161662902366917e-05,
      "loss": 1.5211,
      "step": 12410
    },
    {
      "epoch": 2.0524684982441643,
      "grad_norm": 8.91051197052002,
      "learning_rate": 4.41524816834683e-05,
      "loss": 1.6806,
      "step": 12420
    },
    {
      "epoch": 2.054121049369965,
      "grad_norm": 26.088380813598633,
      "learning_rate": 4.4143300464569675e-05,
      "loss": 1.6835,
      "step": 12430
    },
    {
      "epoch": 2.0557736004957654,
      "grad_norm": 9.795445442199707,
      "learning_rate": 4.413411924567106e-05,
      "loss": 1.7466,
      "step": 12440
    },
    {
      "epoch": 2.057426151621566,
      "grad_norm": 16.947595596313477,
      "learning_rate": 4.412493802677244e-05,
      "loss": 1.6179,
      "step": 12450
    },
    {
      "epoch": 2.059078702747366,
      "grad_norm": 14.341743469238281,
      "learning_rate": 4.4115756807873816e-05,
      "loss": 1.6692,
      "step": 12460
    },
    {
      "epoch": 2.0607312538731666,
      "grad_norm": 8.210740089416504,
      "learning_rate": 4.41065755889752e-05,
      "loss": 1.5425,
      "step": 12470
    },
    {
      "epoch": 2.062383804998967,
      "grad_norm": 16.021188735961914,
      "learning_rate": 4.4097394370076575e-05,
      "loss": 1.466,
      "step": 12480
    },
    {
      "epoch": 2.0640363561247677,
      "grad_norm": 25.511781692504883,
      "learning_rate": 4.408821315117795e-05,
      "loss": 1.6227,
      "step": 12490
    },
    {
      "epoch": 2.065688907250568,
      "grad_norm": 10.479546546936035,
      "learning_rate": 4.407903193227933e-05,
      "loss": 1.6323,
      "step": 12500
    },
    {
      "epoch": 2.0673414583763687,
      "grad_norm": 29.294937133789062,
      "learning_rate": 4.406985071338071e-05,
      "loss": 1.7135,
      "step": 12510
    },
    {
      "epoch": 2.068994009502169,
      "grad_norm": 9.334646224975586,
      "learning_rate": 4.4060669494482085e-05,
      "loss": 1.4985,
      "step": 12520
    },
    {
      "epoch": 2.0706465606279694,
      "grad_norm": 6.601807117462158,
      "learning_rate": 4.405148827558347e-05,
      "loss": 1.6179,
      "step": 12530
    },
    {
      "epoch": 2.07229911175377,
      "grad_norm": 39.724403381347656,
      "learning_rate": 4.4042307056684843e-05,
      "loss": 1.5593,
      "step": 12540
    },
    {
      "epoch": 2.0739516628795704,
      "grad_norm": 21.564411163330078,
      "learning_rate": 4.4033125837786226e-05,
      "loss": 1.6602,
      "step": 12550
    },
    {
      "epoch": 2.075604214005371,
      "grad_norm": 9.211846351623535,
      "learning_rate": 4.402394461888761e-05,
      "loss": 1.5664,
      "step": 12560
    },
    {
      "epoch": 2.077256765131171,
      "grad_norm": 86.83989715576172,
      "learning_rate": 4.4014763399988985e-05,
      "loss": 1.6355,
      "step": 12570
    },
    {
      "epoch": 2.0789093162569716,
      "grad_norm": 15.011252403259277,
      "learning_rate": 4.400558218109037e-05,
      "loss": 1.7233,
      "step": 12580
    },
    {
      "epoch": 2.080561867382772,
      "grad_norm": 22.9763240814209,
      "learning_rate": 4.399640096219174e-05,
      "loss": 1.6327,
      "step": 12590
    },
    {
      "epoch": 2.0822144185085727,
      "grad_norm": 9.363661766052246,
      "learning_rate": 4.3987219743293126e-05,
      "loss": 1.6782,
      "step": 12600
    },
    {
      "epoch": 2.0838669696343732,
      "grad_norm": 10.958879470825195,
      "learning_rate": 4.39780385243945e-05,
      "loss": 1.7072,
      "step": 12610
    },
    {
      "epoch": 2.0855195207601733,
      "grad_norm": 6.636941432952881,
      "learning_rate": 4.3968857305495884e-05,
      "loss": 1.5247,
      "step": 12620
    },
    {
      "epoch": 2.087172071885974,
      "grad_norm": 74.40653228759766,
      "learning_rate": 4.395967608659726e-05,
      "loss": 1.5546,
      "step": 12630
    },
    {
      "epoch": 2.0888246230117744,
      "grad_norm": 9.923933029174805,
      "learning_rate": 4.3950494867698636e-05,
      "loss": 1.5428,
      "step": 12640
    },
    {
      "epoch": 2.090477174137575,
      "grad_norm": 8.562093734741211,
      "learning_rate": 4.394131364880001e-05,
      "loss": 1.5351,
      "step": 12650
    },
    {
      "epoch": 2.0921297252633755,
      "grad_norm": 46.868709564208984,
      "learning_rate": 4.3932132429901394e-05,
      "loss": 1.6106,
      "step": 12660
    },
    {
      "epoch": 2.0937822763891756,
      "grad_norm": 5.111778259277344,
      "learning_rate": 4.392295121100278e-05,
      "loss": 1.5635,
      "step": 12670
    },
    {
      "epoch": 2.095434827514976,
      "grad_norm": 9.51055908203125,
      "learning_rate": 4.391376999210415e-05,
      "loss": 1.5957,
      "step": 12680
    },
    {
      "epoch": 2.0970873786407767,
      "grad_norm": 14.892036437988281,
      "learning_rate": 4.3904588773205535e-05,
      "loss": 1.6939,
      "step": 12690
    },
    {
      "epoch": 2.098739929766577,
      "grad_norm": 6.7437567710876465,
      "learning_rate": 4.389540755430691e-05,
      "loss": 1.5671,
      "step": 12700
    },
    {
      "epoch": 2.1003924808923777,
      "grad_norm": 34.27804946899414,
      "learning_rate": 4.3886226335408294e-05,
      "loss": 1.6071,
      "step": 12710
    },
    {
      "epoch": 2.1020450320181783,
      "grad_norm": 13.502313613891602,
      "learning_rate": 4.387704511650967e-05,
      "loss": 1.5568,
      "step": 12720
    },
    {
      "epoch": 2.1036975831439784,
      "grad_norm": 12.005770683288574,
      "learning_rate": 4.386786389761105e-05,
      "loss": 1.563,
      "step": 12730
    },
    {
      "epoch": 2.105350134269779,
      "grad_norm": 8.112605094909668,
      "learning_rate": 4.385868267871243e-05,
      "loss": 1.7326,
      "step": 12740
    },
    {
      "epoch": 2.1070026853955794,
      "grad_norm": 6.765664100646973,
      "learning_rate": 4.384950145981381e-05,
      "loss": 1.6562,
      "step": 12750
    },
    {
      "epoch": 2.10865523652138,
      "grad_norm": 12.570945739746094,
      "learning_rate": 4.384032024091519e-05,
      "loss": 1.6129,
      "step": 12760
    },
    {
      "epoch": 2.1103077876471805,
      "grad_norm": 10.410295486450195,
      "learning_rate": 4.383113902201656e-05,
      "loss": 1.6615,
      "step": 12770
    },
    {
      "epoch": 2.1119603387729806,
      "grad_norm": 25.10518455505371,
      "learning_rate": 4.3821957803117945e-05,
      "loss": 1.7272,
      "step": 12780
    },
    {
      "epoch": 2.113612889898781,
      "grad_norm": 19.222471237182617,
      "learning_rate": 4.381277658421932e-05,
      "loss": 1.5842,
      "step": 12790
    },
    {
      "epoch": 2.1152654410245817,
      "grad_norm": 9.53968334197998,
      "learning_rate": 4.3803595365320704e-05,
      "loss": 1.5341,
      "step": 12800
    },
    {
      "epoch": 2.1169179921503822,
      "grad_norm": 11.091026306152344,
      "learning_rate": 4.379441414642208e-05,
      "loss": 1.7523,
      "step": 12810
    },
    {
      "epoch": 2.1185705432761828,
      "grad_norm": 5.745179653167725,
      "learning_rate": 4.378523292752346e-05,
      "loss": 1.6201,
      "step": 12820
    },
    {
      "epoch": 2.120223094401983,
      "grad_norm": 21.147724151611328,
      "learning_rate": 4.377605170862484e-05,
      "loss": 1.5084,
      "step": 12830
    },
    {
      "epoch": 2.1218756455277834,
      "grad_norm": 5.363860130310059,
      "learning_rate": 4.376687048972622e-05,
      "loss": 1.6167,
      "step": 12840
    },
    {
      "epoch": 2.123528196653584,
      "grad_norm": 5.5015459060668945,
      "learning_rate": 4.37576892708276e-05,
      "loss": 1.5093,
      "step": 12850
    },
    {
      "epoch": 2.1251807477793845,
      "grad_norm": 6.91679573059082,
      "learning_rate": 4.374850805192898e-05,
      "loss": 1.5668,
      "step": 12860
    },
    {
      "epoch": 2.126833298905185,
      "grad_norm": 15.651091575622559,
      "learning_rate": 4.3739326833030355e-05,
      "loss": 1.6675,
      "step": 12870
    },
    {
      "epoch": 2.1284858500309856,
      "grad_norm": 9.818643569946289,
      "learning_rate": 4.373014561413174e-05,
      "loss": 1.5916,
      "step": 12880
    },
    {
      "epoch": 2.1301384011567857,
      "grad_norm": 11.20659065246582,
      "learning_rate": 4.3720964395233114e-05,
      "loss": 1.6256,
      "step": 12890
    },
    {
      "epoch": 2.131790952282586,
      "grad_norm": 9.242450714111328,
      "learning_rate": 4.371178317633449e-05,
      "loss": 1.5514,
      "step": 12900
    },
    {
      "epoch": 2.1334435034083867,
      "grad_norm": 11.752909660339355,
      "learning_rate": 4.370260195743587e-05,
      "loss": 1.6745,
      "step": 12910
    },
    {
      "epoch": 2.1350960545341873,
      "grad_norm": 19.809019088745117,
      "learning_rate": 4.369342073853725e-05,
      "loss": 1.537,
      "step": 12920
    },
    {
      "epoch": 2.136748605659988,
      "grad_norm": 13.41193962097168,
      "learning_rate": 4.368423951963863e-05,
      "loss": 1.5797,
      "step": 12930
    },
    {
      "epoch": 2.138401156785788,
      "grad_norm": 17.19065284729004,
      "learning_rate": 4.3675058300740007e-05,
      "loss": 1.6646,
      "step": 12940
    },
    {
      "epoch": 2.1400537079115884,
      "grad_norm": 32.37812423706055,
      "learning_rate": 4.366587708184139e-05,
      "loss": 1.6777,
      "step": 12950
    },
    {
      "epoch": 2.141706259037389,
      "grad_norm": 14.590213775634766,
      "learning_rate": 4.3656695862942765e-05,
      "loss": 1.582,
      "step": 12960
    },
    {
      "epoch": 2.1433588101631895,
      "grad_norm": 11.580479621887207,
      "learning_rate": 4.364751464404415e-05,
      "loss": 1.5257,
      "step": 12970
    },
    {
      "epoch": 2.14501136128899,
      "grad_norm": 13.083703994750977,
      "learning_rate": 4.3638333425145524e-05,
      "loss": 1.6449,
      "step": 12980
    },
    {
      "epoch": 2.14666391241479,
      "grad_norm": 44.948280334472656,
      "learning_rate": 4.3629152206246906e-05,
      "loss": 1.6575,
      "step": 12990
    },
    {
      "epoch": 2.1483164635405907,
      "grad_norm": 12.724278450012207,
      "learning_rate": 4.361997098734828e-05,
      "loss": 1.576,
      "step": 13000
    },
    {
      "epoch": 2.1499690146663912,
      "grad_norm": 14.588871002197266,
      "learning_rate": 4.3610789768449665e-05,
      "loss": 1.5038,
      "step": 13010
    },
    {
      "epoch": 2.1516215657921918,
      "grad_norm": 47.780120849609375,
      "learning_rate": 4.360160854955104e-05,
      "loss": 1.6169,
      "step": 13020
    },
    {
      "epoch": 2.1532741169179923,
      "grad_norm": 303.17950439453125,
      "learning_rate": 4.3592427330652416e-05,
      "loss": 1.4563,
      "step": 13030
    },
    {
      "epoch": 2.1549266680437924,
      "grad_norm": 8.716339111328125,
      "learning_rate": 4.35832461117538e-05,
      "loss": 1.5758,
      "step": 13040
    },
    {
      "epoch": 2.156579219169593,
      "grad_norm": 11.586296081542969,
      "learning_rate": 4.3574064892855175e-05,
      "loss": 1.6525,
      "step": 13050
    },
    {
      "epoch": 2.1582317702953935,
      "grad_norm": 20.37527847290039,
      "learning_rate": 4.356488367395656e-05,
      "loss": 1.6296,
      "step": 13060
    },
    {
      "epoch": 2.159884321421194,
      "grad_norm": 258.34210205078125,
      "learning_rate": 4.355570245505793e-05,
      "loss": 1.6202,
      "step": 13070
    },
    {
      "epoch": 2.1615368725469946,
      "grad_norm": 8.469282150268555,
      "learning_rate": 4.3546521236159316e-05,
      "loss": 1.6968,
      "step": 13080
    },
    {
      "epoch": 2.1631894236727947,
      "grad_norm": 16.505489349365234,
      "learning_rate": 4.353734001726069e-05,
      "loss": 1.6207,
      "step": 13090
    },
    {
      "epoch": 2.164841974798595,
      "grad_norm": 12.483043670654297,
      "learning_rate": 4.3528158798362074e-05,
      "loss": 1.674,
      "step": 13100
    },
    {
      "epoch": 2.1664945259243957,
      "grad_norm": 11.628996849060059,
      "learning_rate": 4.351897757946345e-05,
      "loss": 1.6864,
      "step": 13110
    },
    {
      "epoch": 2.1681470770501963,
      "grad_norm": 9.3779878616333,
      "learning_rate": 4.350979636056483e-05,
      "loss": 1.5968,
      "step": 13120
    },
    {
      "epoch": 2.169799628175997,
      "grad_norm": 7.350371360778809,
      "learning_rate": 4.350061514166621e-05,
      "loss": 1.5171,
      "step": 13130
    },
    {
      "epoch": 2.1714521793017973,
      "grad_norm": 17.027692794799805,
      "learning_rate": 4.349143392276759e-05,
      "loss": 1.7044,
      "step": 13140
    },
    {
      "epoch": 2.1731047304275974,
      "grad_norm": 6.943367958068848,
      "learning_rate": 4.348225270386897e-05,
      "loss": 1.585,
      "step": 13150
    },
    {
      "epoch": 2.174757281553398,
      "grad_norm": 14.061978340148926,
      "learning_rate": 4.347307148497034e-05,
      "loss": 1.5373,
      "step": 13160
    },
    {
      "epoch": 2.1764098326791985,
      "grad_norm": 10.38961410522461,
      "learning_rate": 4.3463890266071726e-05,
      "loss": 1.7171,
      "step": 13170
    },
    {
      "epoch": 2.178062383804999,
      "grad_norm": 33.14574432373047,
      "learning_rate": 4.34547090471731e-05,
      "loss": 1.5202,
      "step": 13180
    },
    {
      "epoch": 2.1797149349307996,
      "grad_norm": 34.48433303833008,
      "learning_rate": 4.3445527828274484e-05,
      "loss": 1.671,
      "step": 13190
    },
    {
      "epoch": 2.1813674860565997,
      "grad_norm": 8.59406566619873,
      "learning_rate": 4.343634660937586e-05,
      "loss": 1.6391,
      "step": 13200
    },
    {
      "epoch": 2.1830200371824002,
      "grad_norm": 9.573761940002441,
      "learning_rate": 4.342716539047724e-05,
      "loss": 1.6025,
      "step": 13210
    },
    {
      "epoch": 2.1846725883082008,
      "grad_norm": 11.121373176574707,
      "learning_rate": 4.341798417157862e-05,
      "loss": 1.5423,
      "step": 13220
    },
    {
      "epoch": 2.1863251394340013,
      "grad_norm": 8.483185768127441,
      "learning_rate": 4.340880295268e-05,
      "loss": 1.6423,
      "step": 13230
    },
    {
      "epoch": 2.187977690559802,
      "grad_norm": 11.853544235229492,
      "learning_rate": 4.3399621733781384e-05,
      "loss": 1.6377,
      "step": 13240
    },
    {
      "epoch": 2.189630241685602,
      "grad_norm": 6.522881031036377,
      "learning_rate": 4.339044051488276e-05,
      "loss": 1.5402,
      "step": 13250
    },
    {
      "epoch": 2.1912827928114025,
      "grad_norm": 14.069580078125,
      "learning_rate": 4.338125929598414e-05,
      "loss": 1.5593,
      "step": 13260
    },
    {
      "epoch": 2.192935343937203,
      "grad_norm": 7.207234859466553,
      "learning_rate": 4.337207807708552e-05,
      "loss": 1.6057,
      "step": 13270
    },
    {
      "epoch": 2.1945878950630036,
      "grad_norm": 11.689896583557129,
      "learning_rate": 4.3362896858186894e-05,
      "loss": 1.5226,
      "step": 13280
    },
    {
      "epoch": 2.196240446188804,
      "grad_norm": 57.798797607421875,
      "learning_rate": 4.335371563928827e-05,
      "loss": 1.5292,
      "step": 13290
    },
    {
      "epoch": 2.1978929973146046,
      "grad_norm": 15.47495174407959,
      "learning_rate": 4.334453442038965e-05,
      "loss": 1.5943,
      "step": 13300
    },
    {
      "epoch": 2.1995455484404047,
      "grad_norm": 6.988327980041504,
      "learning_rate": 4.333535320149103e-05,
      "loss": 1.5755,
      "step": 13310
    },
    {
      "epoch": 2.2011980995662053,
      "grad_norm": 10.670233726501465,
      "learning_rate": 4.332617198259241e-05,
      "loss": 1.6494,
      "step": 13320
    },
    {
      "epoch": 2.202850650692006,
      "grad_norm": 9.843840599060059,
      "learning_rate": 4.331699076369379e-05,
      "loss": 1.6159,
      "step": 13330
    },
    {
      "epoch": 2.2045032018178063,
      "grad_norm": 7.6408233642578125,
      "learning_rate": 4.330780954479517e-05,
      "loss": 1.6032,
      "step": 13340
    },
    {
      "epoch": 2.206155752943607,
      "grad_norm": 7.837189197540283,
      "learning_rate": 4.329862832589655e-05,
      "loss": 1.5953,
      "step": 13350
    },
    {
      "epoch": 2.207808304069407,
      "grad_norm": 14.017073631286621,
      "learning_rate": 4.328944710699793e-05,
      "loss": 1.5831,
      "step": 13360
    },
    {
      "epoch": 2.2094608551952075,
      "grad_norm": 19.07843017578125,
      "learning_rate": 4.328026588809931e-05,
      "loss": 1.5628,
      "step": 13370
    },
    {
      "epoch": 2.211113406321008,
      "grad_norm": 8.755454063415527,
      "learning_rate": 4.327108466920069e-05,
      "loss": 1.6262,
      "step": 13380
    },
    {
      "epoch": 2.2127659574468086,
      "grad_norm": 41.75136184692383,
      "learning_rate": 4.326190345030207e-05,
      "loss": 1.6595,
      "step": 13390
    },
    {
      "epoch": 2.214418508572609,
      "grad_norm": 10.803937911987305,
      "learning_rate": 4.3252722231403445e-05,
      "loss": 1.7103,
      "step": 13400
    },
    {
      "epoch": 2.2160710596984092,
      "grad_norm": 64.1683349609375,
      "learning_rate": 4.324354101250482e-05,
      "loss": 1.6162,
      "step": 13410
    },
    {
      "epoch": 2.2177236108242098,
      "grad_norm": 14.838408470153809,
      "learning_rate": 4.32343597936062e-05,
      "loss": 1.6336,
      "step": 13420
    },
    {
      "epoch": 2.2193761619500103,
      "grad_norm": 15.170886039733887,
      "learning_rate": 4.322517857470758e-05,
      "loss": 1.4815,
      "step": 13430
    },
    {
      "epoch": 2.221028713075811,
      "grad_norm": 13.091402053833008,
      "learning_rate": 4.3215997355808955e-05,
      "loss": 1.6045,
      "step": 13440
    },
    {
      "epoch": 2.2226812642016114,
      "grad_norm": 8.757658958435059,
      "learning_rate": 4.320681613691034e-05,
      "loss": 1.7558,
      "step": 13450
    },
    {
      "epoch": 2.2243338153274115,
      "grad_norm": 12.71592903137207,
      "learning_rate": 4.3197634918011714e-05,
      "loss": 1.581,
      "step": 13460
    },
    {
      "epoch": 2.225986366453212,
      "grad_norm": 18.19761085510254,
      "learning_rate": 4.3188453699113096e-05,
      "loss": 1.5034,
      "step": 13470
    },
    {
      "epoch": 2.2276389175790126,
      "grad_norm": 8.124272346496582,
      "learning_rate": 4.317927248021448e-05,
      "loss": 1.561,
      "step": 13480
    },
    {
      "epoch": 2.229291468704813,
      "grad_norm": 6.5509796142578125,
      "learning_rate": 4.3170091261315855e-05,
      "loss": 1.709,
      "step": 13490
    },
    {
      "epoch": 2.2309440198306136,
      "grad_norm": 34.16242599487305,
      "learning_rate": 4.316091004241724e-05,
      "loss": 1.5383,
      "step": 13500
    },
    {
      "epoch": 2.2325965709564137,
      "grad_norm": 60.66366195678711,
      "learning_rate": 4.3151728823518613e-05,
      "loss": 1.6962,
      "step": 13510
    },
    {
      "epoch": 2.2342491220822143,
      "grad_norm": 8.734251976013184,
      "learning_rate": 4.3142547604619996e-05,
      "loss": 1.5747,
      "step": 13520
    },
    {
      "epoch": 2.235901673208015,
      "grad_norm": 7.926821231842041,
      "learning_rate": 4.313336638572137e-05,
      "loss": 1.5384,
      "step": 13530
    },
    {
      "epoch": 2.2375542243338153,
      "grad_norm": 13.231529235839844,
      "learning_rate": 4.312418516682275e-05,
      "loss": 1.6822,
      "step": 13540
    },
    {
      "epoch": 2.239206775459616,
      "grad_norm": 19.836008071899414,
      "learning_rate": 4.3115003947924124e-05,
      "loss": 1.5861,
      "step": 13550
    },
    {
      "epoch": 2.2408593265854164,
      "grad_norm": 9.528962135314941,
      "learning_rate": 4.3105822729025506e-05,
      "loss": 1.7453,
      "step": 13560
    },
    {
      "epoch": 2.2425118777112165,
      "grad_norm": 25.385387420654297,
      "learning_rate": 4.309664151012688e-05,
      "loss": 1.5686,
      "step": 13570
    },
    {
      "epoch": 2.244164428837017,
      "grad_norm": 7.8074870109558105,
      "learning_rate": 4.3087460291228265e-05,
      "loss": 1.6753,
      "step": 13580
    },
    {
      "epoch": 2.2458169799628176,
      "grad_norm": 142.185302734375,
      "learning_rate": 4.307827907232965e-05,
      "loss": 1.5779,
      "step": 13590
    },
    {
      "epoch": 2.247469531088618,
      "grad_norm": 24.071073532104492,
      "learning_rate": 4.306909785343102e-05,
      "loss": 1.5333,
      "step": 13600
    },
    {
      "epoch": 2.2491220822144187,
      "grad_norm": 16.71168327331543,
      "learning_rate": 4.3059916634532406e-05,
      "loss": 1.6225,
      "step": 13610
    },
    {
      "epoch": 2.250774633340219,
      "grad_norm": 14.29904556274414,
      "learning_rate": 4.305073541563378e-05,
      "loss": 1.7723,
      "step": 13620
    },
    {
      "epoch": 2.2524271844660193,
      "grad_norm": 9.134810447692871,
      "learning_rate": 4.3041554196735164e-05,
      "loss": 1.5871,
      "step": 13630
    },
    {
      "epoch": 2.25407973559182,
      "grad_norm": 15.202738761901855,
      "learning_rate": 4.303237297783654e-05,
      "loss": 1.6463,
      "step": 13640
    },
    {
      "epoch": 2.2557322867176204,
      "grad_norm": 61.93202209472656,
      "learning_rate": 4.302319175893792e-05,
      "loss": 1.7039,
      "step": 13650
    },
    {
      "epoch": 2.257384837843421,
      "grad_norm": 48.4030647277832,
      "learning_rate": 4.30140105400393e-05,
      "loss": 1.6503,
      "step": 13660
    },
    {
      "epoch": 2.2590373889692215,
      "grad_norm": 17.932994842529297,
      "learning_rate": 4.3004829321140675e-05,
      "loss": 1.5064,
      "step": 13670
    },
    {
      "epoch": 2.2606899400950216,
      "grad_norm": 12.965964317321777,
      "learning_rate": 4.299564810224205e-05,
      "loss": 1.5113,
      "step": 13680
    },
    {
      "epoch": 2.262342491220822,
      "grad_norm": 34.27570724487305,
      "learning_rate": 4.298646688334343e-05,
      "loss": 1.5163,
      "step": 13690
    },
    {
      "epoch": 2.2639950423466226,
      "grad_norm": 31.845481872558594,
      "learning_rate": 4.2977285664444816e-05,
      "loss": 1.5038,
      "step": 13700
    },
    {
      "epoch": 2.265647593472423,
      "grad_norm": 10.161199569702148,
      "learning_rate": 4.296810444554619e-05,
      "loss": 1.4943,
      "step": 13710
    },
    {
      "epoch": 2.2673001445982237,
      "grad_norm": 14.812753677368164,
      "learning_rate": 4.2958923226647574e-05,
      "loss": 1.5093,
      "step": 13720
    },
    {
      "epoch": 2.268952695724024,
      "grad_norm": 8.536107063293457,
      "learning_rate": 4.294974200774895e-05,
      "loss": 1.5917,
      "step": 13730
    },
    {
      "epoch": 2.2706052468498243,
      "grad_norm": 12.14443302154541,
      "learning_rate": 4.294056078885033e-05,
      "loss": 1.6145,
      "step": 13740
    },
    {
      "epoch": 2.272257797975625,
      "grad_norm": 9.584919929504395,
      "learning_rate": 4.293137956995171e-05,
      "loss": 1.5674,
      "step": 13750
    },
    {
      "epoch": 2.2739103491014254,
      "grad_norm": 8.43072509765625,
      "learning_rate": 4.292219835105309e-05,
      "loss": 1.6024,
      "step": 13760
    },
    {
      "epoch": 2.275562900227226,
      "grad_norm": 10.97778606414795,
      "learning_rate": 4.291301713215447e-05,
      "loss": 1.7366,
      "step": 13770
    },
    {
      "epoch": 2.277215451353026,
      "grad_norm": 7.326660633087158,
      "learning_rate": 4.290383591325585e-05,
      "loss": 1.632,
      "step": 13780
    },
    {
      "epoch": 2.2788680024788266,
      "grad_norm": 6.518946170806885,
      "learning_rate": 4.2894654694357226e-05,
      "loss": 1.5364,
      "step": 13790
    },
    {
      "epoch": 2.280520553604627,
      "grad_norm": 28.530855178833008,
      "learning_rate": 4.28854734754586e-05,
      "loss": 1.7219,
      "step": 13800
    },
    {
      "epoch": 2.2821731047304277,
      "grad_norm": 14.148917198181152,
      "learning_rate": 4.2876292256559984e-05,
      "loss": 1.6488,
      "step": 13810
    },
    {
      "epoch": 2.283825655856228,
      "grad_norm": 40.1848030090332,
      "learning_rate": 4.286711103766136e-05,
      "loss": 1.6645,
      "step": 13820
    },
    {
      "epoch": 2.2854782069820283,
      "grad_norm": 7.235057830810547,
      "learning_rate": 4.285792981876274e-05,
      "loss": 1.584,
      "step": 13830
    },
    {
      "epoch": 2.287130758107829,
      "grad_norm": 10.927480697631836,
      "learning_rate": 4.284874859986412e-05,
      "loss": 1.612,
      "step": 13840
    },
    {
      "epoch": 2.2887833092336294,
      "grad_norm": 9.236492156982422,
      "learning_rate": 4.28395673809655e-05,
      "loss": 1.6889,
      "step": 13850
    },
    {
      "epoch": 2.29043586035943,
      "grad_norm": 13.05184555053711,
      "learning_rate": 4.283038616206688e-05,
      "loss": 1.564,
      "step": 13860
    },
    {
      "epoch": 2.2920884114852305,
      "grad_norm": 17.070354461669922,
      "learning_rate": 4.282120494316826e-05,
      "loss": 1.5551,
      "step": 13870
    },
    {
      "epoch": 2.2937409626110306,
      "grad_norm": 11.861235618591309,
      "learning_rate": 4.2812023724269635e-05,
      "loss": 1.763,
      "step": 13880
    },
    {
      "epoch": 2.295393513736831,
      "grad_norm": 10.376340866088867,
      "learning_rate": 4.280284250537102e-05,
      "loss": 1.6106,
      "step": 13890
    },
    {
      "epoch": 2.2970460648626316,
      "grad_norm": 77.84674835205078,
      "learning_rate": 4.2793661286472394e-05,
      "loss": 1.6489,
      "step": 13900
    },
    {
      "epoch": 2.298698615988432,
      "grad_norm": 8.317307472229004,
      "learning_rate": 4.2784480067573777e-05,
      "loss": 1.6459,
      "step": 13910
    },
    {
      "epoch": 2.3003511671142327,
      "grad_norm": 9.242462158203125,
      "learning_rate": 4.277529884867515e-05,
      "loss": 1.5626,
      "step": 13920
    },
    {
      "epoch": 2.302003718240033,
      "grad_norm": 9.533720016479492,
      "learning_rate": 4.276611762977653e-05,
      "loss": 1.6698,
      "step": 13930
    },
    {
      "epoch": 2.3036562693658333,
      "grad_norm": 8.523353576660156,
      "learning_rate": 4.275693641087791e-05,
      "loss": 1.6613,
      "step": 13940
    },
    {
      "epoch": 2.305308820491634,
      "grad_norm": 7.527460098266602,
      "learning_rate": 4.274775519197929e-05,
      "loss": 1.5093,
      "step": 13950
    },
    {
      "epoch": 2.3069613716174344,
      "grad_norm": 7.213209629058838,
      "learning_rate": 4.273857397308067e-05,
      "loss": 1.6411,
      "step": 13960
    },
    {
      "epoch": 2.308613922743235,
      "grad_norm": 8.763940811157227,
      "learning_rate": 4.2729392754182045e-05,
      "loss": 1.666,
      "step": 13970
    },
    {
      "epoch": 2.3102664738690355,
      "grad_norm": 26.79812240600586,
      "learning_rate": 4.272021153528343e-05,
      "loss": 1.5755,
      "step": 13980
    },
    {
      "epoch": 2.3119190249948356,
      "grad_norm": 20.0844669342041,
      "learning_rate": 4.2711030316384804e-05,
      "loss": 1.6604,
      "step": 13990
    },
    {
      "epoch": 2.313571576120636,
      "grad_norm": 9.126409530639648,
      "learning_rate": 4.2701849097486186e-05,
      "loss": 1.5742,
      "step": 14000
    },
    {
      "epoch": 2.3152241272464367,
      "grad_norm": 24.678070068359375,
      "learning_rate": 4.269266787858756e-05,
      "loss": 1.5018,
      "step": 14010
    },
    {
      "epoch": 2.316876678372237,
      "grad_norm": 6.996845245361328,
      "learning_rate": 4.2683486659688945e-05,
      "loss": 1.5528,
      "step": 14020
    },
    {
      "epoch": 2.3185292294980377,
      "grad_norm": 7.3137617111206055,
      "learning_rate": 4.267430544079032e-05,
      "loss": 1.54,
      "step": 14030
    },
    {
      "epoch": 2.3201817806238383,
      "grad_norm": 10.51693344116211,
      "learning_rate": 4.2665124221891703e-05,
      "loss": 1.7331,
      "step": 14040
    },
    {
      "epoch": 2.3218343317496384,
      "grad_norm": 9.240628242492676,
      "learning_rate": 4.265594300299308e-05,
      "loss": 1.5992,
      "step": 14050
    },
    {
      "epoch": 2.323486882875439,
      "grad_norm": 34.22092056274414,
      "learning_rate": 4.2646761784094455e-05,
      "loss": 1.5359,
      "step": 14060
    },
    {
      "epoch": 2.3251394340012395,
      "grad_norm": 8.278003692626953,
      "learning_rate": 4.263758056519584e-05,
      "loss": 1.587,
      "step": 14070
    },
    {
      "epoch": 2.32679198512704,
      "grad_norm": 9.079841613769531,
      "learning_rate": 4.2628399346297214e-05,
      "loss": 1.615,
      "step": 14080
    },
    {
      "epoch": 2.3284445362528405,
      "grad_norm": 130.9219512939453,
      "learning_rate": 4.2619218127398596e-05,
      "loss": 1.6647,
      "step": 14090
    },
    {
      "epoch": 2.3300970873786406,
      "grad_norm": 8.039260864257812,
      "learning_rate": 4.261003690849997e-05,
      "loss": 1.5117,
      "step": 14100
    },
    {
      "epoch": 2.331749638504441,
      "grad_norm": 20.554414749145508,
      "learning_rate": 4.2600855689601355e-05,
      "loss": 1.659,
      "step": 14110
    },
    {
      "epoch": 2.3334021896302417,
      "grad_norm": 16.2318172454834,
      "learning_rate": 4.259167447070273e-05,
      "loss": 1.7476,
      "step": 14120
    },
    {
      "epoch": 2.3350547407560422,
      "grad_norm": 14.719922065734863,
      "learning_rate": 4.258249325180411e-05,
      "loss": 1.6945,
      "step": 14130
    },
    {
      "epoch": 2.336707291881843,
      "grad_norm": 19.263084411621094,
      "learning_rate": 4.257331203290549e-05,
      "loss": 1.6284,
      "step": 14140
    },
    {
      "epoch": 2.338359843007643,
      "grad_norm": 8.643294334411621,
      "learning_rate": 4.256413081400687e-05,
      "loss": 1.568,
      "step": 14150
    },
    {
      "epoch": 2.3400123941334434,
      "grad_norm": 7.2459797859191895,
      "learning_rate": 4.2554949595108254e-05,
      "loss": 1.6159,
      "step": 14160
    },
    {
      "epoch": 2.341664945259244,
      "grad_norm": 9.25794792175293,
      "learning_rate": 4.254576837620963e-05,
      "loss": 1.548,
      "step": 14170
    },
    {
      "epoch": 2.3433174963850445,
      "grad_norm": 39.79923629760742,
      "learning_rate": 4.2536587157311006e-05,
      "loss": 1.5182,
      "step": 14180
    },
    {
      "epoch": 2.344970047510845,
      "grad_norm": 10.711408615112305,
      "learning_rate": 4.252740593841238e-05,
      "loss": 1.5869,
      "step": 14190
    },
    {
      "epoch": 2.346622598636645,
      "grad_norm": 4.972466945648193,
      "learning_rate": 4.2518224719513765e-05,
      "loss": 1.6212,
      "step": 14200
    },
    {
      "epoch": 2.3482751497624457,
      "grad_norm": 23.379873275756836,
      "learning_rate": 4.250904350061514e-05,
      "loss": 1.581,
      "step": 14210
    },
    {
      "epoch": 2.349927700888246,
      "grad_norm": 9.372620582580566,
      "learning_rate": 4.249986228171652e-05,
      "loss": 1.6083,
      "step": 14220
    },
    {
      "epoch": 2.3515802520140467,
      "grad_norm": 7.492948532104492,
      "learning_rate": 4.24906810628179e-05,
      "loss": 1.5183,
      "step": 14230
    },
    {
      "epoch": 2.3532328031398473,
      "grad_norm": 11.669918060302734,
      "learning_rate": 4.248149984391928e-05,
      "loss": 1.6055,
      "step": 14240
    },
    {
      "epoch": 2.3548853542656474,
      "grad_norm": 14.879387855529785,
      "learning_rate": 4.247231862502066e-05,
      "loss": 1.5403,
      "step": 14250
    },
    {
      "epoch": 2.356537905391448,
      "grad_norm": 10.409502029418945,
      "learning_rate": 4.246313740612204e-05,
      "loss": 1.6121,
      "step": 14260
    },
    {
      "epoch": 2.3581904565172485,
      "grad_norm": 12.416383743286133,
      "learning_rate": 4.245395618722342e-05,
      "loss": 1.7431,
      "step": 14270
    },
    {
      "epoch": 2.359843007643049,
      "grad_norm": 13.863862991333008,
      "learning_rate": 4.24447749683248e-05,
      "loss": 1.5425,
      "step": 14280
    },
    {
      "epoch": 2.3614955587688495,
      "grad_norm": 10.172608375549316,
      "learning_rate": 4.243559374942618e-05,
      "loss": 1.5315,
      "step": 14290
    },
    {
      "epoch": 2.3631481098946496,
      "grad_norm": 21.186439514160156,
      "learning_rate": 4.242641253052756e-05,
      "loss": 1.7509,
      "step": 14300
    },
    {
      "epoch": 2.36480066102045,
      "grad_norm": 14.802777290344238,
      "learning_rate": 4.241723131162893e-05,
      "loss": 1.6284,
      "step": 14310
    },
    {
      "epoch": 2.3664532121462507,
      "grad_norm": 7.647261142730713,
      "learning_rate": 4.240805009273031e-05,
      "loss": 1.6912,
      "step": 14320
    },
    {
      "epoch": 2.3681057632720512,
      "grad_norm": 12.083390235900879,
      "learning_rate": 4.239886887383169e-05,
      "loss": 1.5557,
      "step": 14330
    },
    {
      "epoch": 2.369758314397852,
      "grad_norm": 7.496702194213867,
      "learning_rate": 4.238968765493307e-05,
      "loss": 1.7399,
      "step": 14340
    },
    {
      "epoch": 2.3714108655236523,
      "grad_norm": 13.159797668457031,
      "learning_rate": 4.238050643603445e-05,
      "loss": 1.5162,
      "step": 14350
    },
    {
      "epoch": 2.3730634166494524,
      "grad_norm": 6.241695404052734,
      "learning_rate": 4.2371325217135826e-05,
      "loss": 1.6829,
      "step": 14360
    },
    {
      "epoch": 2.374715967775253,
      "grad_norm": 10.561747550964355,
      "learning_rate": 4.236214399823721e-05,
      "loss": 1.6487,
      "step": 14370
    },
    {
      "epoch": 2.3763685189010535,
      "grad_norm": 24.415855407714844,
      "learning_rate": 4.235296277933859e-05,
      "loss": 1.7504,
      "step": 14380
    },
    {
      "epoch": 2.378021070026854,
      "grad_norm": 6.549388885498047,
      "learning_rate": 4.234378156043997e-05,
      "loss": 1.4467,
      "step": 14390
    },
    {
      "epoch": 2.3796736211526546,
      "grad_norm": 8.12175464630127,
      "learning_rate": 4.233460034154135e-05,
      "loss": 1.6034,
      "step": 14400
    },
    {
      "epoch": 2.381326172278455,
      "grad_norm": 10.860891342163086,
      "learning_rate": 4.2325419122642725e-05,
      "loss": 1.5926,
      "step": 14410
    },
    {
      "epoch": 2.382978723404255,
      "grad_norm": 5.750729560852051,
      "learning_rate": 4.231623790374411e-05,
      "loss": 1.5773,
      "step": 14420
    },
    {
      "epoch": 2.3846312745300557,
      "grad_norm": 4.9887871742248535,
      "learning_rate": 4.2307056684845484e-05,
      "loss": 1.5499,
      "step": 14430
    },
    {
      "epoch": 2.3862838256558563,
      "grad_norm": 15.247535705566406,
      "learning_rate": 4.229787546594686e-05,
      "loss": 1.6275,
      "step": 14440
    },
    {
      "epoch": 2.387936376781657,
      "grad_norm": 5.501838207244873,
      "learning_rate": 4.2288694247048236e-05,
      "loss": 1.507,
      "step": 14450
    },
    {
      "epoch": 2.3895889279074574,
      "grad_norm": 6.640819549560547,
      "learning_rate": 4.227951302814962e-05,
      "loss": 1.5051,
      "step": 14460
    },
    {
      "epoch": 2.3912414790332575,
      "grad_norm": 6.430408477783203,
      "learning_rate": 4.2270331809250994e-05,
      "loss": 1.7827,
      "step": 14470
    },
    {
      "epoch": 2.392894030159058,
      "grad_norm": 10.69271183013916,
      "learning_rate": 4.226115059035238e-05,
      "loss": 1.6772,
      "step": 14480
    },
    {
      "epoch": 2.3945465812848585,
      "grad_norm": 12.075413703918457,
      "learning_rate": 4.225196937145376e-05,
      "loss": 1.586,
      "step": 14490
    },
    {
      "epoch": 2.396199132410659,
      "grad_norm": 12.302436828613281,
      "learning_rate": 4.2242788152555135e-05,
      "loss": 1.7124,
      "step": 14500
    },
    {
      "epoch": 2.3978516835364596,
      "grad_norm": 19.029186248779297,
      "learning_rate": 4.223360693365652e-05,
      "loss": 1.5647,
      "step": 14510
    },
    {
      "epoch": 2.3995042346622597,
      "grad_norm": 10.742748260498047,
      "learning_rate": 4.2224425714757894e-05,
      "loss": 1.6184,
      "step": 14520
    },
    {
      "epoch": 2.4011567857880602,
      "grad_norm": 7.062446594238281,
      "learning_rate": 4.2215244495859276e-05,
      "loss": 1.5524,
      "step": 14530
    },
    {
      "epoch": 2.402809336913861,
      "grad_norm": 9.456984519958496,
      "learning_rate": 4.220606327696065e-05,
      "loss": 1.4961,
      "step": 14540
    },
    {
      "epoch": 2.4044618880396613,
      "grad_norm": 17.99965476989746,
      "learning_rate": 4.2196882058062035e-05,
      "loss": 1.5853,
      "step": 14550
    },
    {
      "epoch": 2.406114439165462,
      "grad_norm": 56.12897491455078,
      "learning_rate": 4.218770083916341e-05,
      "loss": 1.6261,
      "step": 14560
    },
    {
      "epoch": 2.407766990291262,
      "grad_norm": 7.298854827880859,
      "learning_rate": 4.217851962026479e-05,
      "loss": 1.8109,
      "step": 14570
    },
    {
      "epoch": 2.4094195414170625,
      "grad_norm": 6.553171634674072,
      "learning_rate": 4.216933840136616e-05,
      "loss": 1.5705,
      "step": 14580
    },
    {
      "epoch": 2.411072092542863,
      "grad_norm": 18.439796447753906,
      "learning_rate": 4.2160157182467545e-05,
      "loss": 1.4941,
      "step": 14590
    },
    {
      "epoch": 2.4127246436686636,
      "grad_norm": 16.03473663330078,
      "learning_rate": 4.215097596356892e-05,
      "loss": 1.5356,
      "step": 14600
    },
    {
      "epoch": 2.414377194794464,
      "grad_norm": 23.698720932006836,
      "learning_rate": 4.2141794744670304e-05,
      "loss": 1.6713,
      "step": 14610
    },
    {
      "epoch": 2.416029745920264,
      "grad_norm": 12.044432640075684,
      "learning_rate": 4.2132613525771686e-05,
      "loss": 1.6328,
      "step": 14620
    },
    {
      "epoch": 2.4176822970460647,
      "grad_norm": 44.36164093017578,
      "learning_rate": 4.212343230687306e-05,
      "loss": 1.5393,
      "step": 14630
    },
    {
      "epoch": 2.4193348481718653,
      "grad_norm": 8.344030380249023,
      "learning_rate": 4.2114251087974445e-05,
      "loss": 1.7533,
      "step": 14640
    },
    {
      "epoch": 2.420987399297666,
      "grad_norm": 17.99836540222168,
      "learning_rate": 4.210506986907582e-05,
      "loss": 1.7182,
      "step": 14650
    },
    {
      "epoch": 2.4226399504234664,
      "grad_norm": 31.487668991088867,
      "learning_rate": 4.20958886501772e-05,
      "loss": 1.5901,
      "step": 14660
    },
    {
      "epoch": 2.4242925015492665,
      "grad_norm": 10.33210277557373,
      "learning_rate": 4.208670743127858e-05,
      "loss": 1.5997,
      "step": 14670
    },
    {
      "epoch": 2.425945052675067,
      "grad_norm": 7.618146896362305,
      "learning_rate": 4.207752621237996e-05,
      "loss": 1.5562,
      "step": 14680
    },
    {
      "epoch": 2.4275976038008675,
      "grad_norm": 39.23401641845703,
      "learning_rate": 4.206834499348134e-05,
      "loss": 1.5395,
      "step": 14690
    },
    {
      "epoch": 2.429250154926668,
      "grad_norm": 32.31549072265625,
      "learning_rate": 4.2059163774582713e-05,
      "loss": 1.5347,
      "step": 14700
    },
    {
      "epoch": 2.4309027060524686,
      "grad_norm": 17.82357406616211,
      "learning_rate": 4.204998255568409e-05,
      "loss": 1.5538,
      "step": 14710
    },
    {
      "epoch": 2.4325552571782687,
      "grad_norm": 9.20028305053711,
      "learning_rate": 4.204080133678547e-05,
      "loss": 1.6732,
      "step": 14720
    },
    {
      "epoch": 2.4342078083040692,
      "grad_norm": 5.321588039398193,
      "learning_rate": 4.2031620117886855e-05,
      "loss": 1.5215,
      "step": 14730
    },
    {
      "epoch": 2.43586035942987,
      "grad_norm": 21.302366256713867,
      "learning_rate": 4.202243889898823e-05,
      "loss": 1.4867,
      "step": 14740
    },
    {
      "epoch": 2.4375129105556703,
      "grad_norm": 6.692672252655029,
      "learning_rate": 4.201325768008961e-05,
      "loss": 1.5802,
      "step": 14750
    },
    {
      "epoch": 2.439165461681471,
      "grad_norm": 36.045379638671875,
      "learning_rate": 4.200407646119099e-05,
      "loss": 1.465,
      "step": 14760
    },
    {
      "epoch": 2.4408180128072714,
      "grad_norm": 11.485674858093262,
      "learning_rate": 4.199489524229237e-05,
      "loss": 1.5289,
      "step": 14770
    },
    {
      "epoch": 2.4424705639330715,
      "grad_norm": 11.762076377868652,
      "learning_rate": 4.198571402339375e-05,
      "loss": 1.5009,
      "step": 14780
    },
    {
      "epoch": 2.444123115058872,
      "grad_norm": 9.65092658996582,
      "learning_rate": 4.197653280449513e-05,
      "loss": 1.7058,
      "step": 14790
    },
    {
      "epoch": 2.4457756661846726,
      "grad_norm": 12.469741821289062,
      "learning_rate": 4.1967351585596506e-05,
      "loss": 1.5731,
      "step": 14800
    },
    {
      "epoch": 2.447428217310473,
      "grad_norm": 48.536380767822266,
      "learning_rate": 4.195817036669789e-05,
      "loss": 1.509,
      "step": 14810
    },
    {
      "epoch": 2.4490807684362736,
      "grad_norm": 9.138647079467773,
      "learning_rate": 4.1948989147799264e-05,
      "loss": 1.5453,
      "step": 14820
    },
    {
      "epoch": 2.450733319562074,
      "grad_norm": 13.465476036071777,
      "learning_rate": 4.193980792890064e-05,
      "loss": 1.4734,
      "step": 14830
    },
    {
      "epoch": 2.4523858706878743,
      "grad_norm": 9.56730842590332,
      "learning_rate": 4.193062671000202e-05,
      "loss": 1.6971,
      "step": 14840
    },
    {
      "epoch": 2.454038421813675,
      "grad_norm": 30.218923568725586,
      "learning_rate": 4.19214454911034e-05,
      "loss": 1.6432,
      "step": 14850
    },
    {
      "epoch": 2.4556909729394754,
      "grad_norm": 9.52951717376709,
      "learning_rate": 4.191226427220478e-05,
      "loss": 1.5145,
      "step": 14860
    },
    {
      "epoch": 2.457343524065276,
      "grad_norm": 15.97042179107666,
      "learning_rate": 4.190308305330616e-05,
      "loss": 1.5629,
      "step": 14870
    },
    {
      "epoch": 2.4589960751910764,
      "grad_norm": 7.123246669769287,
      "learning_rate": 4.189390183440754e-05,
      "loss": 1.6142,
      "step": 14880
    },
    {
      "epoch": 2.4606486263168765,
      "grad_norm": 10.944896697998047,
      "learning_rate": 4.1884720615508916e-05,
      "loss": 1.4761,
      "step": 14890
    },
    {
      "epoch": 2.462301177442677,
      "grad_norm": 14.6271333694458,
      "learning_rate": 4.18755393966103e-05,
      "loss": 1.6334,
      "step": 14900
    },
    {
      "epoch": 2.4639537285684776,
      "grad_norm": 13.755815505981445,
      "learning_rate": 4.1866358177711674e-05,
      "loss": 1.6396,
      "step": 14910
    },
    {
      "epoch": 2.465606279694278,
      "grad_norm": 5.795435428619385,
      "learning_rate": 4.185717695881306e-05,
      "loss": 1.6156,
      "step": 14920
    },
    {
      "epoch": 2.4672588308200787,
      "grad_norm": 8.652409553527832,
      "learning_rate": 4.184799573991443e-05,
      "loss": 1.5923,
      "step": 14930
    },
    {
      "epoch": 2.468911381945879,
      "grad_norm": 34.65216064453125,
      "learning_rate": 4.1838814521015815e-05,
      "loss": 1.6352,
      "step": 14940
    },
    {
      "epoch": 2.4705639330716793,
      "grad_norm": 8.559181213378906,
      "learning_rate": 4.182963330211719e-05,
      "loss": 1.567,
      "step": 14950
    },
    {
      "epoch": 2.47221648419748,
      "grad_norm": 9.969573020935059,
      "learning_rate": 4.182045208321857e-05,
      "loss": 1.5711,
      "step": 14960
    },
    {
      "epoch": 2.4738690353232804,
      "grad_norm": 18.179960250854492,
      "learning_rate": 4.181127086431995e-05,
      "loss": 1.6054,
      "step": 14970
    },
    {
      "epoch": 2.475521586449081,
      "grad_norm": 14.064473152160645,
      "learning_rate": 4.1802089645421326e-05,
      "loss": 1.6849,
      "step": 14980
    },
    {
      "epoch": 2.477174137574881,
      "grad_norm": 32.33445739746094,
      "learning_rate": 4.179290842652271e-05,
      "loss": 1.6486,
      "step": 14990
    },
    {
      "epoch": 2.4788266887006816,
      "grad_norm": 10.423996925354004,
      "learning_rate": 4.1783727207624084e-05,
      "loss": 1.5323,
      "step": 15000
    },
    {
      "epoch": 2.480479239826482,
      "grad_norm": 18.493953704833984,
      "learning_rate": 4.177454598872547e-05,
      "loss": 1.6395,
      "step": 15010
    },
    {
      "epoch": 2.4821317909522826,
      "grad_norm": 12.233598709106445,
      "learning_rate": 4.176536476982684e-05,
      "loss": 1.6386,
      "step": 15020
    },
    {
      "epoch": 2.483784342078083,
      "grad_norm": 6.207724571228027,
      "learning_rate": 4.1756183550928225e-05,
      "loss": 1.6277,
      "step": 15030
    },
    {
      "epoch": 2.4854368932038833,
      "grad_norm": 29.125558853149414,
      "learning_rate": 4.17470023320296e-05,
      "loss": 1.7678,
      "step": 15040
    },
    {
      "epoch": 2.487089444329684,
      "grad_norm": 155.11293029785156,
      "learning_rate": 4.1737821113130984e-05,
      "loss": 1.6569,
      "step": 15050
    },
    {
      "epoch": 2.4887419954554844,
      "grad_norm": 11.938981056213379,
      "learning_rate": 4.172863989423236e-05,
      "loss": 1.4698,
      "step": 15060
    },
    {
      "epoch": 2.490394546581285,
      "grad_norm": 17.687034606933594,
      "learning_rate": 4.171945867533374e-05,
      "loss": 1.4958,
      "step": 15070
    },
    {
      "epoch": 2.4920470977070854,
      "grad_norm": 18.96551513671875,
      "learning_rate": 4.171027745643512e-05,
      "loss": 1.5655,
      "step": 15080
    },
    {
      "epoch": 2.4936996488328855,
      "grad_norm": 151.6097869873047,
      "learning_rate": 4.1701096237536494e-05,
      "loss": 1.553,
      "step": 15090
    },
    {
      "epoch": 2.495352199958686,
      "grad_norm": 18.68946075439453,
      "learning_rate": 4.1691915018637877e-05,
      "loss": 1.7066,
      "step": 15100
    },
    {
      "epoch": 2.4970047510844866,
      "grad_norm": 6.90726375579834,
      "learning_rate": 4.168273379973925e-05,
      "loss": 1.5377,
      "step": 15110
    },
    {
      "epoch": 2.498657302210287,
      "grad_norm": 5.961545467376709,
      "learning_rate": 4.1673552580840635e-05,
      "loss": 1.6982,
      "step": 15120
    },
    {
      "epoch": 2.5003098533360877,
      "grad_norm": 14.853984832763672,
      "learning_rate": 4.166437136194201e-05,
      "loss": 1.4802,
      "step": 15130
    },
    {
      "epoch": 2.501962404461888,
      "grad_norm": 22.466632843017578,
      "learning_rate": 4.1655190143043394e-05,
      "loss": 1.7366,
      "step": 15140
    },
    {
      "epoch": 2.5036149555876888,
      "grad_norm": 17.365001678466797,
      "learning_rate": 4.164600892414477e-05,
      "loss": 1.574,
      "step": 15150
    },
    {
      "epoch": 2.505267506713489,
      "grad_norm": 14.483482360839844,
      "learning_rate": 4.163682770524615e-05,
      "loss": 1.7487,
      "step": 15160
    },
    {
      "epoch": 2.5069200578392894,
      "grad_norm": 16.953773498535156,
      "learning_rate": 4.162764648634753e-05,
      "loss": 1.6225,
      "step": 15170
    },
    {
      "epoch": 2.50857260896509,
      "grad_norm": 10.457685470581055,
      "learning_rate": 4.161846526744891e-05,
      "loss": 1.7095,
      "step": 15180
    },
    {
      "epoch": 2.5102251600908905,
      "grad_norm": 17.119020462036133,
      "learning_rate": 4.160928404855029e-05,
      "loss": 1.6546,
      "step": 15190
    },
    {
      "epoch": 2.511877711216691,
      "grad_norm": 5.457040309906006,
      "learning_rate": 4.160010282965167e-05,
      "loss": 1.5472,
      "step": 15200
    },
    {
      "epoch": 2.513530262342491,
      "grad_norm": 11.147836685180664,
      "learning_rate": 4.1590921610753045e-05,
      "loss": 1.6512,
      "step": 15210
    },
    {
      "epoch": 2.5151828134682916,
      "grad_norm": 9.593242645263672,
      "learning_rate": 4.158174039185442e-05,
      "loss": 1.6628,
      "step": 15220
    },
    {
      "epoch": 2.516835364594092,
      "grad_norm": 11.19896125793457,
      "learning_rate": 4.1572559172955803e-05,
      "loss": 1.6524,
      "step": 15230
    },
    {
      "epoch": 2.5184879157198927,
      "grad_norm": 7.335904121398926,
      "learning_rate": 4.156337795405718e-05,
      "loss": 1.5542,
      "step": 15240
    },
    {
      "epoch": 2.5201404668456933,
      "grad_norm": 15.49926471710205,
      "learning_rate": 4.155419673515856e-05,
      "loss": 1.5335,
      "step": 15250
    },
    {
      "epoch": 2.5217930179714934,
      "grad_norm": 8.069356918334961,
      "learning_rate": 4.154501551625994e-05,
      "loss": 1.498,
      "step": 15260
    },
    {
      "epoch": 2.523445569097294,
      "grad_norm": 12.631221771240234,
      "learning_rate": 4.153583429736132e-05,
      "loss": 1.5483,
      "step": 15270
    },
    {
      "epoch": 2.5250981202230944,
      "grad_norm": 10.724843978881836,
      "learning_rate": 4.1526653078462696e-05,
      "loss": 1.459,
      "step": 15280
    },
    {
      "epoch": 2.526750671348895,
      "grad_norm": 28.29610252380371,
      "learning_rate": 4.151747185956408e-05,
      "loss": 1.5287,
      "step": 15290
    },
    {
      "epoch": 2.5284032224746955,
      "grad_norm": 11.114155769348145,
      "learning_rate": 4.150829064066546e-05,
      "loss": 1.5773,
      "step": 15300
    },
    {
      "epoch": 2.5300557736004956,
      "grad_norm": 16.751001358032227,
      "learning_rate": 4.149910942176684e-05,
      "loss": 1.6019,
      "step": 15310
    },
    {
      "epoch": 2.531708324726296,
      "grad_norm": 12.741584777832031,
      "learning_rate": 4.148992820286822e-05,
      "loss": 1.602,
      "step": 15320
    },
    {
      "epoch": 2.5333608758520967,
      "grad_norm": 14.595974922180176,
      "learning_rate": 4.1480746983969596e-05,
      "loss": 1.5229,
      "step": 15330
    },
    {
      "epoch": 2.535013426977897,
      "grad_norm": 9.688117027282715,
      "learning_rate": 4.147156576507097e-05,
      "loss": 1.573,
      "step": 15340
    },
    {
      "epoch": 2.5366659781036978,
      "grad_norm": 5.253176212310791,
      "learning_rate": 4.146238454617235e-05,
      "loss": 1.6018,
      "step": 15350
    },
    {
      "epoch": 2.538318529229498,
      "grad_norm": 11.29525375366211,
      "learning_rate": 4.145320332727373e-05,
      "loss": 1.5689,
      "step": 15360
    },
    {
      "epoch": 2.5399710803552984,
      "grad_norm": 10.628171920776367,
      "learning_rate": 4.1444022108375106e-05,
      "loss": 1.4875,
      "step": 15370
    },
    {
      "epoch": 2.541623631481099,
      "grad_norm": 11.296927452087402,
      "learning_rate": 4.143484088947649e-05,
      "loss": 1.6718,
      "step": 15380
    },
    {
      "epoch": 2.5432761826068995,
      "grad_norm": 9.183626174926758,
      "learning_rate": 4.1425659670577865e-05,
      "loss": 1.6317,
      "step": 15390
    },
    {
      "epoch": 2.5449287337327,
      "grad_norm": 7.3979268074035645,
      "learning_rate": 4.141647845167925e-05,
      "loss": 1.6035,
      "step": 15400
    },
    {
      "epoch": 2.5465812848585,
      "grad_norm": 9.273728370666504,
      "learning_rate": 4.140729723278063e-05,
      "loss": 1.596,
      "step": 15410
    },
    {
      "epoch": 2.5482338359843006,
      "grad_norm": 16.016374588012695,
      "learning_rate": 4.1398116013882006e-05,
      "loss": 1.5189,
      "step": 15420
    },
    {
      "epoch": 2.549886387110101,
      "grad_norm": 7.42404317855835,
      "learning_rate": 4.138893479498339e-05,
      "loss": 1.6049,
      "step": 15430
    },
    {
      "epoch": 2.5515389382359017,
      "grad_norm": 14.383243560791016,
      "learning_rate": 4.1379753576084764e-05,
      "loss": 1.6131,
      "step": 15440
    },
    {
      "epoch": 2.5531914893617023,
      "grad_norm": 7.443404197692871,
      "learning_rate": 4.137057235718615e-05,
      "loss": 1.7355,
      "step": 15450
    },
    {
      "epoch": 2.5548440404875024,
      "grad_norm": 7.681070327758789,
      "learning_rate": 4.136139113828752e-05,
      "loss": 1.6039,
      "step": 15460
    },
    {
      "epoch": 2.556496591613303,
      "grad_norm": 17.475801467895508,
      "learning_rate": 4.13522099193889e-05,
      "loss": 1.5287,
      "step": 15470
    },
    {
      "epoch": 2.5581491427391034,
      "grad_norm": 11.472020149230957,
      "learning_rate": 4.1343028700490274e-05,
      "loss": 1.572,
      "step": 15480
    },
    {
      "epoch": 2.559801693864904,
      "grad_norm": 8.961161613464355,
      "learning_rate": 4.133384748159166e-05,
      "loss": 1.6715,
      "step": 15490
    },
    {
      "epoch": 2.5614542449907045,
      "grad_norm": 9.045692443847656,
      "learning_rate": 4.132466626269303e-05,
      "loss": 1.6285,
      "step": 15500
    },
    {
      "epoch": 2.5631067961165046,
      "grad_norm": 9.706321716308594,
      "learning_rate": 4.1315485043794416e-05,
      "loss": 1.6204,
      "step": 15510
    },
    {
      "epoch": 2.564759347242305,
      "grad_norm": 22.907732009887695,
      "learning_rate": 4.13063038248958e-05,
      "loss": 1.6672,
      "step": 15520
    },
    {
      "epoch": 2.5664118983681057,
      "grad_norm": 13.848122596740723,
      "learning_rate": 4.1297122605997174e-05,
      "loss": 1.7706,
      "step": 15530
    },
    {
      "epoch": 2.568064449493906,
      "grad_norm": 9.652771949768066,
      "learning_rate": 4.128794138709856e-05,
      "loss": 1.6259,
      "step": 15540
    },
    {
      "epoch": 2.5697170006197068,
      "grad_norm": 12.885273933410645,
      "learning_rate": 4.127876016819993e-05,
      "loss": 1.7457,
      "step": 15550
    },
    {
      "epoch": 2.571369551745507,
      "grad_norm": 6.977317810058594,
      "learning_rate": 4.1269578949301315e-05,
      "loss": 1.5125,
      "step": 15560
    },
    {
      "epoch": 2.573022102871308,
      "grad_norm": 34.75843811035156,
      "learning_rate": 4.126039773040269e-05,
      "loss": 1.6154,
      "step": 15570
    },
    {
      "epoch": 2.574674653997108,
      "grad_norm": 9.55295181274414,
      "learning_rate": 4.1251216511504074e-05,
      "loss": 1.5819,
      "step": 15580
    },
    {
      "epoch": 2.5763272051229085,
      "grad_norm": 13.800176620483398,
      "learning_rate": 4.124203529260545e-05,
      "loss": 1.5153,
      "step": 15590
    },
    {
      "epoch": 2.577979756248709,
      "grad_norm": 53.50123977661133,
      "learning_rate": 4.1232854073706825e-05,
      "loss": 1.6517,
      "step": 15600
    },
    {
      "epoch": 2.5796323073745095,
      "grad_norm": 24.61274528503418,
      "learning_rate": 4.12236728548082e-05,
      "loss": 1.6843,
      "step": 15610
    },
    {
      "epoch": 2.58128485850031,
      "grad_norm": 8.19063663482666,
      "learning_rate": 4.1214491635909584e-05,
      "loss": 1.5765,
      "step": 15620
    },
    {
      "epoch": 2.58293740962611,
      "grad_norm": 8.86671257019043,
      "learning_rate": 4.120531041701096e-05,
      "loss": 1.5666,
      "step": 15630
    },
    {
      "epoch": 2.5845899607519107,
      "grad_norm": 8.37794303894043,
      "learning_rate": 4.119612919811234e-05,
      "loss": 1.6589,
      "step": 15640
    },
    {
      "epoch": 2.5862425118777113,
      "grad_norm": 12.301908493041992,
      "learning_rate": 4.1186947979213725e-05,
      "loss": 1.5103,
      "step": 15650
    },
    {
      "epoch": 2.587895063003512,
      "grad_norm": 21.372512817382812,
      "learning_rate": 4.11777667603151e-05,
      "loss": 1.5983,
      "step": 15660
    },
    {
      "epoch": 2.5895476141293123,
      "grad_norm": 10.715096473693848,
      "learning_rate": 4.1168585541416484e-05,
      "loss": 1.5617,
      "step": 15670
    },
    {
      "epoch": 2.5912001652551124,
      "grad_norm": 24.612468719482422,
      "learning_rate": 4.115940432251786e-05,
      "loss": 1.6039,
      "step": 15680
    },
    {
      "epoch": 2.592852716380913,
      "grad_norm": 26.50051498413086,
      "learning_rate": 4.115022310361924e-05,
      "loss": 1.5615,
      "step": 15690
    },
    {
      "epoch": 2.5945052675067135,
      "grad_norm": 21.82634162902832,
      "learning_rate": 4.114104188472062e-05,
      "loss": 1.5786,
      "step": 15700
    },
    {
      "epoch": 2.596157818632514,
      "grad_norm": 7.950484275817871,
      "learning_rate": 4.1131860665822e-05,
      "loss": 1.6964,
      "step": 15710
    },
    {
      "epoch": 2.5978103697583146,
      "grad_norm": 30.824180603027344,
      "learning_rate": 4.1122679446923376e-05,
      "loss": 1.514,
      "step": 15720
    },
    {
      "epoch": 2.5994629208841147,
      "grad_norm": 7.380947113037109,
      "learning_rate": 4.111349822802475e-05,
      "loss": 1.6425,
      "step": 15730
    },
    {
      "epoch": 2.601115472009915,
      "grad_norm": 7.936986923217773,
      "learning_rate": 4.110431700912613e-05,
      "loss": 1.5766,
      "step": 15740
    },
    {
      "epoch": 2.6027680231357158,
      "grad_norm": 9.09585189819336,
      "learning_rate": 4.109513579022751e-05,
      "loss": 1.6068,
      "step": 15750
    },
    {
      "epoch": 2.6044205742615163,
      "grad_norm": 23.57267951965332,
      "learning_rate": 4.108595457132889e-05,
      "loss": 1.4544,
      "step": 15760
    },
    {
      "epoch": 2.606073125387317,
      "grad_norm": 8.182823181152344,
      "learning_rate": 4.107677335243027e-05,
      "loss": 1.5769,
      "step": 15770
    },
    {
      "epoch": 2.607725676513117,
      "grad_norm": 15.561851501464844,
      "learning_rate": 4.106759213353165e-05,
      "loss": 1.5978,
      "step": 15780
    },
    {
      "epoch": 2.6093782276389175,
      "grad_norm": 11.034695625305176,
      "learning_rate": 4.105841091463303e-05,
      "loss": 1.608,
      "step": 15790
    },
    {
      "epoch": 2.611030778764718,
      "grad_norm": 63.234703063964844,
      "learning_rate": 4.104922969573441e-05,
      "loss": 1.6947,
      "step": 15800
    },
    {
      "epoch": 2.6126833298905185,
      "grad_norm": 8.965785026550293,
      "learning_rate": 4.1040048476835786e-05,
      "loss": 1.6128,
      "step": 15810
    },
    {
      "epoch": 2.614335881016319,
      "grad_norm": 7.541203022003174,
      "learning_rate": 4.103086725793717e-05,
      "loss": 1.5494,
      "step": 15820
    },
    {
      "epoch": 2.615988432142119,
      "grad_norm": 12.703873634338379,
      "learning_rate": 4.1021686039038545e-05,
      "loss": 1.6318,
      "step": 15830
    },
    {
      "epoch": 2.6176409832679197,
      "grad_norm": 8.111274719238281,
      "learning_rate": 4.101250482013993e-05,
      "loss": 1.5468,
      "step": 15840
    },
    {
      "epoch": 2.6192935343937203,
      "grad_norm": 23.11355209350586,
      "learning_rate": 4.10033236012413e-05,
      "loss": 1.6287,
      "step": 15850
    },
    {
      "epoch": 2.620946085519521,
      "grad_norm": 12.819192886352539,
      "learning_rate": 4.099414238234268e-05,
      "loss": 1.6666,
      "step": 15860
    },
    {
      "epoch": 2.6225986366453213,
      "grad_norm": 9.071503639221191,
      "learning_rate": 4.098496116344406e-05,
      "loss": 1.6456,
      "step": 15870
    },
    {
      "epoch": 2.6242511877711214,
      "grad_norm": 14.59699821472168,
      "learning_rate": 4.097577994454544e-05,
      "loss": 1.5137,
      "step": 15880
    },
    {
      "epoch": 2.625903738896922,
      "grad_norm": 11.511642456054688,
      "learning_rate": 4.096659872564682e-05,
      "loss": 1.5802,
      "step": 15890
    },
    {
      "epoch": 2.6275562900227225,
      "grad_norm": 13.53958511352539,
      "learning_rate": 4.0957417506748196e-05,
      "loss": 1.704,
      "step": 15900
    },
    {
      "epoch": 2.629208841148523,
      "grad_norm": 12.743249893188477,
      "learning_rate": 4.094823628784958e-05,
      "loss": 1.567,
      "step": 15910
    },
    {
      "epoch": 2.6308613922743236,
      "grad_norm": 31.744903564453125,
      "learning_rate": 4.0939055068950955e-05,
      "loss": 1.4322,
      "step": 15920
    },
    {
      "epoch": 2.6325139434001237,
      "grad_norm": 12.78609848022461,
      "learning_rate": 4.092987385005234e-05,
      "loss": 1.7043,
      "step": 15930
    },
    {
      "epoch": 2.6341664945259247,
      "grad_norm": 19.535892486572266,
      "learning_rate": 4.092069263115371e-05,
      "loss": 1.569,
      "step": 15940
    },
    {
      "epoch": 2.6358190456517248,
      "grad_norm": 13.720259666442871,
      "learning_rate": 4.0911511412255096e-05,
      "loss": 1.6444,
      "step": 15950
    },
    {
      "epoch": 2.6374715967775253,
      "grad_norm": 7.525299549102783,
      "learning_rate": 4.090233019335647e-05,
      "loss": 1.511,
      "step": 15960
    },
    {
      "epoch": 2.639124147903326,
      "grad_norm": 12.380800247192383,
      "learning_rate": 4.0893148974457854e-05,
      "loss": 1.6673,
      "step": 15970
    },
    {
      "epoch": 2.6407766990291264,
      "grad_norm": 14.61955451965332,
      "learning_rate": 4.088396775555923e-05,
      "loss": 1.5086,
      "step": 15980
    },
    {
      "epoch": 2.642429250154927,
      "grad_norm": 11.027737617492676,
      "learning_rate": 4.0874786536660606e-05,
      "loss": 1.7163,
      "step": 15990
    },
    {
      "epoch": 2.644081801280727,
      "grad_norm": 19.45004653930664,
      "learning_rate": 4.086560531776199e-05,
      "loss": 1.6217,
      "step": 16000
    },
    {
      "epoch": 2.6457343524065275,
      "grad_norm": 13.399431228637695,
      "learning_rate": 4.0856424098863364e-05,
      "loss": 1.5046,
      "step": 16010
    },
    {
      "epoch": 2.647386903532328,
      "grad_norm": 11.20180892944336,
      "learning_rate": 4.084724287996475e-05,
      "loss": 1.5504,
      "step": 16020
    },
    {
      "epoch": 2.6490394546581286,
      "grad_norm": 24.36643409729004,
      "learning_rate": 4.083806166106612e-05,
      "loss": 1.633,
      "step": 16030
    },
    {
      "epoch": 2.650692005783929,
      "grad_norm": 9.366679191589355,
      "learning_rate": 4.0828880442167506e-05,
      "loss": 1.6683,
      "step": 16040
    },
    {
      "epoch": 2.6523445569097293,
      "grad_norm": 11.785042762756348,
      "learning_rate": 4.081969922326888e-05,
      "loss": 1.6522,
      "step": 16050
    },
    {
      "epoch": 2.65399710803553,
      "grad_norm": 5.310166358947754,
      "learning_rate": 4.0810518004370264e-05,
      "loss": 1.621,
      "step": 16060
    },
    {
      "epoch": 2.6556496591613303,
      "grad_norm": 39.18266677856445,
      "learning_rate": 4.080133678547164e-05,
      "loss": 1.6034,
      "step": 16070
    },
    {
      "epoch": 2.657302210287131,
      "grad_norm": 12.86594009399414,
      "learning_rate": 4.079215556657302e-05,
      "loss": 1.438,
      "step": 16080
    },
    {
      "epoch": 2.6589547614129314,
      "grad_norm": 13.897305488586426,
      "learning_rate": 4.0782974347674405e-05,
      "loss": 1.4932,
      "step": 16090
    },
    {
      "epoch": 2.6606073125387315,
      "grad_norm": 25.27904510498047,
      "learning_rate": 4.077379312877578e-05,
      "loss": 1.7065,
      "step": 16100
    },
    {
      "epoch": 2.662259863664532,
      "grad_norm": 10.040287017822266,
      "learning_rate": 4.076461190987716e-05,
      "loss": 1.571,
      "step": 16110
    },
    {
      "epoch": 2.6639124147903326,
      "grad_norm": 11.403328895568848,
      "learning_rate": 4.075543069097853e-05,
      "loss": 1.6483,
      "step": 16120
    },
    {
      "epoch": 2.665564965916133,
      "grad_norm": 10.510461807250977,
      "learning_rate": 4.0746249472079915e-05,
      "loss": 1.6841,
      "step": 16130
    },
    {
      "epoch": 2.6672175170419337,
      "grad_norm": 6.346688270568848,
      "learning_rate": 4.073706825318129e-05,
      "loss": 1.6634,
      "step": 16140
    },
    {
      "epoch": 2.6688700681677338,
      "grad_norm": 6.1756696701049805,
      "learning_rate": 4.0727887034282674e-05,
      "loss": 1.6516,
      "step": 16150
    },
    {
      "epoch": 2.6705226192935343,
      "grad_norm": 10.464988708496094,
      "learning_rate": 4.071870581538405e-05,
      "loss": 1.6066,
      "step": 16160
    },
    {
      "epoch": 2.672175170419335,
      "grad_norm": 18.168760299682617,
      "learning_rate": 4.070952459648543e-05,
      "loss": 1.6268,
      "step": 16170
    },
    {
      "epoch": 2.6738277215451354,
      "grad_norm": 31.59029197692871,
      "learning_rate": 4.070034337758681e-05,
      "loss": 1.593,
      "step": 16180
    },
    {
      "epoch": 2.675480272670936,
      "grad_norm": 13.246702194213867,
      "learning_rate": 4.069116215868819e-05,
      "loss": 1.6339,
      "step": 16190
    },
    {
      "epoch": 2.677132823796736,
      "grad_norm": 13.714371681213379,
      "learning_rate": 4.068198093978957e-05,
      "loss": 1.6681,
      "step": 16200
    },
    {
      "epoch": 2.6787853749225365,
      "grad_norm": 9.815982818603516,
      "learning_rate": 4.067279972089095e-05,
      "loss": 1.4579,
      "step": 16210
    },
    {
      "epoch": 2.680437926048337,
      "grad_norm": 11.735651016235352,
      "learning_rate": 4.066361850199233e-05,
      "loss": 1.6187,
      "step": 16220
    },
    {
      "epoch": 2.6820904771741376,
      "grad_norm": 10.747529983520508,
      "learning_rate": 4.065443728309371e-05,
      "loss": 1.5022,
      "step": 16230
    },
    {
      "epoch": 2.683743028299938,
      "grad_norm": 19.077959060668945,
      "learning_rate": 4.0645256064195084e-05,
      "loss": 1.5181,
      "step": 16240
    },
    {
      "epoch": 2.6853955794257383,
      "grad_norm": 23.054729461669922,
      "learning_rate": 4.063607484529646e-05,
      "loss": 1.6234,
      "step": 16250
    },
    {
      "epoch": 2.687048130551539,
      "grad_norm": 6.9372992515563965,
      "learning_rate": 4.062689362639784e-05,
      "loss": 1.5558,
      "step": 16260
    },
    {
      "epoch": 2.6887006816773393,
      "grad_norm": 9.9459867477417,
      "learning_rate": 4.061771240749922e-05,
      "loss": 1.4208,
      "step": 16270
    },
    {
      "epoch": 2.69035323280314,
      "grad_norm": 8.664091110229492,
      "learning_rate": 4.06085311886006e-05,
      "loss": 1.6421,
      "step": 16280
    },
    {
      "epoch": 2.6920057839289404,
      "grad_norm": 11.299137115478516,
      "learning_rate": 4.0599349969701977e-05,
      "loss": 1.6188,
      "step": 16290
    },
    {
      "epoch": 2.6936583350547405,
      "grad_norm": 8.935805320739746,
      "learning_rate": 4.059016875080336e-05,
      "loss": 1.6265,
      "step": 16300
    },
    {
      "epoch": 2.6953108861805415,
      "grad_norm": 8.099499702453613,
      "learning_rate": 4.0580987531904735e-05,
      "loss": 1.4238,
      "step": 16310
    },
    {
      "epoch": 2.6969634373063416,
      "grad_norm": 6.492707252502441,
      "learning_rate": 4.057180631300612e-05,
      "loss": 1.5873,
      "step": 16320
    },
    {
      "epoch": 2.698615988432142,
      "grad_norm": 6.262495517730713,
      "learning_rate": 4.05626250941075e-05,
      "loss": 1.639,
      "step": 16330
    },
    {
      "epoch": 2.7002685395579427,
      "grad_norm": 16.449399948120117,
      "learning_rate": 4.0553443875208876e-05,
      "loss": 1.5878,
      "step": 16340
    },
    {
      "epoch": 2.7019210906837428,
      "grad_norm": 21.750577926635742,
      "learning_rate": 4.054426265631026e-05,
      "loss": 1.7173,
      "step": 16350
    },
    {
      "epoch": 2.7035736418095437,
      "grad_norm": 10.185629844665527,
      "learning_rate": 4.0535081437411635e-05,
      "loss": 1.5977,
      "step": 16360
    },
    {
      "epoch": 2.705226192935344,
      "grad_norm": 66.03086853027344,
      "learning_rate": 4.052590021851301e-05,
      "loss": 1.7398,
      "step": 16370
    },
    {
      "epoch": 2.7068787440611444,
      "grad_norm": 6.185853481292725,
      "learning_rate": 4.0516718999614386e-05,
      "loss": 1.5928,
      "step": 16380
    },
    {
      "epoch": 2.708531295186945,
      "grad_norm": 5.825438499450684,
      "learning_rate": 4.050753778071577e-05,
      "loss": 1.5435,
      "step": 16390
    },
    {
      "epoch": 2.7101838463127454,
      "grad_norm": 19.0218505859375,
      "learning_rate": 4.0498356561817145e-05,
      "loss": 1.6741,
      "step": 16400
    },
    {
      "epoch": 2.711836397438546,
      "grad_norm": 12.254176139831543,
      "learning_rate": 4.048917534291853e-05,
      "loss": 1.5611,
      "step": 16410
    },
    {
      "epoch": 2.713488948564346,
      "grad_norm": 18.687061309814453,
      "learning_rate": 4.04799941240199e-05,
      "loss": 1.5091,
      "step": 16420
    },
    {
      "epoch": 2.7151414996901466,
      "grad_norm": 11.243393898010254,
      "learning_rate": 4.0470812905121286e-05,
      "loss": 1.5715,
      "step": 16430
    },
    {
      "epoch": 2.716794050815947,
      "grad_norm": 15.61385726928711,
      "learning_rate": 4.046163168622267e-05,
      "loss": 1.4761,
      "step": 16440
    },
    {
      "epoch": 2.7184466019417477,
      "grad_norm": 6.631096839904785,
      "learning_rate": 4.0452450467324045e-05,
      "loss": 1.6744,
      "step": 16450
    },
    {
      "epoch": 2.7200991530675482,
      "grad_norm": 31.41004753112793,
      "learning_rate": 4.044326924842543e-05,
      "loss": 1.4479,
      "step": 16460
    },
    {
      "epoch": 2.7217517041933483,
      "grad_norm": 8.312516212463379,
      "learning_rate": 4.04340880295268e-05,
      "loss": 1.4798,
      "step": 16470
    },
    {
      "epoch": 2.723404255319149,
      "grad_norm": 16.965839385986328,
      "learning_rate": 4.0424906810628186e-05,
      "loss": 1.5273,
      "step": 16480
    },
    {
      "epoch": 2.7250568064449494,
      "grad_norm": 22.01923179626465,
      "learning_rate": 4.041572559172956e-05,
      "loss": 1.5548,
      "step": 16490
    },
    {
      "epoch": 2.72670935757075,
      "grad_norm": 8.084673881530762,
      "learning_rate": 4.040654437283094e-05,
      "loss": 1.6711,
      "step": 16500
    },
    {
      "epoch": 2.7283619086965505,
      "grad_norm": 8.843356132507324,
      "learning_rate": 4.039736315393231e-05,
      "loss": 1.5687,
      "step": 16510
    },
    {
      "epoch": 2.7300144598223506,
      "grad_norm": 10.197770118713379,
      "learning_rate": 4.0388181935033696e-05,
      "loss": 1.5118,
      "step": 16520
    },
    {
      "epoch": 2.731667010948151,
      "grad_norm": 38.914649963378906,
      "learning_rate": 4.037900071613507e-05,
      "loss": 1.6277,
      "step": 16530
    },
    {
      "epoch": 2.7333195620739517,
      "grad_norm": 5.684174060821533,
      "learning_rate": 4.0369819497236454e-05,
      "loss": 1.7301,
      "step": 16540
    },
    {
      "epoch": 2.734972113199752,
      "grad_norm": 12.711173057556152,
      "learning_rate": 4.036063827833784e-05,
      "loss": 1.4475,
      "step": 16550
    },
    {
      "epoch": 2.7366246643255527,
      "grad_norm": 48.89714813232422,
      "learning_rate": 4.035145705943921e-05,
      "loss": 1.584,
      "step": 16560
    },
    {
      "epoch": 2.738277215451353,
      "grad_norm": 22.189855575561523,
      "learning_rate": 4.0342275840540595e-05,
      "loss": 1.6205,
      "step": 16570
    },
    {
      "epoch": 2.7399297665771534,
      "grad_norm": 7.027297019958496,
      "learning_rate": 4.033309462164197e-05,
      "loss": 1.6204,
      "step": 16580
    },
    {
      "epoch": 2.741582317702954,
      "grad_norm": 16.53078842163086,
      "learning_rate": 4.0323913402743354e-05,
      "loss": 1.5647,
      "step": 16590
    },
    {
      "epoch": 2.7432348688287544,
      "grad_norm": 14.532068252563477,
      "learning_rate": 4.031473218384473e-05,
      "loss": 1.5833,
      "step": 16600
    },
    {
      "epoch": 2.744887419954555,
      "grad_norm": 17.68378448486328,
      "learning_rate": 4.030555096494611e-05,
      "loss": 1.6564,
      "step": 16610
    },
    {
      "epoch": 2.746539971080355,
      "grad_norm": 8.336248397827148,
      "learning_rate": 4.029636974604749e-05,
      "loss": 1.6376,
      "step": 16620
    },
    {
      "epoch": 2.7481925222061556,
      "grad_norm": 8.331178665161133,
      "learning_rate": 4.0287188527148864e-05,
      "loss": 1.5466,
      "step": 16630
    },
    {
      "epoch": 2.749845073331956,
      "grad_norm": 9.454358100891113,
      "learning_rate": 4.027800730825024e-05,
      "loss": 1.6387,
      "step": 16640
    },
    {
      "epoch": 2.7514976244577567,
      "grad_norm": 19.310138702392578,
      "learning_rate": 4.026882608935162e-05,
      "loss": 1.6703,
      "step": 16650
    },
    {
      "epoch": 2.7531501755835572,
      "grad_norm": 7.6864705085754395,
      "learning_rate": 4.0259644870453005e-05,
      "loss": 1.5447,
      "step": 16660
    },
    {
      "epoch": 2.7548027267093573,
      "grad_norm": 5.934784412384033,
      "learning_rate": 4.025046365155438e-05,
      "loss": 1.4783,
      "step": 16670
    },
    {
      "epoch": 2.756455277835158,
      "grad_norm": 53.78288269042969,
      "learning_rate": 4.0241282432655764e-05,
      "loss": 1.7438,
      "step": 16680
    },
    {
      "epoch": 2.7581078289609584,
      "grad_norm": 9.907873153686523,
      "learning_rate": 4.023210121375714e-05,
      "loss": 1.5094,
      "step": 16690
    },
    {
      "epoch": 2.759760380086759,
      "grad_norm": 17.700838088989258,
      "learning_rate": 4.022291999485852e-05,
      "loss": 1.5519,
      "step": 16700
    },
    {
      "epoch": 2.7614129312125595,
      "grad_norm": 11.269119262695312,
      "learning_rate": 4.02137387759599e-05,
      "loss": 1.661,
      "step": 16710
    },
    {
      "epoch": 2.7630654823383596,
      "grad_norm": 12.648497581481934,
      "learning_rate": 4.020455755706128e-05,
      "loss": 1.7044,
      "step": 16720
    },
    {
      "epoch": 2.7647180334641606,
      "grad_norm": 7.257272720336914,
      "learning_rate": 4.019537633816266e-05,
      "loss": 1.5785,
      "step": 16730
    },
    {
      "epoch": 2.7663705845899607,
      "grad_norm": 10.091726303100586,
      "learning_rate": 4.018619511926404e-05,
      "loss": 1.5245,
      "step": 16740
    },
    {
      "epoch": 2.768023135715761,
      "grad_norm": 23.164775848388672,
      "learning_rate": 4.0177013900365415e-05,
      "loss": 1.5764,
      "step": 16750
    },
    {
      "epoch": 2.7696756868415617,
      "grad_norm": 21.203231811523438,
      "learning_rate": 4.016783268146679e-05,
      "loss": 1.5322,
      "step": 16760
    },
    {
      "epoch": 2.7713282379673623,
      "grad_norm": 8.188257217407227,
      "learning_rate": 4.015865146256817e-05,
      "loss": 1.5544,
      "step": 16770
    },
    {
      "epoch": 2.772980789093163,
      "grad_norm": 5.449336051940918,
      "learning_rate": 4.014947024366955e-05,
      "loss": 1.5638,
      "step": 16780
    },
    {
      "epoch": 2.774633340218963,
      "grad_norm": 30.01235008239746,
      "learning_rate": 4.014028902477093e-05,
      "loss": 1.6193,
      "step": 16790
    },
    {
      "epoch": 2.7762858913447634,
      "grad_norm": 7.071667194366455,
      "learning_rate": 4.013110780587231e-05,
      "loss": 1.5834,
      "step": 16800
    },
    {
      "epoch": 2.777938442470564,
      "grad_norm": 7.050570011138916,
      "learning_rate": 4.012192658697369e-05,
      "loss": 1.4917,
      "step": 16810
    },
    {
      "epoch": 2.7795909935963645,
      "grad_norm": 6.538548469543457,
      "learning_rate": 4.0112745368075067e-05,
      "loss": 1.6527,
      "step": 16820
    },
    {
      "epoch": 2.781243544722165,
      "grad_norm": 10.154224395751953,
      "learning_rate": 4.010356414917645e-05,
      "loss": 1.5722,
      "step": 16830
    },
    {
      "epoch": 2.782896095847965,
      "grad_norm": 13.937148094177246,
      "learning_rate": 4.0094382930277825e-05,
      "loss": 1.6655,
      "step": 16840
    },
    {
      "epoch": 2.7845486469737657,
      "grad_norm": 10.971857070922852,
      "learning_rate": 4.008520171137921e-05,
      "loss": 1.6278,
      "step": 16850
    },
    {
      "epoch": 2.7862011980995662,
      "grad_norm": 19.598770141601562,
      "learning_rate": 4.0076020492480584e-05,
      "loss": 1.6375,
      "step": 16860
    },
    {
      "epoch": 2.7878537492253668,
      "grad_norm": 13.447637557983398,
      "learning_rate": 4.0066839273581966e-05,
      "loss": 1.5709,
      "step": 16870
    },
    {
      "epoch": 2.7895063003511673,
      "grad_norm": 12.843521118164062,
      "learning_rate": 4.005765805468334e-05,
      "loss": 1.6029,
      "step": 16880
    },
    {
      "epoch": 2.7911588514769674,
      "grad_norm": 58.23180389404297,
      "learning_rate": 4.004847683578472e-05,
      "loss": 1.5457,
      "step": 16890
    },
    {
      "epoch": 2.792811402602768,
      "grad_norm": 9.176880836486816,
      "learning_rate": 4.00392956168861e-05,
      "loss": 1.5729,
      "step": 16900
    },
    {
      "epoch": 2.7944639537285685,
      "grad_norm": 5.380598545074463,
      "learning_rate": 4.0030114397987476e-05,
      "loss": 1.5439,
      "step": 16910
    },
    {
      "epoch": 2.796116504854369,
      "grad_norm": 8.638605117797852,
      "learning_rate": 4.002093317908886e-05,
      "loss": 1.5537,
      "step": 16920
    },
    {
      "epoch": 2.7977690559801696,
      "grad_norm": 5.76012659072876,
      "learning_rate": 4.0011751960190235e-05,
      "loss": 1.521,
      "step": 16930
    },
    {
      "epoch": 2.7994216071059697,
      "grad_norm": 8.071260452270508,
      "learning_rate": 4.000257074129162e-05,
      "loss": 1.559,
      "step": 16940
    },
    {
      "epoch": 2.80107415823177,
      "grad_norm": 13.014986991882324,
      "learning_rate": 3.999338952239299e-05,
      "loss": 1.6444,
      "step": 16950
    },
    {
      "epoch": 2.8027267093575707,
      "grad_norm": 5.6536431312561035,
      "learning_rate": 3.9984208303494376e-05,
      "loss": 1.6925,
      "step": 16960
    },
    {
      "epoch": 2.8043792604833713,
      "grad_norm": 6.50076961517334,
      "learning_rate": 3.997502708459575e-05,
      "loss": 1.5527,
      "step": 16970
    },
    {
      "epoch": 2.806031811609172,
      "grad_norm": 11.952144622802734,
      "learning_rate": 3.9965845865697134e-05,
      "loss": 1.6111,
      "step": 16980
    },
    {
      "epoch": 2.807684362734972,
      "grad_norm": 12.421814918518066,
      "learning_rate": 3.995666464679851e-05,
      "loss": 1.5704,
      "step": 16990
    },
    {
      "epoch": 2.8093369138607724,
      "grad_norm": 7.645977973937988,
      "learning_rate": 3.994748342789989e-05,
      "loss": 1.6001,
      "step": 17000
    },
    {
      "epoch": 2.810989464986573,
      "grad_norm": 9.946296691894531,
      "learning_rate": 3.993830220900127e-05,
      "loss": 1.5803,
      "step": 17010
    },
    {
      "epoch": 2.8126420161123735,
      "grad_norm": 61.43771743774414,
      "learning_rate": 3.9929120990102645e-05,
      "loss": 1.5299,
      "step": 17020
    },
    {
      "epoch": 2.814294567238174,
      "grad_norm": 13.09609317779541,
      "learning_rate": 3.991993977120403e-05,
      "loss": 1.5403,
      "step": 17030
    },
    {
      "epoch": 2.815947118363974,
      "grad_norm": 10.614474296569824,
      "learning_rate": 3.99107585523054e-05,
      "loss": 1.5166,
      "step": 17040
    },
    {
      "epoch": 2.8175996694897747,
      "grad_norm": 64.39093780517578,
      "learning_rate": 3.9901577333406786e-05,
      "loss": 1.6604,
      "step": 17050
    },
    {
      "epoch": 2.8192522206155752,
      "grad_norm": 27.681407928466797,
      "learning_rate": 3.989239611450816e-05,
      "loss": 1.6366,
      "step": 17060
    },
    {
      "epoch": 2.8209047717413758,
      "grad_norm": 9.592436790466309,
      "learning_rate": 3.9883214895609544e-05,
      "loss": 1.5281,
      "step": 17070
    },
    {
      "epoch": 2.8225573228671763,
      "grad_norm": 10.364959716796875,
      "learning_rate": 3.987403367671092e-05,
      "loss": 1.6947,
      "step": 17080
    },
    {
      "epoch": 2.8242098739929764,
      "grad_norm": 8.182236671447754,
      "learning_rate": 3.98648524578123e-05,
      "loss": 1.6698,
      "step": 17090
    },
    {
      "epoch": 2.8258624251187774,
      "grad_norm": 13.211899757385254,
      "learning_rate": 3.985567123891368e-05,
      "loss": 1.5847,
      "step": 17100
    },
    {
      "epoch": 2.8275149762445775,
      "grad_norm": 12.27530288696289,
      "learning_rate": 3.984649002001506e-05,
      "loss": 1.6346,
      "step": 17110
    },
    {
      "epoch": 2.829167527370378,
      "grad_norm": 27.445775985717773,
      "learning_rate": 3.9837308801116444e-05,
      "loss": 1.5896,
      "step": 17120
    },
    {
      "epoch": 2.8308200784961786,
      "grad_norm": 107.77072143554688,
      "learning_rate": 3.982812758221782e-05,
      "loss": 1.5398,
      "step": 17130
    },
    {
      "epoch": 2.8324726296219787,
      "grad_norm": 6.350532531738281,
      "learning_rate": 3.9818946363319196e-05,
      "loss": 1.6412,
      "step": 17140
    },
    {
      "epoch": 2.8341251807477796,
      "grad_norm": 10.610336303710938,
      "learning_rate": 3.980976514442057e-05,
      "loss": 1.5777,
      "step": 17150
    },
    {
      "epoch": 2.8357777318735797,
      "grad_norm": 6.003240585327148,
      "learning_rate": 3.9800583925521954e-05,
      "loss": 1.5696,
      "step": 17160
    },
    {
      "epoch": 2.8374302829993803,
      "grad_norm": 7.450899124145508,
      "learning_rate": 3.979140270662333e-05,
      "loss": 1.6595,
      "step": 17170
    },
    {
      "epoch": 2.839082834125181,
      "grad_norm": 10.081241607666016,
      "learning_rate": 3.978222148772471e-05,
      "loss": 1.5032,
      "step": 17180
    },
    {
      "epoch": 2.8407353852509813,
      "grad_norm": 10.23855972290039,
      "learning_rate": 3.977304026882609e-05,
      "loss": 1.5916,
      "step": 17190
    },
    {
      "epoch": 2.842387936376782,
      "grad_norm": 9.486077308654785,
      "learning_rate": 3.976385904992747e-05,
      "loss": 1.6483,
      "step": 17200
    },
    {
      "epoch": 2.844040487502582,
      "grad_norm": 22.05427360534668,
      "learning_rate": 3.975467783102885e-05,
      "loss": 1.6892,
      "step": 17210
    },
    {
      "epoch": 2.8456930386283825,
      "grad_norm": 16.26018524169922,
      "learning_rate": 3.974549661213023e-05,
      "loss": 1.5048,
      "step": 17220
    },
    {
      "epoch": 2.847345589754183,
      "grad_norm": 14.785170555114746,
      "learning_rate": 3.9736315393231606e-05,
      "loss": 1.7072,
      "step": 17230
    },
    {
      "epoch": 2.8489981408799836,
      "grad_norm": 7.535512924194336,
      "learning_rate": 3.972713417433299e-05,
      "loss": 1.6278,
      "step": 17240
    },
    {
      "epoch": 2.850650692005784,
      "grad_norm": 14.22430419921875,
      "learning_rate": 3.971795295543437e-05,
      "loss": 1.6181,
      "step": 17250
    },
    {
      "epoch": 2.8523032431315842,
      "grad_norm": 9.323792457580566,
      "learning_rate": 3.970877173653575e-05,
      "loss": 1.72,
      "step": 17260
    },
    {
      "epoch": 2.8539557942573848,
      "grad_norm": 9.750657081604004,
      "learning_rate": 3.969959051763712e-05,
      "loss": 1.4657,
      "step": 17270
    },
    {
      "epoch": 2.8556083453831853,
      "grad_norm": 16.884572982788086,
      "learning_rate": 3.96904092987385e-05,
      "loss": 1.6819,
      "step": 17280
    },
    {
      "epoch": 2.857260896508986,
      "grad_norm": 29.167482376098633,
      "learning_rate": 3.968122807983988e-05,
      "loss": 1.6451,
      "step": 17290
    },
    {
      "epoch": 2.8589134476347864,
      "grad_norm": 36.17399215698242,
      "learning_rate": 3.967204686094126e-05,
      "loss": 1.6083,
      "step": 17300
    },
    {
      "epoch": 2.8605659987605865,
      "grad_norm": 13.15809440612793,
      "learning_rate": 3.966286564204264e-05,
      "loss": 1.6859,
      "step": 17310
    },
    {
      "epoch": 2.862218549886387,
      "grad_norm": 9.510421752929688,
      "learning_rate": 3.9653684423144015e-05,
      "loss": 1.6486,
      "step": 17320
    },
    {
      "epoch": 2.8638711010121876,
      "grad_norm": 9.244708061218262,
      "learning_rate": 3.96445032042454e-05,
      "loss": 1.5877,
      "step": 17330
    },
    {
      "epoch": 2.865523652137988,
      "grad_norm": 6.563562393188477,
      "learning_rate": 3.9635321985346774e-05,
      "loss": 1.575,
      "step": 17340
    },
    {
      "epoch": 2.8671762032637886,
      "grad_norm": 15.244311332702637,
      "learning_rate": 3.9626140766448156e-05,
      "loss": 1.55,
      "step": 17350
    },
    {
      "epoch": 2.8688287543895887,
      "grad_norm": 5.563783645629883,
      "learning_rate": 3.961695954754954e-05,
      "loss": 1.5702,
      "step": 17360
    },
    {
      "epoch": 2.8704813055153893,
      "grad_norm": 11.255026817321777,
      "learning_rate": 3.9607778328650915e-05,
      "loss": 1.713,
      "step": 17370
    },
    {
      "epoch": 2.87213385664119,
      "grad_norm": 13.83602237701416,
      "learning_rate": 3.95985971097523e-05,
      "loss": 1.4584,
      "step": 17380
    },
    {
      "epoch": 2.8737864077669903,
      "grad_norm": 10.390022277832031,
      "learning_rate": 3.9589415890853673e-05,
      "loss": 1.5963,
      "step": 17390
    },
    {
      "epoch": 2.875438958892791,
      "grad_norm": 6.194431781768799,
      "learning_rate": 3.958023467195505e-05,
      "loss": 1.5349,
      "step": 17400
    },
    {
      "epoch": 2.877091510018591,
      "grad_norm": 6.946686744689941,
      "learning_rate": 3.9571053453056425e-05,
      "loss": 1.5731,
      "step": 17410
    },
    {
      "epoch": 2.8787440611443915,
      "grad_norm": 11.890841484069824,
      "learning_rate": 3.956187223415781e-05,
      "loss": 1.5738,
      "step": 17420
    },
    {
      "epoch": 2.880396612270192,
      "grad_norm": 4.596217632293701,
      "learning_rate": 3.9552691015259184e-05,
      "loss": 1.5325,
      "step": 17430
    },
    {
      "epoch": 2.8820491633959926,
      "grad_norm": 9.431676864624023,
      "learning_rate": 3.9543509796360566e-05,
      "loss": 1.6274,
      "step": 17440
    },
    {
      "epoch": 2.883701714521793,
      "grad_norm": 9.621706008911133,
      "learning_rate": 3.953432857746194e-05,
      "loss": 1.5976,
      "step": 17450
    },
    {
      "epoch": 2.8853542656475932,
      "grad_norm": 6.648024559020996,
      "learning_rate": 3.9525147358563325e-05,
      "loss": 1.5725,
      "step": 17460
    },
    {
      "epoch": 2.8870068167733938,
      "grad_norm": 33.377403259277344,
      "learning_rate": 3.951596613966471e-05,
      "loss": 1.5394,
      "step": 17470
    },
    {
      "epoch": 2.8886593678991943,
      "grad_norm": 7.782329082489014,
      "learning_rate": 3.950678492076608e-05,
      "loss": 1.5998,
      "step": 17480
    },
    {
      "epoch": 2.890311919024995,
      "grad_norm": 6.886441230773926,
      "learning_rate": 3.9497603701867466e-05,
      "loss": 1.6038,
      "step": 17490
    },
    {
      "epoch": 2.8919644701507954,
      "grad_norm": 7.424654006958008,
      "learning_rate": 3.948842248296884e-05,
      "loss": 1.5715,
      "step": 17500
    },
    {
      "epoch": 2.8936170212765955,
      "grad_norm": 35.0805549621582,
      "learning_rate": 3.9479241264070224e-05,
      "loss": 1.5405,
      "step": 17510
    },
    {
      "epoch": 2.8952695724023965,
      "grad_norm": 8.772028923034668,
      "learning_rate": 3.94700600451716e-05,
      "loss": 1.6665,
      "step": 17520
    },
    {
      "epoch": 2.8969221235281966,
      "grad_norm": 10.116393089294434,
      "learning_rate": 3.9460878826272976e-05,
      "loss": 1.4747,
      "step": 17530
    },
    {
      "epoch": 2.898574674653997,
      "grad_norm": 12.616483688354492,
      "learning_rate": 3.945169760737435e-05,
      "loss": 1.6141,
      "step": 17540
    },
    {
      "epoch": 2.9002272257797976,
      "grad_norm": 10.682483673095703,
      "learning_rate": 3.9442516388475735e-05,
      "loss": 1.526,
      "step": 17550
    },
    {
      "epoch": 2.901879776905598,
      "grad_norm": 12.046555519104004,
      "learning_rate": 3.943333516957711e-05,
      "loss": 1.5962,
      "step": 17560
    },
    {
      "epoch": 2.9035323280313987,
      "grad_norm": 9.173718452453613,
      "learning_rate": 3.942415395067849e-05,
      "loss": 1.5042,
      "step": 17570
    },
    {
      "epoch": 2.905184879157199,
      "grad_norm": 7.157196998596191,
      "learning_rate": 3.9414972731779876e-05,
      "loss": 1.5015,
      "step": 17580
    },
    {
      "epoch": 2.9068374302829993,
      "grad_norm": 9.088848114013672,
      "learning_rate": 3.940579151288125e-05,
      "loss": 1.5697,
      "step": 17590
    },
    {
      "epoch": 2.9084899814088,
      "grad_norm": 7.118719577789307,
      "learning_rate": 3.9396610293982634e-05,
      "loss": 1.5875,
      "step": 17600
    },
    {
      "epoch": 2.9101425325346004,
      "grad_norm": 4.978763103485107,
      "learning_rate": 3.938742907508401e-05,
      "loss": 1.4973,
      "step": 17610
    },
    {
      "epoch": 2.911795083660401,
      "grad_norm": 11.87973403930664,
      "learning_rate": 3.937824785618539e-05,
      "loss": 1.6336,
      "step": 17620
    },
    {
      "epoch": 2.913447634786201,
      "grad_norm": 23.558897018432617,
      "learning_rate": 3.936906663728677e-05,
      "loss": 1.4903,
      "step": 17630
    },
    {
      "epoch": 2.9151001859120016,
      "grad_norm": 8.35072135925293,
      "learning_rate": 3.935988541838815e-05,
      "loss": 1.5911,
      "step": 17640
    },
    {
      "epoch": 2.916752737037802,
      "grad_norm": 7.296316623687744,
      "learning_rate": 3.935070419948953e-05,
      "loss": 1.4861,
      "step": 17650
    },
    {
      "epoch": 2.9184052881636027,
      "grad_norm": 6.886360168457031,
      "learning_rate": 3.93415229805909e-05,
      "loss": 1.4605,
      "step": 17660
    },
    {
      "epoch": 2.920057839289403,
      "grad_norm": 9.66377067565918,
      "learning_rate": 3.933234176169228e-05,
      "loss": 1.4889,
      "step": 17670
    },
    {
      "epoch": 2.9217103904152033,
      "grad_norm": 12.230368614196777,
      "learning_rate": 3.932316054279366e-05,
      "loss": 1.6329,
      "step": 17680
    },
    {
      "epoch": 2.923362941541004,
      "grad_norm": 23.454057693481445,
      "learning_rate": 3.9313979323895044e-05,
      "loss": 1.5017,
      "step": 17690
    },
    {
      "epoch": 2.9250154926668044,
      "grad_norm": 10.6051025390625,
      "learning_rate": 3.930479810499642e-05,
      "loss": 1.6555,
      "step": 17700
    },
    {
      "epoch": 2.926668043792605,
      "grad_norm": 5.926885604858398,
      "learning_rate": 3.92956168860978e-05,
      "loss": 1.5747,
      "step": 17710
    },
    {
      "epoch": 2.9283205949184055,
      "grad_norm": 9.678142547607422,
      "learning_rate": 3.928643566719918e-05,
      "loss": 1.7058,
      "step": 17720
    },
    {
      "epoch": 2.9299731460442056,
      "grad_norm": 6.809134483337402,
      "learning_rate": 3.927725444830056e-05,
      "loss": 1.5812,
      "step": 17730
    },
    {
      "epoch": 2.931625697170006,
      "grad_norm": 25.03977394104004,
      "learning_rate": 3.926807322940194e-05,
      "loss": 1.5576,
      "step": 17740
    },
    {
      "epoch": 2.9332782482958066,
      "grad_norm": 6.897164344787598,
      "learning_rate": 3.925889201050332e-05,
      "loss": 1.5183,
      "step": 17750
    },
    {
      "epoch": 2.934930799421607,
      "grad_norm": 6.26050329208374,
      "learning_rate": 3.9249710791604695e-05,
      "loss": 1.7254,
      "step": 17760
    },
    {
      "epoch": 2.9365833505474077,
      "grad_norm": 7.9940385818481445,
      "learning_rate": 3.924052957270608e-05,
      "loss": 1.5969,
      "step": 17770
    },
    {
      "epoch": 2.938235901673208,
      "grad_norm": 13.388635635375977,
      "learning_rate": 3.9231348353807454e-05,
      "loss": 1.5729,
      "step": 17780
    },
    {
      "epoch": 2.9398884527990083,
      "grad_norm": 5.230599880218506,
      "learning_rate": 3.922216713490883e-05,
      "loss": 1.6222,
      "step": 17790
    },
    {
      "epoch": 2.941541003924809,
      "grad_norm": 13.589559555053711,
      "learning_rate": 3.9212985916010206e-05,
      "loss": 1.6322,
      "step": 17800
    },
    {
      "epoch": 2.9431935550506094,
      "grad_norm": 22.378572463989258,
      "learning_rate": 3.920380469711159e-05,
      "loss": 1.6475,
      "step": 17810
    },
    {
      "epoch": 2.94484610617641,
      "grad_norm": 6.242623805999756,
      "learning_rate": 3.919462347821297e-05,
      "loss": 1.6062,
      "step": 17820
    },
    {
      "epoch": 2.94649865730221,
      "grad_norm": 14.694461822509766,
      "learning_rate": 3.918544225931435e-05,
      "loss": 1.6167,
      "step": 17830
    },
    {
      "epoch": 2.9481512084280106,
      "grad_norm": 6.985626697540283,
      "learning_rate": 3.917626104041573e-05,
      "loss": 1.7133,
      "step": 17840
    },
    {
      "epoch": 2.949803759553811,
      "grad_norm": 30.49583625793457,
      "learning_rate": 3.9167079821517105e-05,
      "loss": 1.6603,
      "step": 17850
    },
    {
      "epoch": 2.9514563106796117,
      "grad_norm": 16.23304557800293,
      "learning_rate": 3.915789860261849e-05,
      "loss": 1.5594,
      "step": 17860
    },
    {
      "epoch": 2.953108861805412,
      "grad_norm": 8.873785972595215,
      "learning_rate": 3.9148717383719864e-05,
      "loss": 1.6136,
      "step": 17870
    },
    {
      "epoch": 2.9547614129312123,
      "grad_norm": 10.205568313598633,
      "learning_rate": 3.9139536164821246e-05,
      "loss": 1.5299,
      "step": 17880
    },
    {
      "epoch": 2.9564139640570133,
      "grad_norm": 7.908629894256592,
      "learning_rate": 3.913035494592262e-05,
      "loss": 1.5855,
      "step": 17890
    },
    {
      "epoch": 2.9580665151828134,
      "grad_norm": 14.08654499053955,
      "learning_rate": 3.9121173727024005e-05,
      "loss": 1.6594,
      "step": 17900
    },
    {
      "epoch": 2.959719066308614,
      "grad_norm": 6.650778293609619,
      "learning_rate": 3.911199250812538e-05,
      "loss": 1.5049,
      "step": 17910
    },
    {
      "epoch": 2.9613716174344145,
      "grad_norm": 14.497530937194824,
      "learning_rate": 3.910281128922676e-05,
      "loss": 1.5831,
      "step": 17920
    },
    {
      "epoch": 2.9630241685602146,
      "grad_norm": 9.797993659973145,
      "learning_rate": 3.909363007032814e-05,
      "loss": 1.6278,
      "step": 17930
    },
    {
      "epoch": 2.9646767196860155,
      "grad_norm": 8.208698272705078,
      "learning_rate": 3.9084448851429515e-05,
      "loss": 1.7192,
      "step": 17940
    },
    {
      "epoch": 2.9663292708118156,
      "grad_norm": 27.290040969848633,
      "learning_rate": 3.90752676325309e-05,
      "loss": 1.5312,
      "step": 17950
    },
    {
      "epoch": 2.967981821937616,
      "grad_norm": 7.827337741851807,
      "learning_rate": 3.9066086413632274e-05,
      "loss": 1.4619,
      "step": 17960
    },
    {
      "epoch": 2.9696343730634167,
      "grad_norm": 9.8374605178833,
      "learning_rate": 3.9056905194733656e-05,
      "loss": 1.6841,
      "step": 17970
    },
    {
      "epoch": 2.9712869241892172,
      "grad_norm": 6.857816219329834,
      "learning_rate": 3.904772397583503e-05,
      "loss": 1.6227,
      "step": 17980
    },
    {
      "epoch": 2.972939475315018,
      "grad_norm": 12.391343116760254,
      "learning_rate": 3.9038542756936415e-05,
      "loss": 1.6735,
      "step": 17990
    },
    {
      "epoch": 2.974592026440818,
      "grad_norm": 9.983665466308594,
      "learning_rate": 3.902936153803779e-05,
      "loss": 1.5625,
      "step": 18000
    },
    {
      "epoch": 2.9762445775666184,
      "grad_norm": 6.569074630737305,
      "learning_rate": 3.902018031913917e-05,
      "loss": 1.408,
      "step": 18010
    },
    {
      "epoch": 2.977897128692419,
      "grad_norm": 10.561470985412598,
      "learning_rate": 3.901099910024055e-05,
      "loss": 1.5342,
      "step": 18020
    },
    {
      "epoch": 2.9795496798182195,
      "grad_norm": 7.086844444274902,
      "learning_rate": 3.900181788134193e-05,
      "loss": 1.7286,
      "step": 18030
    },
    {
      "epoch": 2.98120223094402,
      "grad_norm": 15.396596908569336,
      "learning_rate": 3.899263666244331e-05,
      "loss": 1.4908,
      "step": 18040
    },
    {
      "epoch": 2.98285478206982,
      "grad_norm": 16.880029678344727,
      "learning_rate": 3.8983455443544683e-05,
      "loss": 1.623,
      "step": 18050
    },
    {
      "epoch": 2.9845073331956207,
      "grad_norm": 35.76949691772461,
      "learning_rate": 3.8974274224646066e-05,
      "loss": 1.6083,
      "step": 18060
    },
    {
      "epoch": 2.986159884321421,
      "grad_norm": 11.887407302856445,
      "learning_rate": 3.896509300574744e-05,
      "loss": 1.5319,
      "step": 18070
    },
    {
      "epoch": 2.9878124354472217,
      "grad_norm": 15.86429500579834,
      "learning_rate": 3.8955911786848825e-05,
      "loss": 1.6974,
      "step": 18080
    },
    {
      "epoch": 2.9894649865730223,
      "grad_norm": 45.645751953125,
      "learning_rate": 3.89467305679502e-05,
      "loss": 1.5547,
      "step": 18090
    },
    {
      "epoch": 2.9911175376988224,
      "grad_norm": 4.425729274749756,
      "learning_rate": 3.893754934905158e-05,
      "loss": 1.6069,
      "step": 18100
    },
    {
      "epoch": 2.992770088824623,
      "grad_norm": 14.894420623779297,
      "learning_rate": 3.892836813015296e-05,
      "loss": 1.4857,
      "step": 18110
    },
    {
      "epoch": 2.9944226399504235,
      "grad_norm": 12.32193660736084,
      "learning_rate": 3.891918691125434e-05,
      "loss": 1.7373,
      "step": 18120
    },
    {
      "epoch": 2.996075191076224,
      "grad_norm": 8.957423210144043,
      "learning_rate": 3.891000569235572e-05,
      "loss": 1.5757,
      "step": 18130
    },
    {
      "epoch": 2.9977277422020245,
      "grad_norm": 12.539956092834473,
      "learning_rate": 3.89008244734571e-05,
      "loss": 1.6558,
      "step": 18140
    },
    {
      "epoch": 2.9993802933278246,
      "grad_norm": 18.08677864074707,
      "learning_rate": 3.889164325455848e-05,
      "loss": 1.6866,
      "step": 18150
    },
    {
      "epoch": 2.999876058665565,
      "eval_accuracy": 0.2888352957866364,
      "eval_loss": 2.114047050476074,
      "eval_runtime": 817.3743,
      "eval_samples_per_second": 34.496,
      "eval_steps_per_second": 8.624,
      "step": 18153
    },
    {
      "epoch": 3.001032844453625,
      "grad_norm": 7.226766586303711,
      "learning_rate": 3.888246203565986e-05,
      "loss": 1.481,
      "step": 18160
    },
    {
      "epoch": 3.0026853955794257,
      "grad_norm": 6.795170307159424,
      "learning_rate": 3.8873280816761234e-05,
      "loss": 1.5827,
      "step": 18170
    },
    {
      "epoch": 3.0043379467052262,
      "grad_norm": 9.203761100769043,
      "learning_rate": 3.886409959786261e-05,
      "loss": 1.5858,
      "step": 18180
    },
    {
      "epoch": 3.005990497831027,
      "grad_norm": 8.978983879089355,
      "learning_rate": 3.885491837896399e-05,
      "loss": 1.6911,
      "step": 18190
    },
    {
      "epoch": 3.007643048956827,
      "grad_norm": 6.757550239562988,
      "learning_rate": 3.884573716006537e-05,
      "loss": 1.4933,
      "step": 18200
    },
    {
      "epoch": 3.0092956000826274,
      "grad_norm": 6.979025363922119,
      "learning_rate": 3.883655594116675e-05,
      "loss": 1.4942,
      "step": 18210
    },
    {
      "epoch": 3.010948151208428,
      "grad_norm": 8.403514862060547,
      "learning_rate": 3.882737472226813e-05,
      "loss": 1.5651,
      "step": 18220
    },
    {
      "epoch": 3.0126007023342285,
      "grad_norm": 16.475433349609375,
      "learning_rate": 3.881819350336951e-05,
      "loss": 1.5098,
      "step": 18230
    },
    {
      "epoch": 3.014253253460029,
      "grad_norm": 10.676935195922852,
      "learning_rate": 3.8809012284470886e-05,
      "loss": 1.5301,
      "step": 18240
    },
    {
      "epoch": 3.0159058045858296,
      "grad_norm": 80.31954193115234,
      "learning_rate": 3.879983106557227e-05,
      "loss": 1.5747,
      "step": 18250
    },
    {
      "epoch": 3.0175583557116297,
      "grad_norm": 9.030322074890137,
      "learning_rate": 3.879064984667365e-05,
      "loss": 1.6205,
      "step": 18260
    },
    {
      "epoch": 3.01921090683743,
      "grad_norm": 9.721772193908691,
      "learning_rate": 3.878146862777503e-05,
      "loss": 1.4312,
      "step": 18270
    },
    {
      "epoch": 3.0208634579632307,
      "grad_norm": 8.545890808105469,
      "learning_rate": 3.877228740887641e-05,
      "loss": 1.6293,
      "step": 18280
    },
    {
      "epoch": 3.0225160090890313,
      "grad_norm": 15.735355377197266,
      "learning_rate": 3.8763106189977785e-05,
      "loss": 1.5873,
      "step": 18290
    },
    {
      "epoch": 3.024168560214832,
      "grad_norm": 7.491411209106445,
      "learning_rate": 3.875392497107916e-05,
      "loss": 1.5171,
      "step": 18300
    },
    {
      "epoch": 3.025821111340632,
      "grad_norm": 8.309893608093262,
      "learning_rate": 3.874474375218054e-05,
      "loss": 1.4795,
      "step": 18310
    },
    {
      "epoch": 3.0274736624664325,
      "grad_norm": 19.71273422241211,
      "learning_rate": 3.873556253328192e-05,
      "loss": 1.5505,
      "step": 18320
    },
    {
      "epoch": 3.029126213592233,
      "grad_norm": 9.179100036621094,
      "learning_rate": 3.8726381314383296e-05,
      "loss": 1.6547,
      "step": 18330
    },
    {
      "epoch": 3.0307787647180335,
      "grad_norm": 6.885425090789795,
      "learning_rate": 3.871720009548468e-05,
      "loss": 1.5285,
      "step": 18340
    },
    {
      "epoch": 3.032431315843834,
      "grad_norm": 7.15316104888916,
      "learning_rate": 3.8708018876586054e-05,
      "loss": 1.4998,
      "step": 18350
    },
    {
      "epoch": 3.034083866969634,
      "grad_norm": 7.689057350158691,
      "learning_rate": 3.869883765768744e-05,
      "loss": 1.5134,
      "step": 18360
    },
    {
      "epoch": 3.0357364180954347,
      "grad_norm": 6.1142449378967285,
      "learning_rate": 3.868965643878881e-05,
      "loss": 1.5328,
      "step": 18370
    },
    {
      "epoch": 3.0373889692212352,
      "grad_norm": 19.263931274414062,
      "learning_rate": 3.8680475219890195e-05,
      "loss": 1.5652,
      "step": 18380
    },
    {
      "epoch": 3.039041520347036,
      "grad_norm": 17.659536361694336,
      "learning_rate": 3.867129400099158e-05,
      "loss": 1.6269,
      "step": 18390
    },
    {
      "epoch": 3.0406940714728363,
      "grad_norm": 8.499923706054688,
      "learning_rate": 3.8662112782092954e-05,
      "loss": 1.5685,
      "step": 18400
    },
    {
      "epoch": 3.042346622598637,
      "grad_norm": 7.187222480773926,
      "learning_rate": 3.8652931563194336e-05,
      "loss": 1.6992,
      "step": 18410
    },
    {
      "epoch": 3.043999173724437,
      "grad_norm": 5.891327381134033,
      "learning_rate": 3.864375034429571e-05,
      "loss": 1.4802,
      "step": 18420
    },
    {
      "epoch": 3.0456517248502375,
      "grad_norm": 17.82583999633789,
      "learning_rate": 3.863456912539709e-05,
      "loss": 1.6052,
      "step": 18430
    },
    {
      "epoch": 3.047304275976038,
      "grad_norm": 19.411766052246094,
      "learning_rate": 3.8625387906498464e-05,
      "loss": 1.4634,
      "step": 18440
    },
    {
      "epoch": 3.0489568271018386,
      "grad_norm": 9.680952072143555,
      "learning_rate": 3.861620668759985e-05,
      "loss": 1.6466,
      "step": 18450
    },
    {
      "epoch": 3.050609378227639,
      "grad_norm": 7.9444990158081055,
      "learning_rate": 3.860702546870122e-05,
      "loss": 1.5706,
      "step": 18460
    },
    {
      "epoch": 3.052261929353439,
      "grad_norm": 7.4515814781188965,
      "learning_rate": 3.8597844249802605e-05,
      "loss": 1.6711,
      "step": 18470
    },
    {
      "epoch": 3.0539144804792397,
      "grad_norm": 14.548356056213379,
      "learning_rate": 3.858866303090398e-05,
      "loss": 1.6145,
      "step": 18480
    },
    {
      "epoch": 3.0555670316050403,
      "grad_norm": 24.74275016784668,
      "learning_rate": 3.8579481812005364e-05,
      "loss": 1.542,
      "step": 18490
    },
    {
      "epoch": 3.057219582730841,
      "grad_norm": 5.011672496795654,
      "learning_rate": 3.8570300593106746e-05,
      "loss": 1.5655,
      "step": 18500
    },
    {
      "epoch": 3.0588721338566414,
      "grad_norm": 9.48055648803711,
      "learning_rate": 3.856111937420812e-05,
      "loss": 1.5704,
      "step": 18510
    },
    {
      "epoch": 3.0605246849824415,
      "grad_norm": 17.483966827392578,
      "learning_rate": 3.8551938155309505e-05,
      "loss": 1.5234,
      "step": 18520
    },
    {
      "epoch": 3.062177236108242,
      "grad_norm": 6.578785419464111,
      "learning_rate": 3.854275693641088e-05,
      "loss": 1.4352,
      "step": 18530
    },
    {
      "epoch": 3.0638297872340425,
      "grad_norm": 9.402719497680664,
      "learning_rate": 3.853357571751226e-05,
      "loss": 1.6183,
      "step": 18540
    },
    {
      "epoch": 3.065482338359843,
      "grad_norm": 17.517202377319336,
      "learning_rate": 3.852439449861364e-05,
      "loss": 1.6068,
      "step": 18550
    },
    {
      "epoch": 3.0671348894856436,
      "grad_norm": 8.648771286010742,
      "learning_rate": 3.8515213279715015e-05,
      "loss": 1.6782,
      "step": 18560
    },
    {
      "epoch": 3.0687874406114437,
      "grad_norm": 7.555728912353516,
      "learning_rate": 3.850603206081639e-05,
      "loss": 1.5315,
      "step": 18570
    },
    {
      "epoch": 3.0704399917372442,
      "grad_norm": 43.67970657348633,
      "learning_rate": 3.8496850841917773e-05,
      "loss": 1.5716,
      "step": 18580
    },
    {
      "epoch": 3.072092542863045,
      "grad_norm": 11.917938232421875,
      "learning_rate": 3.848766962301915e-05,
      "loss": 1.4323,
      "step": 18590
    },
    {
      "epoch": 3.0737450939888453,
      "grad_norm": 6.986249923706055,
      "learning_rate": 3.847848840412053e-05,
      "loss": 1.4702,
      "step": 18600
    },
    {
      "epoch": 3.075397645114646,
      "grad_norm": 12.20637035369873,
      "learning_rate": 3.8469307185221915e-05,
      "loss": 1.568,
      "step": 18610
    },
    {
      "epoch": 3.0770501962404464,
      "grad_norm": 29.53895378112793,
      "learning_rate": 3.846012596632329e-05,
      "loss": 1.5619,
      "step": 18620
    },
    {
      "epoch": 3.0787027473662465,
      "grad_norm": 19.85051918029785,
      "learning_rate": 3.845094474742467e-05,
      "loss": 1.6694,
      "step": 18630
    },
    {
      "epoch": 3.080355298492047,
      "grad_norm": 6.126530170440674,
      "learning_rate": 3.844176352852605e-05,
      "loss": 1.5997,
      "step": 18640
    },
    {
      "epoch": 3.0820078496178476,
      "grad_norm": 6.928379058837891,
      "learning_rate": 3.843258230962743e-05,
      "loss": 1.6335,
      "step": 18650
    },
    {
      "epoch": 3.083660400743648,
      "grad_norm": 6.429804801940918,
      "learning_rate": 3.842340109072881e-05,
      "loss": 1.6117,
      "step": 18660
    },
    {
      "epoch": 3.0853129518694486,
      "grad_norm": 6.85563850402832,
      "learning_rate": 3.841421987183019e-05,
      "loss": 1.5142,
      "step": 18670
    },
    {
      "epoch": 3.0869655029952487,
      "grad_norm": 8.979350090026855,
      "learning_rate": 3.8405038652931566e-05,
      "loss": 1.5052,
      "step": 18680
    },
    {
      "epoch": 3.0886180541210493,
      "grad_norm": 8.090424537658691,
      "learning_rate": 3.839585743403294e-05,
      "loss": 1.5765,
      "step": 18690
    },
    {
      "epoch": 3.09027060524685,
      "grad_norm": 12.26007080078125,
      "learning_rate": 3.838667621513432e-05,
      "loss": 1.6073,
      "step": 18700
    },
    {
      "epoch": 3.0919231563726504,
      "grad_norm": 13.63984489440918,
      "learning_rate": 3.83774949962357e-05,
      "loss": 1.6378,
      "step": 18710
    },
    {
      "epoch": 3.093575707498451,
      "grad_norm": 35.192535400390625,
      "learning_rate": 3.836831377733708e-05,
      "loss": 1.7231,
      "step": 18720
    },
    {
      "epoch": 3.095228258624251,
      "grad_norm": 10.192204475402832,
      "learning_rate": 3.835913255843846e-05,
      "loss": 1.5279,
      "step": 18730
    },
    {
      "epoch": 3.0968808097500515,
      "grad_norm": 5.712940216064453,
      "learning_rate": 3.834995133953984e-05,
      "loss": 1.5624,
      "step": 18740
    },
    {
      "epoch": 3.098533360875852,
      "grad_norm": 7.214580535888672,
      "learning_rate": 3.834077012064122e-05,
      "loss": 1.672,
      "step": 18750
    },
    {
      "epoch": 3.1001859120016526,
      "grad_norm": 19.402013778686523,
      "learning_rate": 3.83315889017426e-05,
      "loss": 1.6678,
      "step": 18760
    },
    {
      "epoch": 3.101838463127453,
      "grad_norm": 7.78207540512085,
      "learning_rate": 3.8322407682843976e-05,
      "loss": 1.6826,
      "step": 18770
    },
    {
      "epoch": 3.1034910142532537,
      "grad_norm": 6.110933303833008,
      "learning_rate": 3.831322646394536e-05,
      "loss": 1.5121,
      "step": 18780
    },
    {
      "epoch": 3.105143565379054,
      "grad_norm": 5.636849403381348,
      "learning_rate": 3.8304045245046734e-05,
      "loss": 1.4881,
      "step": 18790
    },
    {
      "epoch": 3.1067961165048543,
      "grad_norm": 8.485923767089844,
      "learning_rate": 3.829486402614812e-05,
      "loss": 1.6185,
      "step": 18800
    },
    {
      "epoch": 3.108448667630655,
      "grad_norm": 7.813500881195068,
      "learning_rate": 3.828568280724949e-05,
      "loss": 1.645,
      "step": 18810
    },
    {
      "epoch": 3.1101012187564554,
      "grad_norm": 8.218082427978516,
      "learning_rate": 3.827650158835087e-05,
      "loss": 1.601,
      "step": 18820
    },
    {
      "epoch": 3.111753769882256,
      "grad_norm": 11.229850769042969,
      "learning_rate": 3.826732036945225e-05,
      "loss": 1.5679,
      "step": 18830
    },
    {
      "epoch": 3.113406321008056,
      "grad_norm": 8.275188446044922,
      "learning_rate": 3.825813915055363e-05,
      "loss": 1.5716,
      "step": 18840
    },
    {
      "epoch": 3.1150588721338566,
      "grad_norm": 7.776087284088135,
      "learning_rate": 3.824895793165501e-05,
      "loss": 1.4384,
      "step": 18850
    },
    {
      "epoch": 3.116711423259657,
      "grad_norm": 37.47963333129883,
      "learning_rate": 3.8239776712756386e-05,
      "loss": 1.6417,
      "step": 18860
    },
    {
      "epoch": 3.1183639743854576,
      "grad_norm": 8.274860382080078,
      "learning_rate": 3.823059549385777e-05,
      "loss": 1.5703,
      "step": 18870
    },
    {
      "epoch": 3.120016525511258,
      "grad_norm": 8.022534370422363,
      "learning_rate": 3.8221414274959144e-05,
      "loss": 1.5707,
      "step": 18880
    },
    {
      "epoch": 3.1216690766370583,
      "grad_norm": 5.824304103851318,
      "learning_rate": 3.821223305606053e-05,
      "loss": 1.4785,
      "step": 18890
    },
    {
      "epoch": 3.123321627762859,
      "grad_norm": 9.422294616699219,
      "learning_rate": 3.82030518371619e-05,
      "loss": 1.5626,
      "step": 18900
    },
    {
      "epoch": 3.1249741788886594,
      "grad_norm": 6.564004898071289,
      "learning_rate": 3.8193870618263285e-05,
      "loss": 1.6056,
      "step": 18910
    },
    {
      "epoch": 3.12662673001446,
      "grad_norm": 10.041267395019531,
      "learning_rate": 3.818468939936466e-05,
      "loss": 1.4272,
      "step": 18920
    },
    {
      "epoch": 3.1282792811402604,
      "grad_norm": 36.88932800292969,
      "learning_rate": 3.8175508180466044e-05,
      "loss": 1.3849,
      "step": 18930
    },
    {
      "epoch": 3.1299318322660605,
      "grad_norm": 13.796693801879883,
      "learning_rate": 3.816632696156742e-05,
      "loss": 1.5946,
      "step": 18940
    },
    {
      "epoch": 3.131584383391861,
      "grad_norm": 14.519100189208984,
      "learning_rate": 3.8157145742668795e-05,
      "loss": 1.532,
      "step": 18950
    },
    {
      "epoch": 3.1332369345176616,
      "grad_norm": 9.376446723937988,
      "learning_rate": 3.814796452377018e-05,
      "loss": 1.683,
      "step": 18960
    },
    {
      "epoch": 3.134889485643462,
      "grad_norm": 9.751054763793945,
      "learning_rate": 3.8138783304871554e-05,
      "loss": 1.5191,
      "step": 18970
    },
    {
      "epoch": 3.1365420367692627,
      "grad_norm": 11.590496063232422,
      "learning_rate": 3.8129602085972937e-05,
      "loss": 1.4696,
      "step": 18980
    },
    {
      "epoch": 3.138194587895063,
      "grad_norm": 18.807069778442383,
      "learning_rate": 3.812042086707431e-05,
      "loss": 1.5602,
      "step": 18990
    },
    {
      "epoch": 3.1398471390208633,
      "grad_norm": 6.233638286590576,
      "learning_rate": 3.8111239648175695e-05,
      "loss": 1.5646,
      "step": 19000
    },
    {
      "epoch": 3.141499690146664,
      "grad_norm": 6.848485469818115,
      "learning_rate": 3.810205842927707e-05,
      "loss": 1.5728,
      "step": 19010
    },
    {
      "epoch": 3.1431522412724644,
      "grad_norm": 18.452594757080078,
      "learning_rate": 3.8092877210378454e-05,
      "loss": 1.6127,
      "step": 19020
    },
    {
      "epoch": 3.144804792398265,
      "grad_norm": 59.3503303527832,
      "learning_rate": 3.808369599147983e-05,
      "loss": 1.6197,
      "step": 19030
    },
    {
      "epoch": 3.1464573435240655,
      "grad_norm": 11.30215835571289,
      "learning_rate": 3.807451477258121e-05,
      "loss": 1.5164,
      "step": 19040
    },
    {
      "epoch": 3.1481098946498656,
      "grad_norm": 8.094250679016113,
      "learning_rate": 3.806533355368259e-05,
      "loss": 1.4852,
      "step": 19050
    },
    {
      "epoch": 3.149762445775666,
      "grad_norm": 8.159111022949219,
      "learning_rate": 3.805615233478397e-05,
      "loss": 1.6521,
      "step": 19060
    },
    {
      "epoch": 3.1514149969014666,
      "grad_norm": 27.37004280090332,
      "learning_rate": 3.8046971115885346e-05,
      "loss": 1.4812,
      "step": 19070
    },
    {
      "epoch": 3.153067548027267,
      "grad_norm": 9.601912498474121,
      "learning_rate": 3.803778989698672e-05,
      "loss": 1.6315,
      "step": 19080
    },
    {
      "epoch": 3.1547200991530677,
      "grad_norm": 10.574562072753906,
      "learning_rate": 3.8028608678088105e-05,
      "loss": 1.5586,
      "step": 19090
    },
    {
      "epoch": 3.156372650278868,
      "grad_norm": 5.611733913421631,
      "learning_rate": 3.801942745918948e-05,
      "loss": 1.5879,
      "step": 19100
    },
    {
      "epoch": 3.1580252014046684,
      "grad_norm": 24.429452896118164,
      "learning_rate": 3.8010246240290863e-05,
      "loss": 1.5259,
      "step": 19110
    },
    {
      "epoch": 3.159677752530469,
      "grad_norm": 6.225391864776611,
      "learning_rate": 3.800106502139224e-05,
      "loss": 1.5108,
      "step": 19120
    },
    {
      "epoch": 3.1613303036562694,
      "grad_norm": 8.502754211425781,
      "learning_rate": 3.799188380249362e-05,
      "loss": 1.5145,
      "step": 19130
    },
    {
      "epoch": 3.16298285478207,
      "grad_norm": 17.88811683654785,
      "learning_rate": 3.7982702583595e-05,
      "loss": 1.6389,
      "step": 19140
    },
    {
      "epoch": 3.16463540590787,
      "grad_norm": 13.118366241455078,
      "learning_rate": 3.797352136469638e-05,
      "loss": 1.5567,
      "step": 19150
    },
    {
      "epoch": 3.1662879570336706,
      "grad_norm": 8.860665321350098,
      "learning_rate": 3.7964340145797756e-05,
      "loss": 1.6773,
      "step": 19160
    },
    {
      "epoch": 3.167940508159471,
      "grad_norm": 6.436194896697998,
      "learning_rate": 3.795515892689914e-05,
      "loss": 1.4645,
      "step": 19170
    },
    {
      "epoch": 3.1695930592852717,
      "grad_norm": 11.198339462280273,
      "learning_rate": 3.794597770800052e-05,
      "loss": 1.5443,
      "step": 19180
    },
    {
      "epoch": 3.1712456104110722,
      "grad_norm": 15.714473724365234,
      "learning_rate": 3.79367964891019e-05,
      "loss": 1.5452,
      "step": 19190
    },
    {
      "epoch": 3.1728981615368728,
      "grad_norm": 17.801097869873047,
      "learning_rate": 3.792761527020327e-05,
      "loss": 1.5366,
      "step": 19200
    },
    {
      "epoch": 3.174550712662673,
      "grad_norm": 11.515276908874512,
      "learning_rate": 3.791843405130465e-05,
      "loss": 1.6625,
      "step": 19210
    },
    {
      "epoch": 3.1762032637884734,
      "grad_norm": 6.000487804412842,
      "learning_rate": 3.790925283240603e-05,
      "loss": 1.4931,
      "step": 19220
    },
    {
      "epoch": 3.177855814914274,
      "grad_norm": 8.129054069519043,
      "learning_rate": 3.790007161350741e-05,
      "loss": 1.7251,
      "step": 19230
    },
    {
      "epoch": 3.1795083660400745,
      "grad_norm": 14.319493293762207,
      "learning_rate": 3.789089039460879e-05,
      "loss": 1.6495,
      "step": 19240
    },
    {
      "epoch": 3.181160917165875,
      "grad_norm": 30.3303165435791,
      "learning_rate": 3.7881709175710166e-05,
      "loss": 1.4484,
      "step": 19250
    },
    {
      "epoch": 3.182813468291675,
      "grad_norm": 10.565496444702148,
      "learning_rate": 3.787252795681155e-05,
      "loss": 1.6871,
      "step": 19260
    },
    {
      "epoch": 3.1844660194174756,
      "grad_norm": 9.682084083557129,
      "learning_rate": 3.7863346737912925e-05,
      "loss": 1.6799,
      "step": 19270
    },
    {
      "epoch": 3.186118570543276,
      "grad_norm": 6.963440418243408,
      "learning_rate": 3.785416551901431e-05,
      "loss": 1.6251,
      "step": 19280
    },
    {
      "epoch": 3.1877711216690767,
      "grad_norm": 20.987451553344727,
      "learning_rate": 3.784498430011569e-05,
      "loss": 1.4718,
      "step": 19290
    },
    {
      "epoch": 3.1894236727948773,
      "grad_norm": 12.118252754211426,
      "learning_rate": 3.7835803081217066e-05,
      "loss": 1.5291,
      "step": 19300
    },
    {
      "epoch": 3.1910762239206774,
      "grad_norm": 8.367793083190918,
      "learning_rate": 3.782662186231845e-05,
      "loss": 1.5914,
      "step": 19310
    },
    {
      "epoch": 3.192728775046478,
      "grad_norm": 7.3973164558410645,
      "learning_rate": 3.7817440643419824e-05,
      "loss": 1.4778,
      "step": 19320
    },
    {
      "epoch": 3.1943813261722784,
      "grad_norm": 28.844614028930664,
      "learning_rate": 3.78082594245212e-05,
      "loss": 1.4982,
      "step": 19330
    },
    {
      "epoch": 3.196033877298079,
      "grad_norm": 14.039192199707031,
      "learning_rate": 3.7799078205622576e-05,
      "loss": 1.5076,
      "step": 19340
    },
    {
      "epoch": 3.1976864284238795,
      "grad_norm": 6.318631172180176,
      "learning_rate": 3.778989698672396e-05,
      "loss": 1.5983,
      "step": 19350
    },
    {
      "epoch": 3.1993389795496796,
      "grad_norm": 11.539376258850098,
      "learning_rate": 3.7780715767825334e-05,
      "loss": 1.6026,
      "step": 19360
    },
    {
      "epoch": 3.20099153067548,
      "grad_norm": 15.837471961975098,
      "learning_rate": 3.777153454892672e-05,
      "loss": 1.6435,
      "step": 19370
    },
    {
      "epoch": 3.2026440818012807,
      "grad_norm": 10.379616737365723,
      "learning_rate": 3.776235333002809e-05,
      "loss": 1.6636,
      "step": 19380
    },
    {
      "epoch": 3.2042966329270812,
      "grad_norm": 18.38787841796875,
      "learning_rate": 3.7753172111129476e-05,
      "loss": 1.5559,
      "step": 19390
    },
    {
      "epoch": 3.2059491840528818,
      "grad_norm": 22.003950119018555,
      "learning_rate": 3.774399089223085e-05,
      "loss": 1.5326,
      "step": 19400
    },
    {
      "epoch": 3.207601735178682,
      "grad_norm": 6.592588424682617,
      "learning_rate": 3.7734809673332234e-05,
      "loss": 1.4978,
      "step": 19410
    },
    {
      "epoch": 3.2092542863044824,
      "grad_norm": 23.952259063720703,
      "learning_rate": 3.772562845443362e-05,
      "loss": 1.5836,
      "step": 19420
    },
    {
      "epoch": 3.210906837430283,
      "grad_norm": 16.6333065032959,
      "learning_rate": 3.771644723553499e-05,
      "loss": 1.61,
      "step": 19430
    },
    {
      "epoch": 3.2125593885560835,
      "grad_norm": 11.940427780151367,
      "learning_rate": 3.7707266016636375e-05,
      "loss": 1.4965,
      "step": 19440
    },
    {
      "epoch": 3.214211939681884,
      "grad_norm": 49.953617095947266,
      "learning_rate": 3.769808479773775e-05,
      "loss": 1.3932,
      "step": 19450
    },
    {
      "epoch": 3.2158644908076846,
      "grad_norm": 11.84276008605957,
      "learning_rate": 3.768890357883913e-05,
      "loss": 1.5016,
      "step": 19460
    },
    {
      "epoch": 3.2175170419334846,
      "grad_norm": 17.914264678955078,
      "learning_rate": 3.76797223599405e-05,
      "loss": 1.5665,
      "step": 19470
    },
    {
      "epoch": 3.219169593059285,
      "grad_norm": 8.291590690612793,
      "learning_rate": 3.7670541141041885e-05,
      "loss": 1.5544,
      "step": 19480
    },
    {
      "epoch": 3.2208221441850857,
      "grad_norm": 11.181132316589355,
      "learning_rate": 3.766135992214326e-05,
      "loss": 1.6092,
      "step": 19490
    },
    {
      "epoch": 3.2224746953108863,
      "grad_norm": 7.685586452484131,
      "learning_rate": 3.7652178703244644e-05,
      "loss": 1.6567,
      "step": 19500
    },
    {
      "epoch": 3.224127246436687,
      "grad_norm": 9.388480186462402,
      "learning_rate": 3.764299748434602e-05,
      "loss": 1.7372,
      "step": 19510
    },
    {
      "epoch": 3.225779797562487,
      "grad_norm": 6.502718448638916,
      "learning_rate": 3.76338162654474e-05,
      "loss": 1.5202,
      "step": 19520
    },
    {
      "epoch": 3.2274323486882874,
      "grad_norm": 19.254741668701172,
      "learning_rate": 3.7624635046548785e-05,
      "loss": 1.5534,
      "step": 19530
    },
    {
      "epoch": 3.229084899814088,
      "grad_norm": 12.590047836303711,
      "learning_rate": 3.761545382765016e-05,
      "loss": 1.4716,
      "step": 19540
    },
    {
      "epoch": 3.2307374509398885,
      "grad_norm": 58.66154479980469,
      "learning_rate": 3.7606272608751544e-05,
      "loss": 1.5091,
      "step": 19550
    },
    {
      "epoch": 3.232390002065689,
      "grad_norm": 9.476922035217285,
      "learning_rate": 3.759709138985292e-05,
      "loss": 1.5869,
      "step": 19560
    },
    {
      "epoch": 3.2340425531914896,
      "grad_norm": 62.604740142822266,
      "learning_rate": 3.75879101709543e-05,
      "loss": 1.5135,
      "step": 19570
    },
    {
      "epoch": 3.2356951043172897,
      "grad_norm": 7.957396507263184,
      "learning_rate": 3.757872895205568e-05,
      "loss": 1.5217,
      "step": 19580
    },
    {
      "epoch": 3.23734765544309,
      "grad_norm": 6.035177230834961,
      "learning_rate": 3.7569547733157054e-05,
      "loss": 1.4088,
      "step": 19590
    },
    {
      "epoch": 3.2390002065688908,
      "grad_norm": 33.28099822998047,
      "learning_rate": 3.756036651425843e-05,
      "loss": 1.5084,
      "step": 19600
    },
    {
      "epoch": 3.2406527576946913,
      "grad_norm": 10.489121437072754,
      "learning_rate": 3.755118529535981e-05,
      "loss": 1.5133,
      "step": 19610
    },
    {
      "epoch": 3.242305308820492,
      "grad_norm": 9.935800552368164,
      "learning_rate": 3.754200407646119e-05,
      "loss": 1.5428,
      "step": 19620
    },
    {
      "epoch": 3.243957859946292,
      "grad_norm": 8.586844444274902,
      "learning_rate": 3.753282285756257e-05,
      "loss": 1.4201,
      "step": 19630
    },
    {
      "epoch": 3.2456104110720925,
      "grad_norm": 24.38202476501465,
      "learning_rate": 3.752364163866395e-05,
      "loss": 1.5493,
      "step": 19640
    },
    {
      "epoch": 3.247262962197893,
      "grad_norm": 9.439826965332031,
      "learning_rate": 3.751446041976533e-05,
      "loss": 1.5399,
      "step": 19650
    },
    {
      "epoch": 3.2489155133236935,
      "grad_norm": 12.375255584716797,
      "learning_rate": 3.750527920086671e-05,
      "loss": 1.511,
      "step": 19660
    },
    {
      "epoch": 3.250568064449494,
      "grad_norm": 9.748537063598633,
      "learning_rate": 3.749609798196809e-05,
      "loss": 1.5526,
      "step": 19670
    },
    {
      "epoch": 3.252220615575294,
      "grad_norm": 6.342364311218262,
      "learning_rate": 3.748691676306947e-05,
      "loss": 1.513,
      "step": 19680
    },
    {
      "epoch": 3.2538731667010947,
      "grad_norm": 23.94266700744629,
      "learning_rate": 3.7477735544170846e-05,
      "loss": 1.5372,
      "step": 19690
    },
    {
      "epoch": 3.2555257178268953,
      "grad_norm": 15.126330375671387,
      "learning_rate": 3.746855432527223e-05,
      "loss": 1.5006,
      "step": 19700
    },
    {
      "epoch": 3.257178268952696,
      "grad_norm": 23.147714614868164,
      "learning_rate": 3.7459373106373605e-05,
      "loss": 1.6257,
      "step": 19710
    },
    {
      "epoch": 3.2588308200784963,
      "grad_norm": 6.821142673492432,
      "learning_rate": 3.745019188747498e-05,
      "loss": 1.3936,
      "step": 19720
    },
    {
      "epoch": 3.2604833712042964,
      "grad_norm": 7.183291912078857,
      "learning_rate": 3.7441010668576356e-05,
      "loss": 1.4941,
      "step": 19730
    },
    {
      "epoch": 3.262135922330097,
      "grad_norm": 9.05080509185791,
      "learning_rate": 3.743182944967774e-05,
      "loss": 1.466,
      "step": 19740
    },
    {
      "epoch": 3.2637884734558975,
      "grad_norm": 30.209848403930664,
      "learning_rate": 3.742264823077912e-05,
      "loss": 1.6091,
      "step": 19750
    },
    {
      "epoch": 3.265441024581698,
      "grad_norm": 7.082402229309082,
      "learning_rate": 3.74134670118805e-05,
      "loss": 1.6003,
      "step": 19760
    },
    {
      "epoch": 3.2670935757074986,
      "grad_norm": 17.827167510986328,
      "learning_rate": 3.740428579298188e-05,
      "loss": 1.4811,
      "step": 19770
    },
    {
      "epoch": 3.2687461268332987,
      "grad_norm": 10.344867706298828,
      "learning_rate": 3.7395104574083256e-05,
      "loss": 1.5169,
      "step": 19780
    },
    {
      "epoch": 3.270398677959099,
      "grad_norm": 12.192914009094238,
      "learning_rate": 3.738592335518464e-05,
      "loss": 1.4465,
      "step": 19790
    },
    {
      "epoch": 3.2720512290848998,
      "grad_norm": 5.389448165893555,
      "learning_rate": 3.7376742136286015e-05,
      "loss": 1.4501,
      "step": 19800
    },
    {
      "epoch": 3.2737037802107003,
      "grad_norm": 10.874370574951172,
      "learning_rate": 3.73675609173874e-05,
      "loss": 1.5779,
      "step": 19810
    },
    {
      "epoch": 3.275356331336501,
      "grad_norm": 12.324655532836914,
      "learning_rate": 3.735837969848877e-05,
      "loss": 1.459,
      "step": 19820
    },
    {
      "epoch": 3.277008882462301,
      "grad_norm": 10.794198036193848,
      "learning_rate": 3.7349198479590156e-05,
      "loss": 1.4958,
      "step": 19830
    },
    {
      "epoch": 3.2786614335881015,
      "grad_norm": 14.633699417114258,
      "learning_rate": 3.734001726069153e-05,
      "loss": 1.5338,
      "step": 19840
    },
    {
      "epoch": 3.280313984713902,
      "grad_norm": 8.332018852233887,
      "learning_rate": 3.733083604179291e-05,
      "loss": 1.4962,
      "step": 19850
    },
    {
      "epoch": 3.2819665358397025,
      "grad_norm": 9.715315818786621,
      "learning_rate": 3.732165482289429e-05,
      "loss": 1.5335,
      "step": 19860
    },
    {
      "epoch": 3.283619086965503,
      "grad_norm": 13.441153526306152,
      "learning_rate": 3.7312473603995666e-05,
      "loss": 1.5538,
      "step": 19870
    },
    {
      "epoch": 3.2852716380913036,
      "grad_norm": 16.49589729309082,
      "learning_rate": 3.730329238509705e-05,
      "loss": 1.5301,
      "step": 19880
    },
    {
      "epoch": 3.2869241892171037,
      "grad_norm": 31.22637939453125,
      "learning_rate": 3.7294111166198424e-05,
      "loss": 1.6213,
      "step": 19890
    },
    {
      "epoch": 3.2885767403429043,
      "grad_norm": 8.242454528808594,
      "learning_rate": 3.728492994729981e-05,
      "loss": 1.6577,
      "step": 19900
    },
    {
      "epoch": 3.290229291468705,
      "grad_norm": 8.728832244873047,
      "learning_rate": 3.727574872840118e-05,
      "loss": 1.5487,
      "step": 19910
    },
    {
      "epoch": 3.2918818425945053,
      "grad_norm": 7.110820770263672,
      "learning_rate": 3.7266567509502566e-05,
      "loss": 1.6357,
      "step": 19920
    },
    {
      "epoch": 3.293534393720306,
      "grad_norm": 6.833790302276611,
      "learning_rate": 3.725738629060394e-05,
      "loss": 1.5007,
      "step": 19930
    },
    {
      "epoch": 3.2951869448461064,
      "grad_norm": 10.365285873413086,
      "learning_rate": 3.7248205071705324e-05,
      "loss": 1.6454,
      "step": 19940
    },
    {
      "epoch": 3.2968394959719065,
      "grad_norm": 10.187037467956543,
      "learning_rate": 3.72390238528067e-05,
      "loss": 1.4288,
      "step": 19950
    },
    {
      "epoch": 3.298492047097707,
      "grad_norm": 33.366790771484375,
      "learning_rate": 3.722984263390808e-05,
      "loss": 1.6844,
      "step": 19960
    },
    {
      "epoch": 3.3001445982235076,
      "grad_norm": 6.164004802703857,
      "learning_rate": 3.722066141500946e-05,
      "loss": 1.5483,
      "step": 19970
    },
    {
      "epoch": 3.301797149349308,
      "grad_norm": 14.696124076843262,
      "learning_rate": 3.7211480196110834e-05,
      "loss": 1.5542,
      "step": 19980
    },
    {
      "epoch": 3.3034497004751087,
      "grad_norm": 11.939413070678711,
      "learning_rate": 3.720229897721222e-05,
      "loss": 1.5707,
      "step": 19990
    },
    {
      "epoch": 3.3051022516009088,
      "grad_norm": 6.623092174530029,
      "learning_rate": 3.719311775831359e-05,
      "loss": 1.5703,
      "step": 20000
    },
    {
      "epoch": 3.3067548027267093,
      "grad_norm": 7.930966854095459,
      "learning_rate": 3.7183936539414975e-05,
      "loss": 1.4843,
      "step": 20010
    },
    {
      "epoch": 3.30840735385251,
      "grad_norm": 5.481695652008057,
      "learning_rate": 3.717475532051635e-05,
      "loss": 1.5541,
      "step": 20020
    },
    {
      "epoch": 3.3100599049783104,
      "grad_norm": 6.865151405334473,
      "learning_rate": 3.7165574101617734e-05,
      "loss": 1.5562,
      "step": 20030
    },
    {
      "epoch": 3.311712456104111,
      "grad_norm": 9.590388298034668,
      "learning_rate": 3.715639288271911e-05,
      "loss": 1.6177,
      "step": 20040
    },
    {
      "epoch": 3.313365007229911,
      "grad_norm": 7.312343120574951,
      "learning_rate": 3.714721166382049e-05,
      "loss": 1.7381,
      "step": 20050
    },
    {
      "epoch": 3.3150175583557115,
      "grad_norm": 9.11748218536377,
      "learning_rate": 3.713803044492187e-05,
      "loss": 1.5066,
      "step": 20060
    },
    {
      "epoch": 3.316670109481512,
      "grad_norm": 7.51953649520874,
      "learning_rate": 3.712884922602325e-05,
      "loss": 1.7568,
      "step": 20070
    },
    {
      "epoch": 3.3183226606073126,
      "grad_norm": 7.698461532592773,
      "learning_rate": 3.711966800712463e-05,
      "loss": 1.4384,
      "step": 20080
    },
    {
      "epoch": 3.319975211733113,
      "grad_norm": 4.527771949768066,
      "learning_rate": 3.711048678822601e-05,
      "loss": 1.5915,
      "step": 20090
    },
    {
      "epoch": 3.3216277628589133,
      "grad_norm": 7.3180131912231445,
      "learning_rate": 3.7101305569327385e-05,
      "loss": 1.5506,
      "step": 20100
    },
    {
      "epoch": 3.323280313984714,
      "grad_norm": 5.010993480682373,
      "learning_rate": 3.709212435042876e-05,
      "loss": 1.4742,
      "step": 20110
    },
    {
      "epoch": 3.3249328651105143,
      "grad_norm": 15.535231590270996,
      "learning_rate": 3.7082943131530144e-05,
      "loss": 1.5647,
      "step": 20120
    },
    {
      "epoch": 3.326585416236315,
      "grad_norm": 7.405104637145996,
      "learning_rate": 3.707376191263152e-05,
      "loss": 1.3815,
      "step": 20130
    },
    {
      "epoch": 3.3282379673621154,
      "grad_norm": 6.594423770904541,
      "learning_rate": 3.70645806937329e-05,
      "loss": 1.5013,
      "step": 20140
    },
    {
      "epoch": 3.3298905184879155,
      "grad_norm": 16.589862823486328,
      "learning_rate": 3.705539947483428e-05,
      "loss": 1.5586,
      "step": 20150
    },
    {
      "epoch": 3.331543069613716,
      "grad_norm": 21.43290138244629,
      "learning_rate": 3.704621825593566e-05,
      "loss": 1.4964,
      "step": 20160
    },
    {
      "epoch": 3.3331956207395166,
      "grad_norm": 13.260900497436523,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 1.5953,
      "step": 20170
    },
    {
      "epoch": 3.334848171865317,
      "grad_norm": 7.0593037605285645,
      "learning_rate": 3.702785581813842e-05,
      "loss": 1.6032,
      "step": 20180
    },
    {
      "epoch": 3.3365007229911177,
      "grad_norm": 5.94556188583374,
      "learning_rate": 3.7018674599239795e-05,
      "loss": 1.677,
      "step": 20190
    },
    {
      "epoch": 3.3381532741169178,
      "grad_norm": 6.961794376373291,
      "learning_rate": 3.700949338034118e-05,
      "loss": 1.3566,
      "step": 20200
    },
    {
      "epoch": 3.3398058252427183,
      "grad_norm": 12.444746017456055,
      "learning_rate": 3.700031216144256e-05,
      "loss": 1.6006,
      "step": 20210
    },
    {
      "epoch": 3.341458376368519,
      "grad_norm": 7.811690330505371,
      "learning_rate": 3.6991130942543936e-05,
      "loss": 1.557,
      "step": 20220
    },
    {
      "epoch": 3.3431109274943194,
      "grad_norm": 58.86392593383789,
      "learning_rate": 3.698194972364531e-05,
      "loss": 1.6786,
      "step": 20230
    },
    {
      "epoch": 3.34476347862012,
      "grad_norm": 13.110153198242188,
      "learning_rate": 3.697276850474669e-05,
      "loss": 1.5049,
      "step": 20240
    },
    {
      "epoch": 3.3464160297459205,
      "grad_norm": 26.155275344848633,
      "learning_rate": 3.696358728584807e-05,
      "loss": 1.6196,
      "step": 20250
    },
    {
      "epoch": 3.3480685808717205,
      "grad_norm": 9.868633270263672,
      "learning_rate": 3.6954406066949446e-05,
      "loss": 1.52,
      "step": 20260
    },
    {
      "epoch": 3.349721131997521,
      "grad_norm": 6.902059078216553,
      "learning_rate": 3.694522484805083e-05,
      "loss": 1.6092,
      "step": 20270
    },
    {
      "epoch": 3.3513736831233216,
      "grad_norm": 13.972870826721191,
      "learning_rate": 3.6936043629152205e-05,
      "loss": 1.4675,
      "step": 20280
    },
    {
      "epoch": 3.353026234249122,
      "grad_norm": 4.600803852081299,
      "learning_rate": 3.692686241025359e-05,
      "loss": 1.5782,
      "step": 20290
    },
    {
      "epoch": 3.3546787853749227,
      "grad_norm": 29.91995620727539,
      "learning_rate": 3.691768119135496e-05,
      "loss": 1.6141,
      "step": 20300
    },
    {
      "epoch": 3.3563313365007232,
      "grad_norm": 13.344480514526367,
      "learning_rate": 3.6908499972456346e-05,
      "loss": 1.5646,
      "step": 20310
    },
    {
      "epoch": 3.3579838876265233,
      "grad_norm": 36.023399353027344,
      "learning_rate": 3.689931875355773e-05,
      "loss": 1.5179,
      "step": 20320
    },
    {
      "epoch": 3.359636438752324,
      "grad_norm": 14.964554786682129,
      "learning_rate": 3.6890137534659105e-05,
      "loss": 1.523,
      "step": 20330
    },
    {
      "epoch": 3.3612889898781244,
      "grad_norm": 10.848349571228027,
      "learning_rate": 3.688095631576049e-05,
      "loss": 1.8119,
      "step": 20340
    },
    {
      "epoch": 3.362941541003925,
      "grad_norm": 13.974653244018555,
      "learning_rate": 3.687177509686186e-05,
      "loss": 1.6156,
      "step": 20350
    },
    {
      "epoch": 3.3645940921297255,
      "grad_norm": 6.851309776306152,
      "learning_rate": 3.686259387796324e-05,
      "loss": 1.5522,
      "step": 20360
    },
    {
      "epoch": 3.3662466432555256,
      "grad_norm": 7.36599063873291,
      "learning_rate": 3.6853412659064615e-05,
      "loss": 1.6563,
      "step": 20370
    },
    {
      "epoch": 3.367899194381326,
      "grad_norm": 8.092686653137207,
      "learning_rate": 3.6844231440166e-05,
      "loss": 1.4674,
      "step": 20380
    },
    {
      "epoch": 3.3695517455071267,
      "grad_norm": 13.80720329284668,
      "learning_rate": 3.683505022126737e-05,
      "loss": 1.6301,
      "step": 20390
    },
    {
      "epoch": 3.371204296632927,
      "grad_norm": 15.420303344726562,
      "learning_rate": 3.6825869002368756e-05,
      "loss": 1.5869,
      "step": 20400
    },
    {
      "epoch": 3.3728568477587277,
      "grad_norm": 8.8609037399292,
      "learning_rate": 3.681668778347013e-05,
      "loss": 1.3465,
      "step": 20410
    },
    {
      "epoch": 3.374509398884528,
      "grad_norm": 12.477286338806152,
      "learning_rate": 3.6807506564571514e-05,
      "loss": 1.6163,
      "step": 20420
    },
    {
      "epoch": 3.3761619500103284,
      "grad_norm": 18.01897430419922,
      "learning_rate": 3.67983253456729e-05,
      "loss": 1.5364,
      "step": 20430
    },
    {
      "epoch": 3.377814501136129,
      "grad_norm": 13.501622200012207,
      "learning_rate": 3.678914412677427e-05,
      "loss": 1.4659,
      "step": 20440
    },
    {
      "epoch": 3.3794670522619294,
      "grad_norm": 16.007644653320312,
      "learning_rate": 3.6779962907875655e-05,
      "loss": 1.6269,
      "step": 20450
    },
    {
      "epoch": 3.38111960338773,
      "grad_norm": 6.28301477432251,
      "learning_rate": 3.677078168897703e-05,
      "loss": 1.4513,
      "step": 20460
    },
    {
      "epoch": 3.38277215451353,
      "grad_norm": 8.616754531860352,
      "learning_rate": 3.6761600470078414e-05,
      "loss": 1.4903,
      "step": 20470
    },
    {
      "epoch": 3.3844247056393306,
      "grad_norm": 9.9006986618042,
      "learning_rate": 3.675241925117979e-05,
      "loss": 1.5364,
      "step": 20480
    },
    {
      "epoch": 3.386077256765131,
      "grad_norm": 11.879949569702148,
      "learning_rate": 3.6743238032281166e-05,
      "loss": 1.6025,
      "step": 20490
    },
    {
      "epoch": 3.3877298078909317,
      "grad_norm": 28.69721221923828,
      "learning_rate": 3.673405681338254e-05,
      "loss": 1.5899,
      "step": 20500
    },
    {
      "epoch": 3.3893823590167322,
      "grad_norm": 10.507150650024414,
      "learning_rate": 3.6724875594483924e-05,
      "loss": 1.6434,
      "step": 20510
    },
    {
      "epoch": 3.3910349101425323,
      "grad_norm": 10.478775978088379,
      "learning_rate": 3.67156943755853e-05,
      "loss": 1.5265,
      "step": 20520
    },
    {
      "epoch": 3.392687461268333,
      "grad_norm": 24.88822364807129,
      "learning_rate": 3.670651315668668e-05,
      "loss": 1.6751,
      "step": 20530
    },
    {
      "epoch": 3.3943400123941334,
      "grad_norm": 9.983197212219238,
      "learning_rate": 3.669733193778806e-05,
      "loss": 1.5644,
      "step": 20540
    },
    {
      "epoch": 3.395992563519934,
      "grad_norm": 7.989924430847168,
      "learning_rate": 3.668815071888944e-05,
      "loss": 1.6401,
      "step": 20550
    },
    {
      "epoch": 3.3976451146457345,
      "grad_norm": 5.701867580413818,
      "learning_rate": 3.6678969499990824e-05,
      "loss": 1.5425,
      "step": 20560
    },
    {
      "epoch": 3.3992976657715346,
      "grad_norm": 14.750271797180176,
      "learning_rate": 3.66697882810922e-05,
      "loss": 1.585,
      "step": 20570
    },
    {
      "epoch": 3.400950216897335,
      "grad_norm": 21.3060245513916,
      "learning_rate": 3.666060706219358e-05,
      "loss": 1.5328,
      "step": 20580
    },
    {
      "epoch": 3.4026027680231357,
      "grad_norm": 14.30643367767334,
      "learning_rate": 3.665142584329496e-05,
      "loss": 1.5749,
      "step": 20590
    },
    {
      "epoch": 3.404255319148936,
      "grad_norm": 10.9791898727417,
      "learning_rate": 3.664224462439634e-05,
      "loss": 1.6841,
      "step": 20600
    },
    {
      "epoch": 3.4059078702747367,
      "grad_norm": 10.282464981079102,
      "learning_rate": 3.663306340549772e-05,
      "loss": 1.586,
      "step": 20610
    },
    {
      "epoch": 3.407560421400537,
      "grad_norm": 10.054344177246094,
      "learning_rate": 3.662388218659909e-05,
      "loss": 1.5445,
      "step": 20620
    },
    {
      "epoch": 3.4092129725263374,
      "grad_norm": 6.583102703094482,
      "learning_rate": 3.661470096770047e-05,
      "loss": 1.507,
      "step": 20630
    },
    {
      "epoch": 3.410865523652138,
      "grad_norm": 7.482510566711426,
      "learning_rate": 3.660551974880185e-05,
      "loss": 1.5865,
      "step": 20640
    },
    {
      "epoch": 3.4125180747779384,
      "grad_norm": 23.863901138305664,
      "learning_rate": 3.659633852990323e-05,
      "loss": 1.7097,
      "step": 20650
    },
    {
      "epoch": 3.414170625903739,
      "grad_norm": 5.8396830558776855,
      "learning_rate": 3.658715731100461e-05,
      "loss": 1.4904,
      "step": 20660
    },
    {
      "epoch": 3.4158231770295395,
      "grad_norm": 14.669455528259277,
      "learning_rate": 3.657797609210599e-05,
      "loss": 1.5476,
      "step": 20670
    },
    {
      "epoch": 3.4174757281553396,
      "grad_norm": 9.562002182006836,
      "learning_rate": 3.656879487320737e-05,
      "loss": 1.474,
      "step": 20680
    },
    {
      "epoch": 3.41912827928114,
      "grad_norm": 13.14786434173584,
      "learning_rate": 3.655961365430875e-05,
      "loss": 1.6046,
      "step": 20690
    },
    {
      "epoch": 3.4207808304069407,
      "grad_norm": 27.712749481201172,
      "learning_rate": 3.6550432435410127e-05,
      "loss": 1.6887,
      "step": 20700
    },
    {
      "epoch": 3.4224333815327412,
      "grad_norm": 8.249465942382812,
      "learning_rate": 3.654125121651151e-05,
      "loss": 1.5875,
      "step": 20710
    },
    {
      "epoch": 3.4240859326585418,
      "grad_norm": 34.13796615600586,
      "learning_rate": 3.6532069997612885e-05,
      "loss": 1.5839,
      "step": 20720
    },
    {
      "epoch": 3.4257384837843423,
      "grad_norm": 9.606948852539062,
      "learning_rate": 3.652288877871427e-05,
      "loss": 1.6466,
      "step": 20730
    },
    {
      "epoch": 3.4273910349101424,
      "grad_norm": 11.323090553283691,
      "learning_rate": 3.6513707559815644e-05,
      "loss": 1.4478,
      "step": 20740
    },
    {
      "epoch": 3.429043586035943,
      "grad_norm": 6.942873477935791,
      "learning_rate": 3.650452634091702e-05,
      "loss": 1.5668,
      "step": 20750
    },
    {
      "epoch": 3.4306961371617435,
      "grad_norm": 6.9645867347717285,
      "learning_rate": 3.6495345122018395e-05,
      "loss": 1.3887,
      "step": 20760
    },
    {
      "epoch": 3.432348688287544,
      "grad_norm": 9.70297622680664,
      "learning_rate": 3.648616390311978e-05,
      "loss": 1.5917,
      "step": 20770
    },
    {
      "epoch": 3.4340012394133446,
      "grad_norm": 19.935123443603516,
      "learning_rate": 3.647698268422116e-05,
      "loss": 1.6505,
      "step": 20780
    },
    {
      "epoch": 3.4356537905391447,
      "grad_norm": 8.549393653869629,
      "learning_rate": 3.6467801465322536e-05,
      "loss": 1.7168,
      "step": 20790
    },
    {
      "epoch": 3.437306341664945,
      "grad_norm": 7.79632568359375,
      "learning_rate": 3.645862024642392e-05,
      "loss": 1.5747,
      "step": 20800
    },
    {
      "epoch": 3.4389588927907457,
      "grad_norm": 15.404546737670898,
      "learning_rate": 3.6449439027525295e-05,
      "loss": 1.54,
      "step": 20810
    },
    {
      "epoch": 3.4406114439165463,
      "grad_norm": 7.910075664520264,
      "learning_rate": 3.644025780862668e-05,
      "loss": 1.4787,
      "step": 20820
    },
    {
      "epoch": 3.442263995042347,
      "grad_norm": 38.832923889160156,
      "learning_rate": 3.643107658972805e-05,
      "loss": 1.3986,
      "step": 20830
    },
    {
      "epoch": 3.443916546168147,
      "grad_norm": 15.27536392211914,
      "learning_rate": 3.6421895370829436e-05,
      "loss": 1.5966,
      "step": 20840
    },
    {
      "epoch": 3.4455690972939474,
      "grad_norm": 5.51830530166626,
      "learning_rate": 3.641271415193081e-05,
      "loss": 1.4576,
      "step": 20850
    },
    {
      "epoch": 3.447221648419748,
      "grad_norm": 9.22146987915039,
      "learning_rate": 3.6403532933032194e-05,
      "loss": 1.5793,
      "step": 20860
    },
    {
      "epoch": 3.4488741995455485,
      "grad_norm": 18.014942169189453,
      "learning_rate": 3.639435171413357e-05,
      "loss": 1.4621,
      "step": 20870
    },
    {
      "epoch": 3.450526750671349,
      "grad_norm": 11.17994499206543,
      "learning_rate": 3.6385170495234946e-05,
      "loss": 1.5456,
      "step": 20880
    },
    {
      "epoch": 3.452179301797149,
      "grad_norm": 7.204450607299805,
      "learning_rate": 3.637598927633633e-05,
      "loss": 1.5712,
      "step": 20890
    },
    {
      "epoch": 3.4538318529229497,
      "grad_norm": 9.417290687561035,
      "learning_rate": 3.6366808057437705e-05,
      "loss": 1.4765,
      "step": 20900
    },
    {
      "epoch": 3.4554844040487502,
      "grad_norm": 8.76371955871582,
      "learning_rate": 3.635762683853909e-05,
      "loss": 1.6961,
      "step": 20910
    },
    {
      "epoch": 3.4571369551745508,
      "grad_norm": 9.267240524291992,
      "learning_rate": 3.634844561964046e-05,
      "loss": 1.5875,
      "step": 20920
    },
    {
      "epoch": 3.4587895063003513,
      "grad_norm": 11.603541374206543,
      "learning_rate": 3.6339264400741846e-05,
      "loss": 1.6531,
      "step": 20930
    },
    {
      "epoch": 3.4604420574261514,
      "grad_norm": 8.296456336975098,
      "learning_rate": 3.633008318184322e-05,
      "loss": 1.6704,
      "step": 20940
    },
    {
      "epoch": 3.462094608551952,
      "grad_norm": 8.473135948181152,
      "learning_rate": 3.6320901962944604e-05,
      "loss": 1.4977,
      "step": 20950
    },
    {
      "epoch": 3.4637471596777525,
      "grad_norm": 18.611042022705078,
      "learning_rate": 3.631172074404598e-05,
      "loss": 1.475,
      "step": 20960
    },
    {
      "epoch": 3.465399710803553,
      "grad_norm": 7.0088090896606445,
      "learning_rate": 3.630253952514736e-05,
      "loss": 1.4484,
      "step": 20970
    },
    {
      "epoch": 3.4670522619293536,
      "grad_norm": 9.421944618225098,
      "learning_rate": 3.629335830624874e-05,
      "loss": 1.4781,
      "step": 20980
    },
    {
      "epoch": 3.4687048130551537,
      "grad_norm": 35.80681610107422,
      "learning_rate": 3.628417708735012e-05,
      "loss": 1.5584,
      "step": 20990
    },
    {
      "epoch": 3.470357364180954,
      "grad_norm": 7.813302040100098,
      "learning_rate": 3.62749958684515e-05,
      "loss": 1.5762,
      "step": 21000
    },
    {
      "epoch": 3.4720099153067547,
      "grad_norm": 21.3023681640625,
      "learning_rate": 3.626581464955287e-05,
      "loss": 1.6249,
      "step": 21010
    },
    {
      "epoch": 3.4736624664325553,
      "grad_norm": 62.12176513671875,
      "learning_rate": 3.6256633430654256e-05,
      "loss": 1.5029,
      "step": 21020
    },
    {
      "epoch": 3.475315017558356,
      "grad_norm": 8.91295337677002,
      "learning_rate": 3.624745221175563e-05,
      "loss": 1.5845,
      "step": 21030
    },
    {
      "epoch": 3.4769675686841564,
      "grad_norm": 10.726168632507324,
      "learning_rate": 3.6238270992857014e-05,
      "loss": 1.6008,
      "step": 21040
    },
    {
      "epoch": 3.4786201198099564,
      "grad_norm": 7.324353218078613,
      "learning_rate": 3.622908977395839e-05,
      "loss": 1.5174,
      "step": 21050
    },
    {
      "epoch": 3.480272670935757,
      "grad_norm": 14.042418479919434,
      "learning_rate": 3.621990855505977e-05,
      "loss": 1.6201,
      "step": 21060
    },
    {
      "epoch": 3.4819252220615575,
      "grad_norm": 22.140363693237305,
      "learning_rate": 3.621072733616115e-05,
      "loss": 1.6755,
      "step": 21070
    },
    {
      "epoch": 3.483577773187358,
      "grad_norm": 10.497157096862793,
      "learning_rate": 3.620154611726253e-05,
      "loss": 1.4485,
      "step": 21080
    },
    {
      "epoch": 3.4852303243131586,
      "grad_norm": 7.269160270690918,
      "learning_rate": 3.619236489836391e-05,
      "loss": 1.5914,
      "step": 21090
    },
    {
      "epoch": 3.486882875438959,
      "grad_norm": 11.066585540771484,
      "learning_rate": 3.618318367946529e-05,
      "loss": 1.4845,
      "step": 21100
    },
    {
      "epoch": 3.4885354265647592,
      "grad_norm": 16.51125144958496,
      "learning_rate": 3.6174002460566666e-05,
      "loss": 1.5517,
      "step": 21110
    },
    {
      "epoch": 3.4901879776905598,
      "grad_norm": 35.069183349609375,
      "learning_rate": 3.616482124166805e-05,
      "loss": 1.5281,
      "step": 21120
    },
    {
      "epoch": 3.4918405288163603,
      "grad_norm": 16.97264289855957,
      "learning_rate": 3.6155640022769424e-05,
      "loss": 1.469,
      "step": 21130
    },
    {
      "epoch": 3.493493079942161,
      "grad_norm": 21.5349178314209,
      "learning_rate": 3.61464588038708e-05,
      "loss": 1.6392,
      "step": 21140
    },
    {
      "epoch": 3.4951456310679614,
      "grad_norm": 8.533900260925293,
      "learning_rate": 3.613727758497218e-05,
      "loss": 1.5645,
      "step": 21150
    },
    {
      "epoch": 3.4967981821937615,
      "grad_norm": 59.57920455932617,
      "learning_rate": 3.612809636607356e-05,
      "loss": 1.598,
      "step": 21160
    },
    {
      "epoch": 3.498450733319562,
      "grad_norm": 34.87241744995117,
      "learning_rate": 3.611891514717494e-05,
      "loss": 1.6019,
      "step": 21170
    },
    {
      "epoch": 3.5001032844453626,
      "grad_norm": 12.19272232055664,
      "learning_rate": 3.610973392827632e-05,
      "loss": 1.5174,
      "step": 21180
    },
    {
      "epoch": 3.501755835571163,
      "grad_norm": 23.954174041748047,
      "learning_rate": 3.61005527093777e-05,
      "loss": 1.6589,
      "step": 21190
    },
    {
      "epoch": 3.5034083866969636,
      "grad_norm": 11.947124481201172,
      "learning_rate": 3.6091371490479075e-05,
      "loss": 1.6736,
      "step": 21200
    },
    {
      "epoch": 3.5050609378227637,
      "grad_norm": 30.422679901123047,
      "learning_rate": 3.608219027158046e-05,
      "loss": 1.5366,
      "step": 21210
    },
    {
      "epoch": 3.5067134889485643,
      "grad_norm": 5.95852518081665,
      "learning_rate": 3.6073009052681834e-05,
      "loss": 1.6103,
      "step": 21220
    },
    {
      "epoch": 3.508366040074365,
      "grad_norm": 7.431918144226074,
      "learning_rate": 3.6063827833783216e-05,
      "loss": 1.4234,
      "step": 21230
    },
    {
      "epoch": 3.5100185912001653,
      "grad_norm": 19.810720443725586,
      "learning_rate": 3.60546466148846e-05,
      "loss": 1.5673,
      "step": 21240
    },
    {
      "epoch": 3.511671142325966,
      "grad_norm": 9.25375747680664,
      "learning_rate": 3.6045465395985975e-05,
      "loss": 1.5157,
      "step": 21250
    },
    {
      "epoch": 3.513323693451766,
      "grad_norm": 15.374961853027344,
      "learning_rate": 3.603628417708735e-05,
      "loss": 1.5093,
      "step": 21260
    },
    {
      "epoch": 3.5149762445775665,
      "grad_norm": 15.200456619262695,
      "learning_rate": 3.602710295818873e-05,
      "loss": 1.6222,
      "step": 21270
    },
    {
      "epoch": 3.516628795703367,
      "grad_norm": 10.129278182983398,
      "learning_rate": 3.601792173929011e-05,
      "loss": 1.5672,
      "step": 21280
    },
    {
      "epoch": 3.5182813468291676,
      "grad_norm": 15.771454811096191,
      "learning_rate": 3.6008740520391485e-05,
      "loss": 1.6731,
      "step": 21290
    },
    {
      "epoch": 3.519933897954968,
      "grad_norm": 10.804917335510254,
      "learning_rate": 3.599955930149287e-05,
      "loss": 1.6508,
      "step": 21300
    },
    {
      "epoch": 3.5215864490807682,
      "grad_norm": 10.056406021118164,
      "learning_rate": 3.5990378082594244e-05,
      "loss": 1.527,
      "step": 21310
    },
    {
      "epoch": 3.5232390002065688,
      "grad_norm": 14.683306694030762,
      "learning_rate": 3.5981196863695626e-05,
      "loss": 1.6713,
      "step": 21320
    },
    {
      "epoch": 3.5248915513323693,
      "grad_norm": 12.720149040222168,
      "learning_rate": 3.5972015644797e-05,
      "loss": 1.5369,
      "step": 21330
    },
    {
      "epoch": 3.52654410245817,
      "grad_norm": 20.362594604492188,
      "learning_rate": 3.5962834425898385e-05,
      "loss": 1.4948,
      "step": 21340
    },
    {
      "epoch": 3.5281966535839704,
      "grad_norm": 9.18614387512207,
      "learning_rate": 3.595365320699977e-05,
      "loss": 1.5522,
      "step": 21350
    },
    {
      "epoch": 3.5298492047097705,
      "grad_norm": 15.578404426574707,
      "learning_rate": 3.594447198810114e-05,
      "loss": 1.5016,
      "step": 21360
    },
    {
      "epoch": 3.531501755835571,
      "grad_norm": 28.72294807434082,
      "learning_rate": 3.5935290769202526e-05,
      "loss": 1.5429,
      "step": 21370
    },
    {
      "epoch": 3.5331543069613716,
      "grad_norm": 8.129132270812988,
      "learning_rate": 3.59261095503039e-05,
      "loss": 1.6308,
      "step": 21380
    },
    {
      "epoch": 3.534806858087172,
      "grad_norm": 7.975843906402588,
      "learning_rate": 3.591692833140528e-05,
      "loss": 1.5086,
      "step": 21390
    },
    {
      "epoch": 3.5364594092129726,
      "grad_norm": 69.10977935791016,
      "learning_rate": 3.5907747112506654e-05,
      "loss": 1.6113,
      "step": 21400
    },
    {
      "epoch": 3.5381119603387727,
      "grad_norm": 28.328746795654297,
      "learning_rate": 3.5898565893608036e-05,
      "loss": 1.5441,
      "step": 21410
    },
    {
      "epoch": 3.5397645114645733,
      "grad_norm": 15.218031883239746,
      "learning_rate": 3.588938467470941e-05,
      "loss": 1.6617,
      "step": 21420
    },
    {
      "epoch": 3.541417062590374,
      "grad_norm": 7.890848636627197,
      "learning_rate": 3.5880203455810795e-05,
      "loss": 1.4507,
      "step": 21430
    },
    {
      "epoch": 3.5430696137161743,
      "grad_norm": 23.32213592529297,
      "learning_rate": 3.587102223691217e-05,
      "loss": 1.4829,
      "step": 21440
    },
    {
      "epoch": 3.544722164841975,
      "grad_norm": 9.921598434448242,
      "learning_rate": 3.586184101801355e-05,
      "loss": 1.5639,
      "step": 21450
    },
    {
      "epoch": 3.546374715967775,
      "grad_norm": 10.007643699645996,
      "learning_rate": 3.5852659799114936e-05,
      "loss": 1.6078,
      "step": 21460
    },
    {
      "epoch": 3.548027267093576,
      "grad_norm": 6.750604629516602,
      "learning_rate": 3.584347858021631e-05,
      "loss": 1.5142,
      "step": 21470
    },
    {
      "epoch": 3.549679818219376,
      "grad_norm": 12.836925506591797,
      "learning_rate": 3.5834297361317694e-05,
      "loss": 1.5415,
      "step": 21480
    },
    {
      "epoch": 3.5513323693451766,
      "grad_norm": 9.204744338989258,
      "learning_rate": 3.582511614241907e-05,
      "loss": 1.5137,
      "step": 21490
    },
    {
      "epoch": 3.552984920470977,
      "grad_norm": 22.862333297729492,
      "learning_rate": 3.581593492352045e-05,
      "loss": 1.5273,
      "step": 21500
    },
    {
      "epoch": 3.5546374715967777,
      "grad_norm": 8.714118003845215,
      "learning_rate": 3.580675370462183e-05,
      "loss": 1.6176,
      "step": 21510
    },
    {
      "epoch": 3.556290022722578,
      "grad_norm": 32.12099838256836,
      "learning_rate": 3.5797572485723204e-05,
      "loss": 1.5262,
      "step": 21520
    },
    {
      "epoch": 3.5579425738483783,
      "grad_norm": 33.427494049072266,
      "learning_rate": 3.578839126682458e-05,
      "loss": 1.6296,
      "step": 21530
    },
    {
      "epoch": 3.559595124974179,
      "grad_norm": 7.42344856262207,
      "learning_rate": 3.577921004792596e-05,
      "loss": 1.6682,
      "step": 21540
    },
    {
      "epoch": 3.5612476760999794,
      "grad_norm": 7.865045547485352,
      "learning_rate": 3.577002882902734e-05,
      "loss": 1.6207,
      "step": 21550
    },
    {
      "epoch": 3.56290022722578,
      "grad_norm": 8.04334831237793,
      "learning_rate": 3.576084761012872e-05,
      "loss": 1.5475,
      "step": 21560
    },
    {
      "epoch": 3.5645527783515805,
      "grad_norm": 9.874261856079102,
      "learning_rate": 3.5751666391230104e-05,
      "loss": 1.5455,
      "step": 21570
    },
    {
      "epoch": 3.5662053294773806,
      "grad_norm": 18.90970802307129,
      "learning_rate": 3.574248517233148e-05,
      "loss": 1.6134,
      "step": 21580
    },
    {
      "epoch": 3.567857880603181,
      "grad_norm": 7.690478324890137,
      "learning_rate": 3.573330395343286e-05,
      "loss": 1.6368,
      "step": 21590
    },
    {
      "epoch": 3.5695104317289816,
      "grad_norm": 12.998335838317871,
      "learning_rate": 3.572412273453424e-05,
      "loss": 1.6065,
      "step": 21600
    },
    {
      "epoch": 3.571162982854782,
      "grad_norm": 6.6522979736328125,
      "learning_rate": 3.571494151563562e-05,
      "loss": 1.3663,
      "step": 21610
    },
    {
      "epoch": 3.5728155339805827,
      "grad_norm": 9.580485343933105,
      "learning_rate": 3.5705760296737e-05,
      "loss": 1.5932,
      "step": 21620
    },
    {
      "epoch": 3.574468085106383,
      "grad_norm": 14.493224143981934,
      "learning_rate": 3.569657907783838e-05,
      "loss": 1.548,
      "step": 21630
    },
    {
      "epoch": 3.5761206362321833,
      "grad_norm": 9.22100830078125,
      "learning_rate": 3.5687397858939755e-05,
      "loss": 1.4881,
      "step": 21640
    },
    {
      "epoch": 3.577773187357984,
      "grad_norm": 32.90215301513672,
      "learning_rate": 3.567821664004113e-05,
      "loss": 1.5178,
      "step": 21650
    },
    {
      "epoch": 3.5794257384837844,
      "grad_norm": 10.091255187988281,
      "learning_rate": 3.566903542114251e-05,
      "loss": 1.5195,
      "step": 21660
    },
    {
      "epoch": 3.581078289609585,
      "grad_norm": 6.877391815185547,
      "learning_rate": 3.565985420224389e-05,
      "loss": 1.5511,
      "step": 21670
    },
    {
      "epoch": 3.582730840735385,
      "grad_norm": 15.701227188110352,
      "learning_rate": 3.5650672983345266e-05,
      "loss": 1.6004,
      "step": 21680
    },
    {
      "epoch": 3.5843833918611856,
      "grad_norm": 5.971979141235352,
      "learning_rate": 3.564149176444665e-05,
      "loss": 1.4233,
      "step": 21690
    },
    {
      "epoch": 3.586035942986986,
      "grad_norm": 28.64961051940918,
      "learning_rate": 3.563231054554803e-05,
      "loss": 1.5402,
      "step": 21700
    },
    {
      "epoch": 3.5876884941127867,
      "grad_norm": 18.97611427307129,
      "learning_rate": 3.562312932664941e-05,
      "loss": 1.6188,
      "step": 21710
    },
    {
      "epoch": 3.589341045238587,
      "grad_norm": 11.403149604797363,
      "learning_rate": 3.561394810775079e-05,
      "loss": 1.6067,
      "step": 21720
    },
    {
      "epoch": 3.5909935963643873,
      "grad_norm": 10.633176803588867,
      "learning_rate": 3.5604766888852165e-05,
      "loss": 1.5884,
      "step": 21730
    },
    {
      "epoch": 3.592646147490188,
      "grad_norm": 6.8395490646362305,
      "learning_rate": 3.559558566995355e-05,
      "loss": 1.5274,
      "step": 21740
    },
    {
      "epoch": 3.5942986986159884,
      "grad_norm": 7.640970706939697,
      "learning_rate": 3.5586404451054924e-05,
      "loss": 1.4888,
      "step": 21750
    },
    {
      "epoch": 3.595951249741789,
      "grad_norm": 27.678407669067383,
      "learning_rate": 3.5577223232156306e-05,
      "loss": 1.4142,
      "step": 21760
    },
    {
      "epoch": 3.5976038008675895,
      "grad_norm": 15.568962097167969,
      "learning_rate": 3.556804201325768e-05,
      "loss": 1.5022,
      "step": 21770
    },
    {
      "epoch": 3.5992563519933896,
      "grad_norm": 7.94047737121582,
      "learning_rate": 3.555886079435906e-05,
      "loss": 1.4938,
      "step": 21780
    },
    {
      "epoch": 3.60090890311919,
      "grad_norm": 8.355306625366211,
      "learning_rate": 3.5549679575460434e-05,
      "loss": 1.6414,
      "step": 21790
    },
    {
      "epoch": 3.6025614542449906,
      "grad_norm": 9.412763595581055,
      "learning_rate": 3.554049835656182e-05,
      "loss": 1.5618,
      "step": 21800
    },
    {
      "epoch": 3.604214005370791,
      "grad_norm": 8.5521240234375,
      "learning_rate": 3.55313171376632e-05,
      "loss": 1.5158,
      "step": 21810
    },
    {
      "epoch": 3.6058665564965917,
      "grad_norm": 14.980252265930176,
      "learning_rate": 3.5522135918764575e-05,
      "loss": 1.5169,
      "step": 21820
    },
    {
      "epoch": 3.607519107622392,
      "grad_norm": 5.556758403778076,
      "learning_rate": 3.551295469986596e-05,
      "loss": 1.5842,
      "step": 21830
    },
    {
      "epoch": 3.609171658748193,
      "grad_norm": 18.579702377319336,
      "learning_rate": 3.5503773480967334e-05,
      "loss": 1.5249,
      "step": 21840
    },
    {
      "epoch": 3.610824209873993,
      "grad_norm": 26.00187873840332,
      "learning_rate": 3.5494592262068716e-05,
      "loss": 1.5037,
      "step": 21850
    },
    {
      "epoch": 3.6124767609997934,
      "grad_norm": 34.09498977661133,
      "learning_rate": 3.548541104317009e-05,
      "loss": 1.4003,
      "step": 21860
    },
    {
      "epoch": 3.614129312125594,
      "grad_norm": 15.18822956085205,
      "learning_rate": 3.5476229824271475e-05,
      "loss": 1.6049,
      "step": 21870
    },
    {
      "epoch": 3.6157818632513945,
      "grad_norm": 8.645750999450684,
      "learning_rate": 3.546704860537285e-05,
      "loss": 1.5309,
      "step": 21880
    },
    {
      "epoch": 3.617434414377195,
      "grad_norm": 9.381811141967773,
      "learning_rate": 3.545786738647423e-05,
      "loss": 1.3985,
      "step": 21890
    },
    {
      "epoch": 3.619086965502995,
      "grad_norm": 10.318458557128906,
      "learning_rate": 3.544868616757561e-05,
      "loss": 1.5741,
      "step": 21900
    },
    {
      "epoch": 3.6207395166287957,
      "grad_norm": 7.140669345855713,
      "learning_rate": 3.5439504948676985e-05,
      "loss": 1.5998,
      "step": 21910
    },
    {
      "epoch": 3.622392067754596,
      "grad_norm": 11.092816352844238,
      "learning_rate": 3.543032372977837e-05,
      "loss": 1.6585,
      "step": 21920
    },
    {
      "epoch": 3.6240446188803968,
      "grad_norm": 14.348875045776367,
      "learning_rate": 3.5421142510879743e-05,
      "loss": 1.6889,
      "step": 21930
    },
    {
      "epoch": 3.6256971700061973,
      "grad_norm": 20.467607498168945,
      "learning_rate": 3.5411961291981126e-05,
      "loss": 1.5698,
      "step": 21940
    },
    {
      "epoch": 3.6273497211319974,
      "grad_norm": 6.864068508148193,
      "learning_rate": 3.54027800730825e-05,
      "loss": 1.3678,
      "step": 21950
    },
    {
      "epoch": 3.629002272257798,
      "grad_norm": 10.878697395324707,
      "learning_rate": 3.5393598854183885e-05,
      "loss": 1.5885,
      "step": 21960
    },
    {
      "epoch": 3.6306548233835985,
      "grad_norm": 11.186368942260742,
      "learning_rate": 3.538441763528526e-05,
      "loss": 1.4722,
      "step": 21970
    },
    {
      "epoch": 3.632307374509399,
      "grad_norm": 11.424324989318848,
      "learning_rate": 3.537523641638664e-05,
      "loss": 1.6041,
      "step": 21980
    },
    {
      "epoch": 3.6339599256351995,
      "grad_norm": 11.108640670776367,
      "learning_rate": 3.536605519748802e-05,
      "loss": 1.5555,
      "step": 21990
    },
    {
      "epoch": 3.6356124767609996,
      "grad_norm": 6.946378707885742,
      "learning_rate": 3.53568739785894e-05,
      "loss": 1.5627,
      "step": 22000
    },
    {
      "epoch": 3.6372650278868,
      "grad_norm": 17.414968490600586,
      "learning_rate": 3.534769275969078e-05,
      "loss": 1.5827,
      "step": 22010
    },
    {
      "epoch": 3.6389175790126007,
      "grad_norm": 8.051285743713379,
      "learning_rate": 3.533851154079216e-05,
      "loss": 1.561,
      "step": 22020
    },
    {
      "epoch": 3.6405701301384013,
      "grad_norm": 6.667621612548828,
      "learning_rate": 3.5329330321893536e-05,
      "loss": 1.4156,
      "step": 22030
    },
    {
      "epoch": 3.642222681264202,
      "grad_norm": 19.320453643798828,
      "learning_rate": 3.532014910299491e-05,
      "loss": 1.5852,
      "step": 22040
    },
    {
      "epoch": 3.643875232390002,
      "grad_norm": 8.498446464538574,
      "learning_rate": 3.5310967884096294e-05,
      "loss": 1.5934,
      "step": 22050
    },
    {
      "epoch": 3.6455277835158024,
      "grad_norm": 8.024924278259277,
      "learning_rate": 3.530178666519767e-05,
      "loss": 1.5443,
      "step": 22060
    },
    {
      "epoch": 3.647180334641603,
      "grad_norm": 14.608724594116211,
      "learning_rate": 3.529260544629905e-05,
      "loss": 1.6092,
      "step": 22070
    },
    {
      "epoch": 3.6488328857674035,
      "grad_norm": 8.343809127807617,
      "learning_rate": 3.528342422740043e-05,
      "loss": 1.6338,
      "step": 22080
    },
    {
      "epoch": 3.650485436893204,
      "grad_norm": 8.434662818908691,
      "learning_rate": 3.527424300850181e-05,
      "loss": 1.5697,
      "step": 22090
    },
    {
      "epoch": 3.652137988019004,
      "grad_norm": 7.201797008514404,
      "learning_rate": 3.526506178960319e-05,
      "loss": 1.4681,
      "step": 22100
    },
    {
      "epoch": 3.6537905391448047,
      "grad_norm": 7.6197896003723145,
      "learning_rate": 3.525588057070457e-05,
      "loss": 1.6327,
      "step": 22110
    },
    {
      "epoch": 3.655443090270605,
      "grad_norm": 6.695487022399902,
      "learning_rate": 3.5246699351805946e-05,
      "loss": 1.5547,
      "step": 22120
    },
    {
      "epoch": 3.6570956413964057,
      "grad_norm": 7.320518493652344,
      "learning_rate": 3.523751813290733e-05,
      "loss": 1.376,
      "step": 22130
    },
    {
      "epoch": 3.6587481925222063,
      "grad_norm": 10.218902587890625,
      "learning_rate": 3.5228336914008704e-05,
      "loss": 1.6282,
      "step": 22140
    },
    {
      "epoch": 3.6604007436480064,
      "grad_norm": 26.678739547729492,
      "learning_rate": 3.521915569511009e-05,
      "loss": 1.5529,
      "step": 22150
    },
    {
      "epoch": 3.662053294773807,
      "grad_norm": 7.635467529296875,
      "learning_rate": 3.520997447621146e-05,
      "loss": 1.5517,
      "step": 22160
    },
    {
      "epoch": 3.6637058458996075,
      "grad_norm": 21.92401123046875,
      "learning_rate": 3.520079325731284e-05,
      "loss": 1.5591,
      "step": 22170
    },
    {
      "epoch": 3.665358397025408,
      "grad_norm": 5.554649829864502,
      "learning_rate": 3.519161203841422e-05,
      "loss": 1.5443,
      "step": 22180
    },
    {
      "epoch": 3.6670109481512085,
      "grad_norm": 9.22087287902832,
      "learning_rate": 3.51824308195156e-05,
      "loss": 1.3653,
      "step": 22190
    },
    {
      "epoch": 3.6686634992770086,
      "grad_norm": 9.278251647949219,
      "learning_rate": 3.517324960061698e-05,
      "loss": 1.5557,
      "step": 22200
    },
    {
      "epoch": 3.670316050402809,
      "grad_norm": 9.113847732543945,
      "learning_rate": 3.5164068381718356e-05,
      "loss": 1.5877,
      "step": 22210
    },
    {
      "epoch": 3.6719686015286097,
      "grad_norm": 7.552914619445801,
      "learning_rate": 3.515488716281974e-05,
      "loss": 1.5065,
      "step": 22220
    },
    {
      "epoch": 3.6736211526544102,
      "grad_norm": 10.888140678405762,
      "learning_rate": 3.5145705943921114e-05,
      "loss": 1.4641,
      "step": 22230
    },
    {
      "epoch": 3.675273703780211,
      "grad_norm": 11.267471313476562,
      "learning_rate": 3.51365247250225e-05,
      "loss": 1.6272,
      "step": 22240
    },
    {
      "epoch": 3.676926254906011,
      "grad_norm": 7.349489688873291,
      "learning_rate": 3.512734350612387e-05,
      "loss": 1.5629,
      "step": 22250
    },
    {
      "epoch": 3.678578806031812,
      "grad_norm": 9.634526252746582,
      "learning_rate": 3.5118162287225255e-05,
      "loss": 1.5289,
      "step": 22260
    },
    {
      "epoch": 3.680231357157612,
      "grad_norm": 9.006583213806152,
      "learning_rate": 3.510898106832664e-05,
      "loss": 1.5232,
      "step": 22270
    },
    {
      "epoch": 3.6818839082834125,
      "grad_norm": 9.722323417663574,
      "learning_rate": 3.5099799849428014e-05,
      "loss": 1.4823,
      "step": 22280
    },
    {
      "epoch": 3.683536459409213,
      "grad_norm": 7.501804828643799,
      "learning_rate": 3.509061863052939e-05,
      "loss": 1.609,
      "step": 22290
    },
    {
      "epoch": 3.6851890105350136,
      "grad_norm": 62.3031120300293,
      "learning_rate": 3.5081437411630765e-05,
      "loss": 1.554,
      "step": 22300
    },
    {
      "epoch": 3.686841561660814,
      "grad_norm": 7.480673313140869,
      "learning_rate": 3.507225619273215e-05,
      "loss": 1.4209,
      "step": 22310
    },
    {
      "epoch": 3.688494112786614,
      "grad_norm": 8.595160484313965,
      "learning_rate": 3.5063074973833524e-05,
      "loss": 1.7512,
      "step": 22320
    },
    {
      "epoch": 3.6901466639124147,
      "grad_norm": 5.803942680358887,
      "learning_rate": 3.5053893754934907e-05,
      "loss": 1.5432,
      "step": 22330
    },
    {
      "epoch": 3.6917992150382153,
      "grad_norm": 13.61528491973877,
      "learning_rate": 3.504471253603628e-05,
      "loss": 1.617,
      "step": 22340
    },
    {
      "epoch": 3.693451766164016,
      "grad_norm": 60.33266067504883,
      "learning_rate": 3.5035531317137665e-05,
      "loss": 1.5357,
      "step": 22350
    },
    {
      "epoch": 3.6951043172898164,
      "grad_norm": 11.57695484161377,
      "learning_rate": 3.502635009823904e-05,
      "loss": 1.3428,
      "step": 22360
    },
    {
      "epoch": 3.6967568684156165,
      "grad_norm": 9.341504096984863,
      "learning_rate": 3.5017168879340424e-05,
      "loss": 1.5852,
      "step": 22370
    },
    {
      "epoch": 3.698409419541417,
      "grad_norm": 9.287238121032715,
      "learning_rate": 3.5007987660441806e-05,
      "loss": 1.4823,
      "step": 22380
    },
    {
      "epoch": 3.7000619706672175,
      "grad_norm": 11.68283462524414,
      "learning_rate": 3.499880644154318e-05,
      "loss": 1.5553,
      "step": 22390
    },
    {
      "epoch": 3.701714521793018,
      "grad_norm": 9.870018005371094,
      "learning_rate": 3.4989625222644565e-05,
      "loss": 1.5829,
      "step": 22400
    },
    {
      "epoch": 3.7033670729188186,
      "grad_norm": 41.26398849487305,
      "learning_rate": 3.498044400374594e-05,
      "loss": 1.4629,
      "step": 22410
    },
    {
      "epoch": 3.7050196240446187,
      "grad_norm": 13.389437675476074,
      "learning_rate": 3.4971262784847316e-05,
      "loss": 1.6009,
      "step": 22420
    },
    {
      "epoch": 3.7066721751704192,
      "grad_norm": 8.6530179977417,
      "learning_rate": 3.496208156594869e-05,
      "loss": 1.6486,
      "step": 22430
    },
    {
      "epoch": 3.70832472629622,
      "grad_norm": 8.932863235473633,
      "learning_rate": 3.4952900347050075e-05,
      "loss": 1.6876,
      "step": 22440
    },
    {
      "epoch": 3.7099772774220203,
      "grad_norm": 14.32821273803711,
      "learning_rate": 3.494371912815145e-05,
      "loss": 1.4702,
      "step": 22450
    },
    {
      "epoch": 3.711629828547821,
      "grad_norm": 10.528609275817871,
      "learning_rate": 3.4934537909252833e-05,
      "loss": 1.4973,
      "step": 22460
    },
    {
      "epoch": 3.713282379673621,
      "grad_norm": 13.962540626525879,
      "learning_rate": 3.492535669035421e-05,
      "loss": 1.6201,
      "step": 22470
    },
    {
      "epoch": 3.7149349307994215,
      "grad_norm": 15.784793853759766,
      "learning_rate": 3.491617547145559e-05,
      "loss": 1.6017,
      "step": 22480
    },
    {
      "epoch": 3.716587481925222,
      "grad_norm": 9.065518379211426,
      "learning_rate": 3.4906994252556975e-05,
      "loss": 1.5692,
      "step": 22490
    },
    {
      "epoch": 3.7182400330510226,
      "grad_norm": 10.009034156799316,
      "learning_rate": 3.489781303365835e-05,
      "loss": 1.5791,
      "step": 22500
    },
    {
      "epoch": 3.719892584176823,
      "grad_norm": 8.641139030456543,
      "learning_rate": 3.488863181475973e-05,
      "loss": 1.5955,
      "step": 22510
    },
    {
      "epoch": 3.721545135302623,
      "grad_norm": 10.301948547363281,
      "learning_rate": 3.487945059586111e-05,
      "loss": 1.5108,
      "step": 22520
    },
    {
      "epoch": 3.7231976864284237,
      "grad_norm": 28.2911319732666,
      "learning_rate": 3.487026937696249e-05,
      "loss": 1.5386,
      "step": 22530
    },
    {
      "epoch": 3.7248502375542243,
      "grad_norm": 129.94375610351562,
      "learning_rate": 3.486108815806387e-05,
      "loss": 1.7079,
      "step": 22540
    },
    {
      "epoch": 3.726502788680025,
      "grad_norm": 23.256444931030273,
      "learning_rate": 3.485190693916524e-05,
      "loss": 1.5878,
      "step": 22550
    },
    {
      "epoch": 3.7281553398058254,
      "grad_norm": 7.868270397186279,
      "learning_rate": 3.484272572026662e-05,
      "loss": 1.536,
      "step": 22560
    },
    {
      "epoch": 3.7298078909316255,
      "grad_norm": 14.722084045410156,
      "learning_rate": 3.4833544501368e-05,
      "loss": 1.4831,
      "step": 22570
    },
    {
      "epoch": 3.731460442057426,
      "grad_norm": 19.196247100830078,
      "learning_rate": 3.482436328246938e-05,
      "loss": 1.4927,
      "step": 22580
    },
    {
      "epoch": 3.7331129931832265,
      "grad_norm": 7.337265968322754,
      "learning_rate": 3.481518206357076e-05,
      "loss": 1.6621,
      "step": 22590
    },
    {
      "epoch": 3.734765544309027,
      "grad_norm": 21.33555793762207,
      "learning_rate": 3.480600084467214e-05,
      "loss": 1.4837,
      "step": 22600
    },
    {
      "epoch": 3.7364180954348276,
      "grad_norm": 33.45065689086914,
      "learning_rate": 3.479681962577352e-05,
      "loss": 1.5462,
      "step": 22610
    },
    {
      "epoch": 3.7380706465606277,
      "grad_norm": 13.384827613830566,
      "learning_rate": 3.47876384068749e-05,
      "loss": 1.5441,
      "step": 22620
    },
    {
      "epoch": 3.7397231976864287,
      "grad_norm": 75.69512939453125,
      "learning_rate": 3.477845718797628e-05,
      "loss": 1.4274,
      "step": 22630
    },
    {
      "epoch": 3.741375748812229,
      "grad_norm": 14.372223854064941,
      "learning_rate": 3.476927596907766e-05,
      "loss": 1.6275,
      "step": 22640
    },
    {
      "epoch": 3.7430282999380293,
      "grad_norm": 8.416259765625,
      "learning_rate": 3.4760094750179036e-05,
      "loss": 1.5561,
      "step": 22650
    },
    {
      "epoch": 3.74468085106383,
      "grad_norm": 11.804091453552246,
      "learning_rate": 3.475091353128042e-05,
      "loss": 1.6549,
      "step": 22660
    },
    {
      "epoch": 3.7463334021896304,
      "grad_norm": 11.802560806274414,
      "learning_rate": 3.4741732312381794e-05,
      "loss": 1.5372,
      "step": 22670
    },
    {
      "epoch": 3.747985953315431,
      "grad_norm": 8.755041122436523,
      "learning_rate": 3.473255109348317e-05,
      "loss": 1.6324,
      "step": 22680
    },
    {
      "epoch": 3.749638504441231,
      "grad_norm": 6.660466194152832,
      "learning_rate": 3.4723369874584546e-05,
      "loss": 1.537,
      "step": 22690
    },
    {
      "epoch": 3.7512910555670316,
      "grad_norm": 19.879173278808594,
      "learning_rate": 3.471418865568593e-05,
      "loss": 1.474,
      "step": 22700
    },
    {
      "epoch": 3.752943606692832,
      "grad_norm": 4.748521327972412,
      "learning_rate": 3.4705007436787304e-05,
      "loss": 1.5009,
      "step": 22710
    },
    {
      "epoch": 3.7545961578186327,
      "grad_norm": 30.82050895690918,
      "learning_rate": 3.469582621788869e-05,
      "loss": 1.5368,
      "step": 22720
    },
    {
      "epoch": 3.756248708944433,
      "grad_norm": 28.105743408203125,
      "learning_rate": 3.468664499899007e-05,
      "loss": 1.359,
      "step": 22730
    },
    {
      "epoch": 3.7579012600702333,
      "grad_norm": 20.436500549316406,
      "learning_rate": 3.4677463780091446e-05,
      "loss": 1.4738,
      "step": 22740
    },
    {
      "epoch": 3.759553811196034,
      "grad_norm": 18.063720703125,
      "learning_rate": 3.466828256119283e-05,
      "loss": 1.6101,
      "step": 22750
    },
    {
      "epoch": 3.7612063623218344,
      "grad_norm": 11.609806060791016,
      "learning_rate": 3.4659101342294204e-05,
      "loss": 1.4377,
      "step": 22760
    },
    {
      "epoch": 3.762858913447635,
      "grad_norm": 7.862545490264893,
      "learning_rate": 3.464992012339559e-05,
      "loss": 1.5208,
      "step": 22770
    },
    {
      "epoch": 3.7645114645734354,
      "grad_norm": 13.573543548583984,
      "learning_rate": 3.464073890449696e-05,
      "loss": 1.6314,
      "step": 22780
    },
    {
      "epoch": 3.7661640156992355,
      "grad_norm": 6.686908721923828,
      "learning_rate": 3.4631557685598345e-05,
      "loss": 1.5347,
      "step": 22790
    },
    {
      "epoch": 3.767816566825036,
      "grad_norm": 16.876447677612305,
      "learning_rate": 3.462237646669972e-05,
      "loss": 1.3998,
      "step": 22800
    },
    {
      "epoch": 3.7694691179508366,
      "grad_norm": 5.551175117492676,
      "learning_rate": 3.46131952478011e-05,
      "loss": 1.6144,
      "step": 22810
    },
    {
      "epoch": 3.771121669076637,
      "grad_norm": 12.1668701171875,
      "learning_rate": 3.460401402890247e-05,
      "loss": 1.5202,
      "step": 22820
    },
    {
      "epoch": 3.7727742202024377,
      "grad_norm": 7.639707565307617,
      "learning_rate": 3.4594832810003855e-05,
      "loss": 1.5545,
      "step": 22830
    },
    {
      "epoch": 3.774426771328238,
      "grad_norm": 27.100339889526367,
      "learning_rate": 3.458565159110524e-05,
      "loss": 1.5098,
      "step": 22840
    },
    {
      "epoch": 3.7760793224540383,
      "grad_norm": 6.334820747375488,
      "learning_rate": 3.4576470372206614e-05,
      "loss": 1.4167,
      "step": 22850
    },
    {
      "epoch": 3.777731873579839,
      "grad_norm": 56.109954833984375,
      "learning_rate": 3.4567289153307997e-05,
      "loss": 1.5223,
      "step": 22860
    },
    {
      "epoch": 3.7793844247056394,
      "grad_norm": 9.93230152130127,
      "learning_rate": 3.455810793440937e-05,
      "loss": 1.5619,
      "step": 22870
    },
    {
      "epoch": 3.78103697583144,
      "grad_norm": 9.03853702545166,
      "learning_rate": 3.4548926715510755e-05,
      "loss": 1.468,
      "step": 22880
    },
    {
      "epoch": 3.78268952695724,
      "grad_norm": 11.272610664367676,
      "learning_rate": 3.453974549661213e-05,
      "loss": 1.4569,
      "step": 22890
    },
    {
      "epoch": 3.7843420780830406,
      "grad_norm": 11.486401557922363,
      "learning_rate": 3.4530564277713514e-05,
      "loss": 1.5401,
      "step": 22900
    },
    {
      "epoch": 3.785994629208841,
      "grad_norm": 12.329487800598145,
      "learning_rate": 3.452138305881489e-05,
      "loss": 1.5879,
      "step": 22910
    },
    {
      "epoch": 3.7876471803346417,
      "grad_norm": 13.06454849243164,
      "learning_rate": 3.451220183991627e-05,
      "loss": 1.5854,
      "step": 22920
    },
    {
      "epoch": 3.789299731460442,
      "grad_norm": 17.574142456054688,
      "learning_rate": 3.450302062101765e-05,
      "loss": 1.5392,
      "step": 22930
    },
    {
      "epoch": 3.7909522825862423,
      "grad_norm": 17.874441146850586,
      "learning_rate": 3.4493839402119024e-05,
      "loss": 1.542,
      "step": 22940
    },
    {
      "epoch": 3.792604833712043,
      "grad_norm": 8.613436698913574,
      "learning_rate": 3.4484658183220406e-05,
      "loss": 1.5781,
      "step": 22950
    },
    {
      "epoch": 3.7942573848378434,
      "grad_norm": 9.75782585144043,
      "learning_rate": 3.447547696432178e-05,
      "loss": 1.5628,
      "step": 22960
    },
    {
      "epoch": 3.795909935963644,
      "grad_norm": 13.988187789916992,
      "learning_rate": 3.4466295745423165e-05,
      "loss": 1.5528,
      "step": 22970
    },
    {
      "epoch": 3.7975624870894444,
      "grad_norm": 15.687790870666504,
      "learning_rate": 3.445711452652454e-05,
      "loss": 1.555,
      "step": 22980
    },
    {
      "epoch": 3.7992150382152445,
      "grad_norm": 10.992048263549805,
      "learning_rate": 3.4447933307625923e-05,
      "loss": 1.7061,
      "step": 22990
    },
    {
      "epoch": 3.8008675893410455,
      "grad_norm": 10.867881774902344,
      "learning_rate": 3.44387520887273e-05,
      "loss": 1.6758,
      "step": 23000
    },
    {
      "epoch": 3.8025201404668456,
      "grad_norm": 10.184456825256348,
      "learning_rate": 3.442957086982868e-05,
      "loss": 1.5848,
      "step": 23010
    },
    {
      "epoch": 3.804172691592646,
      "grad_norm": 17.802675247192383,
      "learning_rate": 3.442038965093006e-05,
      "loss": 1.592,
      "step": 23020
    },
    {
      "epoch": 3.8058252427184467,
      "grad_norm": 12.011911392211914,
      "learning_rate": 3.441120843203144e-05,
      "loss": 1.4794,
      "step": 23030
    },
    {
      "epoch": 3.807477793844247,
      "grad_norm": 8.725467681884766,
      "learning_rate": 3.4402027213132816e-05,
      "loss": 1.5482,
      "step": 23040
    },
    {
      "epoch": 3.8091303449700478,
      "grad_norm": 10.62525749206543,
      "learning_rate": 3.43928459942342e-05,
      "loss": 1.4878,
      "step": 23050
    },
    {
      "epoch": 3.810782896095848,
      "grad_norm": 10.504483222961426,
      "learning_rate": 3.4383664775335575e-05,
      "loss": 1.4817,
      "step": 23060
    },
    {
      "epoch": 3.8124354472216484,
      "grad_norm": 12.407063484191895,
      "learning_rate": 3.437448355643695e-05,
      "loss": 1.5458,
      "step": 23070
    },
    {
      "epoch": 3.814087998347449,
      "grad_norm": 17.53013038635254,
      "learning_rate": 3.436530233753833e-05,
      "loss": 1.5761,
      "step": 23080
    },
    {
      "epoch": 3.8157405494732495,
      "grad_norm": 16.574785232543945,
      "learning_rate": 3.435612111863971e-05,
      "loss": 1.5797,
      "step": 23090
    },
    {
      "epoch": 3.81739310059905,
      "grad_norm": 11.763534545898438,
      "learning_rate": 3.434693989974109e-05,
      "loss": 1.5059,
      "step": 23100
    },
    {
      "epoch": 3.81904565172485,
      "grad_norm": 11.488256454467773,
      "learning_rate": 3.433775868084247e-05,
      "loss": 1.6877,
      "step": 23110
    },
    {
      "epoch": 3.8206982028506506,
      "grad_norm": 8.004593849182129,
      "learning_rate": 3.432857746194385e-05,
      "loss": 1.491,
      "step": 23120
    },
    {
      "epoch": 3.822350753976451,
      "grad_norm": 8.863856315612793,
      "learning_rate": 3.4319396243045226e-05,
      "loss": 1.5296,
      "step": 23130
    },
    {
      "epoch": 3.8240033051022517,
      "grad_norm": 9.865023612976074,
      "learning_rate": 3.431021502414661e-05,
      "loss": 1.5924,
      "step": 23140
    },
    {
      "epoch": 3.8256558562280523,
      "grad_norm": 5.761257171630859,
      "learning_rate": 3.4301033805247985e-05,
      "loss": 1.4003,
      "step": 23150
    },
    {
      "epoch": 3.8273084073538524,
      "grad_norm": 14.028019905090332,
      "learning_rate": 3.429185258634937e-05,
      "loss": 1.5787,
      "step": 23160
    },
    {
      "epoch": 3.828960958479653,
      "grad_norm": 11.172537803649902,
      "learning_rate": 3.428267136745075e-05,
      "loss": 1.5284,
      "step": 23170
    },
    {
      "epoch": 3.8306135096054534,
      "grad_norm": 8.914949417114258,
      "learning_rate": 3.4273490148552126e-05,
      "loss": 1.5841,
      "step": 23180
    },
    {
      "epoch": 3.832266060731254,
      "grad_norm": 11.068294525146484,
      "learning_rate": 3.42643089296535e-05,
      "loss": 1.5591,
      "step": 23190
    },
    {
      "epoch": 3.8339186118570545,
      "grad_norm": 7.558146953582764,
      "learning_rate": 3.425512771075488e-05,
      "loss": 1.6917,
      "step": 23200
    },
    {
      "epoch": 3.8355711629828546,
      "grad_norm": 11.731191635131836,
      "learning_rate": 3.424594649185626e-05,
      "loss": 1.5366,
      "step": 23210
    },
    {
      "epoch": 3.837223714108655,
      "grad_norm": 19.394287109375,
      "learning_rate": 3.4236765272957636e-05,
      "loss": 1.5022,
      "step": 23220
    },
    {
      "epoch": 3.8388762652344557,
      "grad_norm": 27.054391860961914,
      "learning_rate": 3.422758405405902e-05,
      "loss": 1.538,
      "step": 23230
    },
    {
      "epoch": 3.8405288163602562,
      "grad_norm": 15.95034408569336,
      "learning_rate": 3.4218402835160394e-05,
      "loss": 1.6006,
      "step": 23240
    },
    {
      "epoch": 3.8421813674860568,
      "grad_norm": 8.388917922973633,
      "learning_rate": 3.420922161626178e-05,
      "loss": 1.4591,
      "step": 23250
    },
    {
      "epoch": 3.843833918611857,
      "grad_norm": 11.338773727416992,
      "learning_rate": 3.420004039736315e-05,
      "loss": 1.6336,
      "step": 23260
    },
    {
      "epoch": 3.8454864697376574,
      "grad_norm": 12.91083812713623,
      "learning_rate": 3.4190859178464536e-05,
      "loss": 1.6015,
      "step": 23270
    },
    {
      "epoch": 3.847139020863458,
      "grad_norm": 10.893755912780762,
      "learning_rate": 3.418167795956591e-05,
      "loss": 1.5369,
      "step": 23280
    },
    {
      "epoch": 3.8487915719892585,
      "grad_norm": 10.60979175567627,
      "learning_rate": 3.4172496740667294e-05,
      "loss": 1.591,
      "step": 23290
    },
    {
      "epoch": 3.850444123115059,
      "grad_norm": 9.582144737243652,
      "learning_rate": 3.416331552176868e-05,
      "loss": 1.6382,
      "step": 23300
    },
    {
      "epoch": 3.852096674240859,
      "grad_norm": 14.1846923828125,
      "learning_rate": 3.415413430287005e-05,
      "loss": 1.5981,
      "step": 23310
    },
    {
      "epoch": 3.8537492253666596,
      "grad_norm": 10.367266654968262,
      "learning_rate": 3.414495308397143e-05,
      "loss": 1.5864,
      "step": 23320
    },
    {
      "epoch": 3.85540177649246,
      "grad_norm": 9.663958549499512,
      "learning_rate": 3.4135771865072804e-05,
      "loss": 1.544,
      "step": 23330
    },
    {
      "epoch": 3.8570543276182607,
      "grad_norm": 13.666308403015137,
      "learning_rate": 3.412659064617419e-05,
      "loss": 1.4994,
      "step": 23340
    },
    {
      "epoch": 3.8587068787440613,
      "grad_norm": 9.828655242919922,
      "learning_rate": 3.411740942727556e-05,
      "loss": 1.5709,
      "step": 23350
    },
    {
      "epoch": 3.8603594298698614,
      "grad_norm": 9.607194900512695,
      "learning_rate": 3.4108228208376945e-05,
      "loss": 1.54,
      "step": 23360
    },
    {
      "epoch": 3.862011980995662,
      "grad_norm": 12.873283386230469,
      "learning_rate": 3.409904698947832e-05,
      "loss": 1.3461,
      "step": 23370
    },
    {
      "epoch": 3.8636645321214624,
      "grad_norm": 8.448955535888672,
      "learning_rate": 3.4089865770579704e-05,
      "loss": 1.5732,
      "step": 23380
    },
    {
      "epoch": 3.865317083247263,
      "grad_norm": 36.90695571899414,
      "learning_rate": 3.408068455168108e-05,
      "loss": 1.539,
      "step": 23390
    },
    {
      "epoch": 3.8669696343730635,
      "grad_norm": 14.371119499206543,
      "learning_rate": 3.407150333278246e-05,
      "loss": 1.542,
      "step": 23400
    },
    {
      "epoch": 3.8686221854988636,
      "grad_norm": 6.606020450592041,
      "learning_rate": 3.4062322113883845e-05,
      "loss": 1.412,
      "step": 23410
    },
    {
      "epoch": 3.8702747366246646,
      "grad_norm": 7.5736083984375,
      "learning_rate": 3.405314089498522e-05,
      "loss": 1.6222,
      "step": 23420
    },
    {
      "epoch": 3.8719272877504647,
      "grad_norm": 184.7552490234375,
      "learning_rate": 3.4043959676086604e-05,
      "loss": 1.5178,
      "step": 23430
    },
    {
      "epoch": 3.8735798388762652,
      "grad_norm": 7.478529930114746,
      "learning_rate": 3.403477845718798e-05,
      "loss": 1.6738,
      "step": 23440
    },
    {
      "epoch": 3.8752323900020658,
      "grad_norm": 14.52708911895752,
      "learning_rate": 3.4025597238289355e-05,
      "loss": 1.6449,
      "step": 23450
    },
    {
      "epoch": 3.8768849411278663,
      "grad_norm": 8.332096099853516,
      "learning_rate": 3.401641601939073e-05,
      "loss": 1.5679,
      "step": 23460
    },
    {
      "epoch": 3.878537492253667,
      "grad_norm": 9.562748908996582,
      "learning_rate": 3.4007234800492114e-05,
      "loss": 1.5742,
      "step": 23470
    },
    {
      "epoch": 3.880190043379467,
      "grad_norm": 7.013890743255615,
      "learning_rate": 3.399805358159349e-05,
      "loss": 1.6466,
      "step": 23480
    },
    {
      "epoch": 3.8818425945052675,
      "grad_norm": 16.500028610229492,
      "learning_rate": 3.398887236269487e-05,
      "loss": 1.6899,
      "step": 23490
    },
    {
      "epoch": 3.883495145631068,
      "grad_norm": 8.28381633758545,
      "learning_rate": 3.397969114379625e-05,
      "loss": 1.5692,
      "step": 23500
    },
    {
      "epoch": 3.8851476967568686,
      "grad_norm": 35.62749099731445,
      "learning_rate": 3.397050992489763e-05,
      "loss": 1.4717,
      "step": 23510
    },
    {
      "epoch": 3.886800247882669,
      "grad_norm": 8.142434120178223,
      "learning_rate": 3.396132870599901e-05,
      "loss": 1.4372,
      "step": 23520
    },
    {
      "epoch": 3.888452799008469,
      "grad_norm": 13.488250732421875,
      "learning_rate": 3.395214748710039e-05,
      "loss": 1.6781,
      "step": 23530
    },
    {
      "epoch": 3.8901053501342697,
      "grad_norm": 8.421558380126953,
      "learning_rate": 3.394296626820177e-05,
      "loss": 1.4748,
      "step": 23540
    },
    {
      "epoch": 3.8917579012600703,
      "grad_norm": 14.573169708251953,
      "learning_rate": 3.393378504930315e-05,
      "loss": 1.5179,
      "step": 23550
    },
    {
      "epoch": 3.893410452385871,
      "grad_norm": 8.99917984008789,
      "learning_rate": 3.392460383040453e-05,
      "loss": 1.5438,
      "step": 23560
    },
    {
      "epoch": 3.8950630035116713,
      "grad_norm": 16.3277645111084,
      "learning_rate": 3.3915422611505906e-05,
      "loss": 1.5428,
      "step": 23570
    },
    {
      "epoch": 3.8967155546374714,
      "grad_norm": 7.9492878913879395,
      "learning_rate": 3.390624139260728e-05,
      "loss": 1.6256,
      "step": 23580
    },
    {
      "epoch": 3.898368105763272,
      "grad_norm": 7.273673057556152,
      "learning_rate": 3.389706017370866e-05,
      "loss": 1.6642,
      "step": 23590
    },
    {
      "epoch": 3.9000206568890725,
      "grad_norm": 8.6654634475708,
      "learning_rate": 3.388787895481004e-05,
      "loss": 1.5932,
      "step": 23600
    },
    {
      "epoch": 3.901673208014873,
      "grad_norm": 10.21167278289795,
      "learning_rate": 3.3878697735911416e-05,
      "loss": 1.5356,
      "step": 23610
    },
    {
      "epoch": 3.9033257591406736,
      "grad_norm": 16.551963806152344,
      "learning_rate": 3.38695165170128e-05,
      "loss": 1.6243,
      "step": 23620
    },
    {
      "epoch": 3.9049783102664737,
      "grad_norm": 28.581851959228516,
      "learning_rate": 3.386033529811418e-05,
      "loss": 1.6437,
      "step": 23630
    },
    {
      "epoch": 3.9066308613922742,
      "grad_norm": 22.335962295532227,
      "learning_rate": 3.385115407921556e-05,
      "loss": 1.4263,
      "step": 23640
    },
    {
      "epoch": 3.9082834125180748,
      "grad_norm": 7.5155158042907715,
      "learning_rate": 3.384197286031694e-05,
      "loss": 1.4715,
      "step": 23650
    },
    {
      "epoch": 3.9099359636438753,
      "grad_norm": 12.375370979309082,
      "learning_rate": 3.3832791641418316e-05,
      "loss": 1.5336,
      "step": 23660
    },
    {
      "epoch": 3.911588514769676,
      "grad_norm": 26.960378646850586,
      "learning_rate": 3.38236104225197e-05,
      "loss": 1.5445,
      "step": 23670
    },
    {
      "epoch": 3.913241065895476,
      "grad_norm": 17.1274471282959,
      "learning_rate": 3.3814429203621075e-05,
      "loss": 1.6372,
      "step": 23680
    },
    {
      "epoch": 3.9148936170212765,
      "grad_norm": 8.658958435058594,
      "learning_rate": 3.380524798472246e-05,
      "loss": 1.5278,
      "step": 23690
    },
    {
      "epoch": 3.916546168147077,
      "grad_norm": 8.599283218383789,
      "learning_rate": 3.379606676582383e-05,
      "loss": 1.5516,
      "step": 23700
    },
    {
      "epoch": 3.9181987192728776,
      "grad_norm": 17.065500259399414,
      "learning_rate": 3.378688554692521e-05,
      "loss": 1.5585,
      "step": 23710
    },
    {
      "epoch": 3.919851270398678,
      "grad_norm": 6.110587120056152,
      "learning_rate": 3.3777704328026585e-05,
      "loss": 1.4451,
      "step": 23720
    },
    {
      "epoch": 3.921503821524478,
      "grad_norm": 11.20459270477295,
      "learning_rate": 3.376852310912797e-05,
      "loss": 1.5366,
      "step": 23730
    },
    {
      "epoch": 3.9231563726502787,
      "grad_norm": 7.95195198059082,
      "learning_rate": 3.375934189022935e-05,
      "loss": 1.3379,
      "step": 23740
    },
    {
      "epoch": 3.9248089237760793,
      "grad_norm": 17.207660675048828,
      "learning_rate": 3.3750160671330726e-05,
      "loss": 1.5098,
      "step": 23750
    },
    {
      "epoch": 3.92646147490188,
      "grad_norm": 18.350038528442383,
      "learning_rate": 3.374097945243211e-05,
      "loss": 1.421,
      "step": 23760
    },
    {
      "epoch": 3.9281140260276803,
      "grad_norm": 21.70857048034668,
      "learning_rate": 3.3731798233533484e-05,
      "loss": 1.5718,
      "step": 23770
    },
    {
      "epoch": 3.9297665771534804,
      "grad_norm": 12.964973449707031,
      "learning_rate": 3.372261701463487e-05,
      "loss": 1.5394,
      "step": 23780
    },
    {
      "epoch": 3.9314191282792814,
      "grad_norm": 7.696381092071533,
      "learning_rate": 3.371343579573624e-05,
      "loss": 1.392,
      "step": 23790
    },
    {
      "epoch": 3.9330716794050815,
      "grad_norm": 8.877033233642578,
      "learning_rate": 3.3704254576837626e-05,
      "loss": 1.6341,
      "step": 23800
    },
    {
      "epoch": 3.934724230530882,
      "grad_norm": 6.651622772216797,
      "learning_rate": 3.3695073357939e-05,
      "loss": 1.3865,
      "step": 23810
    },
    {
      "epoch": 3.9363767816566826,
      "grad_norm": 13.928619384765625,
      "learning_rate": 3.3685892139040384e-05,
      "loss": 1.5814,
      "step": 23820
    },
    {
      "epoch": 3.9380293327824827,
      "grad_norm": 14.732012748718262,
      "learning_rate": 3.367671092014176e-05,
      "loss": 1.4712,
      "step": 23830
    },
    {
      "epoch": 3.9396818839082837,
      "grad_norm": 14.067058563232422,
      "learning_rate": 3.3667529701243136e-05,
      "loss": 1.5988,
      "step": 23840
    },
    {
      "epoch": 3.9413344350340838,
      "grad_norm": 10.48046875,
      "learning_rate": 3.365834848234452e-05,
      "loss": 1.6534,
      "step": 23850
    },
    {
      "epoch": 3.9429869861598843,
      "grad_norm": 25.77431297302246,
      "learning_rate": 3.3649167263445894e-05,
      "loss": 1.4357,
      "step": 23860
    },
    {
      "epoch": 3.944639537285685,
      "grad_norm": 39.13468551635742,
      "learning_rate": 3.363998604454728e-05,
      "loss": 1.5746,
      "step": 23870
    },
    {
      "epoch": 3.9462920884114854,
      "grad_norm": 12.897835731506348,
      "learning_rate": 3.363080482564865e-05,
      "loss": 1.435,
      "step": 23880
    },
    {
      "epoch": 3.947944639537286,
      "grad_norm": 18.15728187561035,
      "learning_rate": 3.3621623606750035e-05,
      "loss": 1.5303,
      "step": 23890
    },
    {
      "epoch": 3.949597190663086,
      "grad_norm": 20.04810905456543,
      "learning_rate": 3.361244238785141e-05,
      "loss": 1.5076,
      "step": 23900
    },
    {
      "epoch": 3.9512497417888865,
      "grad_norm": 6.630679607391357,
      "learning_rate": 3.3603261168952794e-05,
      "loss": 1.4879,
      "step": 23910
    },
    {
      "epoch": 3.952902292914687,
      "grad_norm": 30.30918312072754,
      "learning_rate": 3.359407995005417e-05,
      "loss": 1.5553,
      "step": 23920
    },
    {
      "epoch": 3.9545548440404876,
      "grad_norm": 7.379817008972168,
      "learning_rate": 3.358489873115555e-05,
      "loss": 1.5888,
      "step": 23930
    },
    {
      "epoch": 3.956207395166288,
      "grad_norm": 8.264845848083496,
      "learning_rate": 3.357571751225693e-05,
      "loss": 1.637,
      "step": 23940
    },
    {
      "epoch": 3.9578599462920883,
      "grad_norm": 14.769917488098145,
      "learning_rate": 3.356653629335831e-05,
      "loss": 1.5499,
      "step": 23950
    },
    {
      "epoch": 3.959512497417889,
      "grad_norm": 6.285022735595703,
      "learning_rate": 3.355735507445969e-05,
      "loss": 1.6016,
      "step": 23960
    },
    {
      "epoch": 3.9611650485436893,
      "grad_norm": 8.565293312072754,
      "learning_rate": 3.354817385556106e-05,
      "loss": 1.4791,
      "step": 23970
    },
    {
      "epoch": 3.96281759966949,
      "grad_norm": 13.735636711120605,
      "learning_rate": 3.3538992636662445e-05,
      "loss": 1.4269,
      "step": 23980
    },
    {
      "epoch": 3.9644701507952904,
      "grad_norm": 6.193691253662109,
      "learning_rate": 3.352981141776382e-05,
      "loss": 1.4687,
      "step": 23990
    },
    {
      "epoch": 3.9661227019210905,
      "grad_norm": 9.597599029541016,
      "learning_rate": 3.3520630198865204e-05,
      "loss": 1.4732,
      "step": 24000
    },
    {
      "epoch": 3.967775253046891,
      "grad_norm": 12.805542945861816,
      "learning_rate": 3.351144897996658e-05,
      "loss": 1.7671,
      "step": 24010
    },
    {
      "epoch": 3.9694278041726916,
      "grad_norm": 11.650763511657715,
      "learning_rate": 3.350226776106796e-05,
      "loss": 1.4774,
      "step": 24020
    },
    {
      "epoch": 3.971080355298492,
      "grad_norm": 32.57954406738281,
      "learning_rate": 3.349308654216934e-05,
      "loss": 1.6312,
      "step": 24030
    },
    {
      "epoch": 3.9727329064242927,
      "grad_norm": 10.600648880004883,
      "learning_rate": 3.348390532327072e-05,
      "loss": 1.4607,
      "step": 24040
    },
    {
      "epoch": 3.9743854575500928,
      "grad_norm": 7.16843843460083,
      "learning_rate": 3.3474724104372097e-05,
      "loss": 1.5808,
      "step": 24050
    },
    {
      "epoch": 3.9760380086758933,
      "grad_norm": 60.35005187988281,
      "learning_rate": 3.346554288547348e-05,
      "loss": 1.4766,
      "step": 24060
    },
    {
      "epoch": 3.977690559801694,
      "grad_norm": 8.780489921569824,
      "learning_rate": 3.3456361666574855e-05,
      "loss": 1.5446,
      "step": 24070
    },
    {
      "epoch": 3.9793431109274944,
      "grad_norm": 17.84938621520996,
      "learning_rate": 3.344718044767624e-05,
      "loss": 1.569,
      "step": 24080
    },
    {
      "epoch": 3.980995662053295,
      "grad_norm": 10.445038795471191,
      "learning_rate": 3.3437999228777614e-05,
      "loss": 1.6016,
      "step": 24090
    },
    {
      "epoch": 3.982648213179095,
      "grad_norm": 6.451947212219238,
      "learning_rate": 3.342881800987899e-05,
      "loss": 1.5652,
      "step": 24100
    },
    {
      "epoch": 3.9843007643048955,
      "grad_norm": 13.274555206298828,
      "learning_rate": 3.341963679098037e-05,
      "loss": 1.3444,
      "step": 24110
    },
    {
      "epoch": 3.985953315430696,
      "grad_norm": 92.55748748779297,
      "learning_rate": 3.341045557208175e-05,
      "loss": 1.5804,
      "step": 24120
    },
    {
      "epoch": 3.9876058665564966,
      "grad_norm": 6.433228015899658,
      "learning_rate": 3.340127435318313e-05,
      "loss": 1.5241,
      "step": 24130
    },
    {
      "epoch": 3.989258417682297,
      "grad_norm": 7.102395534515381,
      "learning_rate": 3.3392093134284506e-05,
      "loss": 1.4866,
      "step": 24140
    },
    {
      "epoch": 3.9909109688080973,
      "grad_norm": 7.430931568145752,
      "learning_rate": 3.338291191538589e-05,
      "loss": 1.5228,
      "step": 24150
    },
    {
      "epoch": 3.992563519933898,
      "grad_norm": 7.019854545593262,
      "learning_rate": 3.3373730696487265e-05,
      "loss": 1.6071,
      "step": 24160
    },
    {
      "epoch": 3.9942160710596983,
      "grad_norm": 5.7070512771606445,
      "learning_rate": 3.336454947758865e-05,
      "loss": 1.4218,
      "step": 24170
    },
    {
      "epoch": 3.995868622185499,
      "grad_norm": 15.087320327758789,
      "learning_rate": 3.335536825869002e-05,
      "loss": 1.4703,
      "step": 24180
    },
    {
      "epoch": 3.9975211733112994,
      "grad_norm": 10.09599781036377,
      "learning_rate": 3.3346187039791406e-05,
      "loss": 1.5294,
      "step": 24190
    },
    {
      "epoch": 3.9991737244370995,
      "grad_norm": 9.25766658782959,
      "learning_rate": 3.333700582089279e-05,
      "loss": 1.5693,
      "step": 24200
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.3066392396084551,
      "eval_loss": 2.2003278732299805,
      "eval_runtime": 814.2753,
      "eval_samples_per_second": 34.627,
      "eval_steps_per_second": 8.657,
      "step": 24205
    },
    {
      "epoch": 4.0008262755629005,
      "grad_norm": 6.969991207122803,
      "learning_rate": 3.3327824601994165e-05,
      "loss": 1.5264,
      "step": 24210
    },
    {
      "epoch": 4.002478826688701,
      "grad_norm": 6.718509674072266,
      "learning_rate": 3.331864338309554e-05,
      "loss": 1.5123,
      "step": 24220
    },
    {
      "epoch": 4.004131377814502,
      "grad_norm": 13.605385780334473,
      "learning_rate": 3.3309462164196916e-05,
      "loss": 1.5449,
      "step": 24230
    },
    {
      "epoch": 4.005783928940302,
      "grad_norm": 8.160221099853516,
      "learning_rate": 3.33002809452983e-05,
      "loss": 1.6421,
      "step": 24240
    },
    {
      "epoch": 4.007436480066102,
      "grad_norm": 6.459905624389648,
      "learning_rate": 3.3291099726399675e-05,
      "loss": 1.6331,
      "step": 24250
    },
    {
      "epoch": 4.009089031191903,
      "grad_norm": 11.185013771057129,
      "learning_rate": 3.328191850750106e-05,
      "loss": 1.4834,
      "step": 24260
    },
    {
      "epoch": 4.010741582317703,
      "grad_norm": 11.707094192504883,
      "learning_rate": 3.327273728860243e-05,
      "loss": 1.5363,
      "step": 24270
    },
    {
      "epoch": 4.012394133443504,
      "grad_norm": 44.96635055541992,
      "learning_rate": 3.3263556069703816e-05,
      "loss": 1.4972,
      "step": 24280
    },
    {
      "epoch": 4.014046684569304,
      "grad_norm": 15.029733657836914,
      "learning_rate": 3.325437485080519e-05,
      "loss": 1.575,
      "step": 24290
    },
    {
      "epoch": 4.015699235695104,
      "grad_norm": 8.984262466430664,
      "learning_rate": 3.3245193631906574e-05,
      "loss": 1.6195,
      "step": 24300
    },
    {
      "epoch": 4.017351786820905,
      "grad_norm": 6.634871482849121,
      "learning_rate": 3.323601241300795e-05,
      "loss": 1.5722,
      "step": 24310
    },
    {
      "epoch": 4.019004337946705,
      "grad_norm": 13.543932914733887,
      "learning_rate": 3.322683119410933e-05,
      "loss": 1.6134,
      "step": 24320
    },
    {
      "epoch": 4.020656889072506,
      "grad_norm": 7.46361780166626,
      "learning_rate": 3.3217649975210715e-05,
      "loss": 1.41,
      "step": 24330
    },
    {
      "epoch": 4.022309440198306,
      "grad_norm": 16.312719345092773,
      "learning_rate": 3.320846875631209e-05,
      "loss": 1.4198,
      "step": 24340
    },
    {
      "epoch": 4.023961991324106,
      "grad_norm": 16.578506469726562,
      "learning_rate": 3.319928753741347e-05,
      "loss": 1.4846,
      "step": 24350
    },
    {
      "epoch": 4.025614542449907,
      "grad_norm": 25.313661575317383,
      "learning_rate": 3.319010631851484e-05,
      "loss": 1.4667,
      "step": 24360
    },
    {
      "epoch": 4.027267093575707,
      "grad_norm": 71.65775299072266,
      "learning_rate": 3.3180925099616226e-05,
      "loss": 1.6426,
      "step": 24370
    },
    {
      "epoch": 4.028919644701508,
      "grad_norm": 17.943498611450195,
      "learning_rate": 3.31717438807176e-05,
      "loss": 1.5965,
      "step": 24380
    },
    {
      "epoch": 4.030572195827308,
      "grad_norm": 12.518502235412598,
      "learning_rate": 3.3162562661818984e-05,
      "loss": 1.57,
      "step": 24390
    },
    {
      "epoch": 4.0322247469531085,
      "grad_norm": 64.42125701904297,
      "learning_rate": 3.315338144292036e-05,
      "loss": 1.5285,
      "step": 24400
    },
    {
      "epoch": 4.0338772980789095,
      "grad_norm": 12.844326972961426,
      "learning_rate": 3.314420022402174e-05,
      "loss": 1.5768,
      "step": 24410
    },
    {
      "epoch": 4.03552984920471,
      "grad_norm": 14.874149322509766,
      "learning_rate": 3.313501900512312e-05,
      "loss": 1.424,
      "step": 24420
    },
    {
      "epoch": 4.037182400330511,
      "grad_norm": 18.798112869262695,
      "learning_rate": 3.31258377862245e-05,
      "loss": 1.6189,
      "step": 24430
    },
    {
      "epoch": 4.038834951456311,
      "grad_norm": 8.762454986572266,
      "learning_rate": 3.3116656567325884e-05,
      "loss": 1.6398,
      "step": 24440
    },
    {
      "epoch": 4.040487502582111,
      "grad_norm": 47.06496047973633,
      "learning_rate": 3.310747534842726e-05,
      "loss": 1.4831,
      "step": 24450
    },
    {
      "epoch": 4.042140053707912,
      "grad_norm": 7.9549384117126465,
      "learning_rate": 3.309829412952864e-05,
      "loss": 1.5666,
      "step": 24460
    },
    {
      "epoch": 4.043792604833712,
      "grad_norm": 8.519120216369629,
      "learning_rate": 3.308911291063002e-05,
      "loss": 1.5574,
      "step": 24470
    },
    {
      "epoch": 4.045445155959513,
      "grad_norm": 16.42377471923828,
      "learning_rate": 3.3079931691731394e-05,
      "loss": 1.5932,
      "step": 24480
    },
    {
      "epoch": 4.047097707085313,
      "grad_norm": 6.637816429138184,
      "learning_rate": 3.307075047283277e-05,
      "loss": 1.5165,
      "step": 24490
    },
    {
      "epoch": 4.048750258211113,
      "grad_norm": 91.19857025146484,
      "learning_rate": 3.306156925393415e-05,
      "loss": 1.4989,
      "step": 24500
    },
    {
      "epoch": 4.050402809336914,
      "grad_norm": 8.961830139160156,
      "learning_rate": 3.305238803503553e-05,
      "loss": 1.5585,
      "step": 24510
    },
    {
      "epoch": 4.052055360462714,
      "grad_norm": 11.918331146240234,
      "learning_rate": 3.304320681613691e-05,
      "loss": 1.3883,
      "step": 24520
    },
    {
      "epoch": 4.053707911588515,
      "grad_norm": 15.587654113769531,
      "learning_rate": 3.303402559723829e-05,
      "loss": 1.5789,
      "step": 24530
    },
    {
      "epoch": 4.055360462714315,
      "grad_norm": 15.174756050109863,
      "learning_rate": 3.302484437833967e-05,
      "loss": 1.5297,
      "step": 24540
    },
    {
      "epoch": 4.057013013840115,
      "grad_norm": 8.750446319580078,
      "learning_rate": 3.301566315944105e-05,
      "loss": 1.4318,
      "step": 24550
    },
    {
      "epoch": 4.058665564965916,
      "grad_norm": 10.243573188781738,
      "learning_rate": 3.300648194054243e-05,
      "loss": 1.5251,
      "step": 24560
    },
    {
      "epoch": 4.060318116091716,
      "grad_norm": 21.074369430541992,
      "learning_rate": 3.299730072164381e-05,
      "loss": 1.7139,
      "step": 24570
    },
    {
      "epoch": 4.061970667217517,
      "grad_norm": 8.997513771057129,
      "learning_rate": 3.2988119502745187e-05,
      "loss": 1.7064,
      "step": 24580
    },
    {
      "epoch": 4.063623218343317,
      "grad_norm": 29.157140731811523,
      "learning_rate": 3.297893828384657e-05,
      "loss": 1.464,
      "step": 24590
    },
    {
      "epoch": 4.0652757694691175,
      "grad_norm": 11.638101577758789,
      "learning_rate": 3.2969757064947945e-05,
      "loss": 1.5183,
      "step": 24600
    },
    {
      "epoch": 4.0669283205949185,
      "grad_norm": 7.134934425354004,
      "learning_rate": 3.296057584604932e-05,
      "loss": 1.4353,
      "step": 24610
    },
    {
      "epoch": 4.068580871720719,
      "grad_norm": 4.920504093170166,
      "learning_rate": 3.2951394627150704e-05,
      "loss": 1.5342,
      "step": 24620
    },
    {
      "epoch": 4.07023342284652,
      "grad_norm": 8.68477725982666,
      "learning_rate": 3.294221340825208e-05,
      "loss": 1.49,
      "step": 24630
    },
    {
      "epoch": 4.07188597397232,
      "grad_norm": 9.163007736206055,
      "learning_rate": 3.2933032189353455e-05,
      "loss": 1.3947,
      "step": 24640
    },
    {
      "epoch": 4.073538525098121,
      "grad_norm": 32.37632751464844,
      "learning_rate": 3.292385097045484e-05,
      "loss": 1.4837,
      "step": 24650
    },
    {
      "epoch": 4.075191076223921,
      "grad_norm": 9.217835426330566,
      "learning_rate": 3.291466975155622e-05,
      "loss": 1.579,
      "step": 24660
    },
    {
      "epoch": 4.076843627349721,
      "grad_norm": 11.495637893676758,
      "learning_rate": 3.2905488532657596e-05,
      "loss": 1.6953,
      "step": 24670
    },
    {
      "epoch": 4.078496178475522,
      "grad_norm": 7.819090843200684,
      "learning_rate": 3.289630731375898e-05,
      "loss": 1.5833,
      "step": 24680
    },
    {
      "epoch": 4.080148729601322,
      "grad_norm": 17.969079971313477,
      "learning_rate": 3.2887126094860355e-05,
      "loss": 1.3082,
      "step": 24690
    },
    {
      "epoch": 4.081801280727123,
      "grad_norm": 11.410839080810547,
      "learning_rate": 3.287794487596174e-05,
      "loss": 1.5623,
      "step": 24700
    },
    {
      "epoch": 4.083453831852923,
      "grad_norm": 37.196895599365234,
      "learning_rate": 3.286876365706311e-05,
      "loss": 1.5952,
      "step": 24710
    },
    {
      "epoch": 4.085106382978723,
      "grad_norm": 7.64441442489624,
      "learning_rate": 3.2859582438164496e-05,
      "loss": 1.4717,
      "step": 24720
    },
    {
      "epoch": 4.086758934104524,
      "grad_norm": 8.817895889282227,
      "learning_rate": 3.285040121926587e-05,
      "loss": 1.4053,
      "step": 24730
    },
    {
      "epoch": 4.088411485230324,
      "grad_norm": 19.185001373291016,
      "learning_rate": 3.284122000036725e-05,
      "loss": 1.5218,
      "step": 24740
    },
    {
      "epoch": 4.090064036356125,
      "grad_norm": 15.323920249938965,
      "learning_rate": 3.283203878146863e-05,
      "loss": 1.5812,
      "step": 24750
    },
    {
      "epoch": 4.091716587481925,
      "grad_norm": 17.28553009033203,
      "learning_rate": 3.2822857562570006e-05,
      "loss": 1.4426,
      "step": 24760
    },
    {
      "epoch": 4.093369138607725,
      "grad_norm": 10.657970428466797,
      "learning_rate": 3.281367634367139e-05,
      "loss": 1.5622,
      "step": 24770
    },
    {
      "epoch": 4.095021689733526,
      "grad_norm": 10.3989896774292,
      "learning_rate": 3.2804495124772765e-05,
      "loss": 1.5272,
      "step": 24780
    },
    {
      "epoch": 4.096674240859326,
      "grad_norm": 8.306746482849121,
      "learning_rate": 3.279531390587415e-05,
      "loss": 1.6179,
      "step": 24790
    },
    {
      "epoch": 4.098326791985127,
      "grad_norm": 7.192846298217773,
      "learning_rate": 3.278613268697552e-05,
      "loss": 1.5878,
      "step": 24800
    },
    {
      "epoch": 4.0999793431109275,
      "grad_norm": 9.504925727844238,
      "learning_rate": 3.2776951468076906e-05,
      "loss": 1.4805,
      "step": 24810
    },
    {
      "epoch": 4.101631894236728,
      "grad_norm": 7.679511070251465,
      "learning_rate": 3.276777024917828e-05,
      "loss": 1.6694,
      "step": 24820
    },
    {
      "epoch": 4.103284445362529,
      "grad_norm": 8.45281982421875,
      "learning_rate": 3.2758589030279664e-05,
      "loss": 1.4295,
      "step": 24830
    },
    {
      "epoch": 4.104936996488329,
      "grad_norm": 33.410911560058594,
      "learning_rate": 3.274940781138104e-05,
      "loss": 1.6718,
      "step": 24840
    },
    {
      "epoch": 4.10658954761413,
      "grad_norm": 10.545844078063965,
      "learning_rate": 3.274022659248242e-05,
      "loss": 1.4676,
      "step": 24850
    },
    {
      "epoch": 4.10824209873993,
      "grad_norm": 9.887289047241211,
      "learning_rate": 3.27310453735838e-05,
      "loss": 1.4952,
      "step": 24860
    },
    {
      "epoch": 4.10989464986573,
      "grad_norm": 13.104098320007324,
      "learning_rate": 3.2721864154685175e-05,
      "loss": 1.6299,
      "step": 24870
    },
    {
      "epoch": 4.111547200991531,
      "grad_norm": 8.035774230957031,
      "learning_rate": 3.271268293578656e-05,
      "loss": 1.5514,
      "step": 24880
    },
    {
      "epoch": 4.113199752117331,
      "grad_norm": 7.870905876159668,
      "learning_rate": 3.270350171688793e-05,
      "loss": 1.4943,
      "step": 24890
    },
    {
      "epoch": 4.114852303243132,
      "grad_norm": 8.745378494262695,
      "learning_rate": 3.2694320497989316e-05,
      "loss": 1.4858,
      "step": 24900
    },
    {
      "epoch": 4.116504854368932,
      "grad_norm": 10.829198837280273,
      "learning_rate": 3.268513927909069e-05,
      "loss": 1.5582,
      "step": 24910
    },
    {
      "epoch": 4.118157405494732,
      "grad_norm": 19.55958366394043,
      "learning_rate": 3.2675958060192074e-05,
      "loss": 1.5017,
      "step": 24920
    },
    {
      "epoch": 4.119809956620533,
      "grad_norm": 12.78548812866211,
      "learning_rate": 3.266677684129345e-05,
      "loss": 1.4994,
      "step": 24930
    },
    {
      "epoch": 4.121462507746333,
      "grad_norm": 13.971933364868164,
      "learning_rate": 3.265759562239483e-05,
      "loss": 1.41,
      "step": 24940
    },
    {
      "epoch": 4.123115058872134,
      "grad_norm": 7.70231294631958,
      "learning_rate": 3.264841440349621e-05,
      "loss": 1.5179,
      "step": 24950
    },
    {
      "epoch": 4.124767609997934,
      "grad_norm": 11.750558853149414,
      "learning_rate": 3.263923318459759e-05,
      "loss": 1.6527,
      "step": 24960
    },
    {
      "epoch": 4.126420161123734,
      "grad_norm": 12.228842735290527,
      "learning_rate": 3.263005196569897e-05,
      "loss": 1.6643,
      "step": 24970
    },
    {
      "epoch": 4.128072712249535,
      "grad_norm": 8.996533393859863,
      "learning_rate": 3.262087074680035e-05,
      "loss": 1.4033,
      "step": 24980
    },
    {
      "epoch": 4.129725263375335,
      "grad_norm": 5.231132984161377,
      "learning_rate": 3.2611689527901726e-05,
      "loss": 1.4871,
      "step": 24990
    },
    {
      "epoch": 4.131377814501136,
      "grad_norm": 8.555381774902344,
      "learning_rate": 3.26025083090031e-05,
      "loss": 1.5646,
      "step": 25000
    },
    {
      "epoch": 4.1330303656269365,
      "grad_norm": 6.629762649536133,
      "learning_rate": 3.2593327090104484e-05,
      "loss": 1.5089,
      "step": 25010
    },
    {
      "epoch": 4.1346829167527375,
      "grad_norm": 12.015291213989258,
      "learning_rate": 3.258414587120586e-05,
      "loss": 1.5633,
      "step": 25020
    },
    {
      "epoch": 4.136335467878538,
      "grad_norm": 7.017959117889404,
      "learning_rate": 3.257496465230724e-05,
      "loss": 1.5515,
      "step": 25030
    },
    {
      "epoch": 4.137988019004338,
      "grad_norm": 9.60529899597168,
      "learning_rate": 3.256578343340862e-05,
      "loss": 1.4825,
      "step": 25040
    },
    {
      "epoch": 4.139640570130139,
      "grad_norm": 10.894176483154297,
      "learning_rate": 3.255660221451e-05,
      "loss": 1.4634,
      "step": 25050
    },
    {
      "epoch": 4.141293121255939,
      "grad_norm": 11.371975898742676,
      "learning_rate": 3.254742099561138e-05,
      "loss": 1.5382,
      "step": 25060
    },
    {
      "epoch": 4.14294567238174,
      "grad_norm": 6.190928936004639,
      "learning_rate": 3.253823977671276e-05,
      "loss": 1.3551,
      "step": 25070
    },
    {
      "epoch": 4.14459822350754,
      "grad_norm": 36.7093620300293,
      "learning_rate": 3.2529058557814135e-05,
      "loss": 1.4276,
      "step": 25080
    },
    {
      "epoch": 4.14625077463334,
      "grad_norm": 9.693130493164062,
      "learning_rate": 3.251987733891552e-05,
      "loss": 1.6435,
      "step": 25090
    },
    {
      "epoch": 4.147903325759141,
      "grad_norm": 25.358810424804688,
      "learning_rate": 3.2510696120016894e-05,
      "loss": 1.4884,
      "step": 25100
    },
    {
      "epoch": 4.149555876884941,
      "grad_norm": 16.18667984008789,
      "learning_rate": 3.2501514901118276e-05,
      "loss": 1.5553,
      "step": 25110
    },
    {
      "epoch": 4.151208428010742,
      "grad_norm": 7.252718925476074,
      "learning_rate": 3.249233368221965e-05,
      "loss": 1.4565,
      "step": 25120
    },
    {
      "epoch": 4.152860979136542,
      "grad_norm": 14.278046607971191,
      "learning_rate": 3.248315246332103e-05,
      "loss": 1.5002,
      "step": 25130
    },
    {
      "epoch": 4.154513530262342,
      "grad_norm": 33.27809524536133,
      "learning_rate": 3.247397124442241e-05,
      "loss": 1.5514,
      "step": 25140
    },
    {
      "epoch": 4.156166081388143,
      "grad_norm": 11.636425971984863,
      "learning_rate": 3.246479002552379e-05,
      "loss": 1.4199,
      "step": 25150
    },
    {
      "epoch": 4.157818632513943,
      "grad_norm": 13.851326942443848,
      "learning_rate": 3.245560880662517e-05,
      "loss": 1.4482,
      "step": 25160
    },
    {
      "epoch": 4.159471183639744,
      "grad_norm": 11.560588836669922,
      "learning_rate": 3.2446427587726545e-05,
      "loss": 1.3819,
      "step": 25170
    },
    {
      "epoch": 4.161123734765544,
      "grad_norm": 12.765569686889648,
      "learning_rate": 3.243724636882793e-05,
      "loss": 1.5236,
      "step": 25180
    },
    {
      "epoch": 4.162776285891344,
      "grad_norm": 30.970333099365234,
      "learning_rate": 3.2428065149929304e-05,
      "loss": 1.5637,
      "step": 25190
    },
    {
      "epoch": 4.164428837017145,
      "grad_norm": 13.522059440612793,
      "learning_rate": 3.2418883931030686e-05,
      "loss": 1.4642,
      "step": 25200
    },
    {
      "epoch": 4.1660813881429455,
      "grad_norm": 9.605454444885254,
      "learning_rate": 3.240970271213206e-05,
      "loss": 1.4922,
      "step": 25210
    },
    {
      "epoch": 4.1677339392687465,
      "grad_norm": 15.900501251220703,
      "learning_rate": 3.2400521493233445e-05,
      "loss": 1.5052,
      "step": 25220
    },
    {
      "epoch": 4.169386490394547,
      "grad_norm": 21.861635208129883,
      "learning_rate": 3.239134027433483e-05,
      "loss": 1.5523,
      "step": 25230
    },
    {
      "epoch": 4.171039041520347,
      "grad_norm": 10.886978149414062,
      "learning_rate": 3.23821590554362e-05,
      "loss": 1.5286,
      "step": 25240
    },
    {
      "epoch": 4.172691592646148,
      "grad_norm": 13.990205764770508,
      "learning_rate": 3.237297783653758e-05,
      "loss": 1.5499,
      "step": 25250
    },
    {
      "epoch": 4.174344143771948,
      "grad_norm": 17.176815032958984,
      "learning_rate": 3.236379661763896e-05,
      "loss": 1.5911,
      "step": 25260
    },
    {
      "epoch": 4.175996694897749,
      "grad_norm": 7.191830635070801,
      "learning_rate": 3.235461539874034e-05,
      "loss": 1.4213,
      "step": 25270
    },
    {
      "epoch": 4.177649246023549,
      "grad_norm": 8.074373245239258,
      "learning_rate": 3.2345434179841714e-05,
      "loss": 1.4857,
      "step": 25280
    },
    {
      "epoch": 4.179301797149349,
      "grad_norm": 26.866851806640625,
      "learning_rate": 3.2336252960943096e-05,
      "loss": 1.629,
      "step": 25290
    },
    {
      "epoch": 4.18095434827515,
      "grad_norm": 9.89456558227539,
      "learning_rate": 3.232707174204447e-05,
      "loss": 1.5325,
      "step": 25300
    },
    {
      "epoch": 4.18260689940095,
      "grad_norm": 9.872194290161133,
      "learning_rate": 3.2317890523145855e-05,
      "loss": 1.5802,
      "step": 25310
    },
    {
      "epoch": 4.184259450526751,
      "grad_norm": 8.017595291137695,
      "learning_rate": 3.230870930424723e-05,
      "loss": 1.4805,
      "step": 25320
    },
    {
      "epoch": 4.185912001652551,
      "grad_norm": 6.862139701843262,
      "learning_rate": 3.229952808534861e-05,
      "loss": 1.4683,
      "step": 25330
    },
    {
      "epoch": 4.187564552778351,
      "grad_norm": 10.27568244934082,
      "learning_rate": 3.2290346866449996e-05,
      "loss": 1.5286,
      "step": 25340
    },
    {
      "epoch": 4.189217103904152,
      "grad_norm": 26.036792755126953,
      "learning_rate": 3.228116564755137e-05,
      "loss": 1.5395,
      "step": 25350
    },
    {
      "epoch": 4.190869655029952,
      "grad_norm": 6.976101398468018,
      "learning_rate": 3.2271984428652754e-05,
      "loss": 1.5876,
      "step": 25360
    },
    {
      "epoch": 4.192522206155753,
      "grad_norm": 13.752799987792969,
      "learning_rate": 3.226280320975413e-05,
      "loss": 1.4922,
      "step": 25370
    },
    {
      "epoch": 4.194174757281553,
      "grad_norm": 12.30428409576416,
      "learning_rate": 3.2253621990855506e-05,
      "loss": 1.5245,
      "step": 25380
    },
    {
      "epoch": 4.195827308407354,
      "grad_norm": 8.185288429260254,
      "learning_rate": 3.224444077195689e-05,
      "loss": 1.4679,
      "step": 25390
    },
    {
      "epoch": 4.197479859533154,
      "grad_norm": 11.520121574401855,
      "learning_rate": 3.2235259553058264e-05,
      "loss": 1.5309,
      "step": 25400
    },
    {
      "epoch": 4.1991324106589545,
      "grad_norm": 23.830415725708008,
      "learning_rate": 3.222607833415964e-05,
      "loss": 1.521,
      "step": 25410
    },
    {
      "epoch": 4.2007849617847555,
      "grad_norm": 13.380767822265625,
      "learning_rate": 3.221689711526102e-05,
      "loss": 1.6148,
      "step": 25420
    },
    {
      "epoch": 4.202437512910556,
      "grad_norm": 11.789327621459961,
      "learning_rate": 3.22077158963624e-05,
      "loss": 1.5611,
      "step": 25430
    },
    {
      "epoch": 4.2040900640363565,
      "grad_norm": 8.8134765625,
      "learning_rate": 3.219853467746378e-05,
      "loss": 1.4858,
      "step": 25440
    },
    {
      "epoch": 4.205742615162157,
      "grad_norm": 9.195273399353027,
      "learning_rate": 3.218935345856516e-05,
      "loss": 1.457,
      "step": 25450
    },
    {
      "epoch": 4.207395166287957,
      "grad_norm": 12.645697593688965,
      "learning_rate": 3.218017223966654e-05,
      "loss": 1.5891,
      "step": 25460
    },
    {
      "epoch": 4.209047717413758,
      "grad_norm": 14.299896240234375,
      "learning_rate": 3.217099102076792e-05,
      "loss": 1.6082,
      "step": 25470
    },
    {
      "epoch": 4.210700268539558,
      "grad_norm": 16.324865341186523,
      "learning_rate": 3.21618098018693e-05,
      "loss": 1.5836,
      "step": 25480
    },
    {
      "epoch": 4.212352819665359,
      "grad_norm": 16.070384979248047,
      "learning_rate": 3.215262858297068e-05,
      "loss": 1.5849,
      "step": 25490
    },
    {
      "epoch": 4.214005370791159,
      "grad_norm": 20.05198097229004,
      "learning_rate": 3.214344736407206e-05,
      "loss": 1.6418,
      "step": 25500
    },
    {
      "epoch": 4.215657921916959,
      "grad_norm": 15.854426383972168,
      "learning_rate": 3.213426614517343e-05,
      "loss": 1.5086,
      "step": 25510
    },
    {
      "epoch": 4.21731047304276,
      "grad_norm": 13.941832542419434,
      "learning_rate": 3.2125084926274815e-05,
      "loss": 1.6784,
      "step": 25520
    },
    {
      "epoch": 4.21896302416856,
      "grad_norm": 11.559526443481445,
      "learning_rate": 3.211590370737619e-05,
      "loss": 1.5038,
      "step": 25530
    },
    {
      "epoch": 4.220615575294361,
      "grad_norm": 8.316058158874512,
      "learning_rate": 3.210672248847757e-05,
      "loss": 1.6032,
      "step": 25540
    },
    {
      "epoch": 4.222268126420161,
      "grad_norm": 9.870329856872559,
      "learning_rate": 3.209754126957895e-05,
      "loss": 1.4683,
      "step": 25550
    },
    {
      "epoch": 4.223920677545961,
      "grad_norm": 7.676739692687988,
      "learning_rate": 3.2088360050680326e-05,
      "loss": 1.499,
      "step": 25560
    },
    {
      "epoch": 4.225573228671762,
      "grad_norm": 10.399463653564453,
      "learning_rate": 3.207917883178171e-05,
      "loss": 1.4412,
      "step": 25570
    },
    {
      "epoch": 4.227225779797562,
      "grad_norm": 15.944662094116211,
      "learning_rate": 3.206999761288309e-05,
      "loss": 1.5239,
      "step": 25580
    },
    {
      "epoch": 4.228878330923363,
      "grad_norm": 10.55685806274414,
      "learning_rate": 3.206081639398447e-05,
      "loss": 1.4544,
      "step": 25590
    },
    {
      "epoch": 4.230530882049163,
      "grad_norm": 16.161338806152344,
      "learning_rate": 3.205163517508585e-05,
      "loss": 1.4372,
      "step": 25600
    },
    {
      "epoch": 4.2321834331749635,
      "grad_norm": 11.07420539855957,
      "learning_rate": 3.2042453956187225e-05,
      "loss": 1.5521,
      "step": 25610
    },
    {
      "epoch": 4.2338359843007645,
      "grad_norm": 19.852649688720703,
      "learning_rate": 3.203327273728861e-05,
      "loss": 1.5212,
      "step": 25620
    },
    {
      "epoch": 4.235488535426565,
      "grad_norm": 12.890336036682129,
      "learning_rate": 3.2024091518389984e-05,
      "loss": 1.6077,
      "step": 25630
    },
    {
      "epoch": 4.2371410865523655,
      "grad_norm": 17.776052474975586,
      "learning_rate": 3.201491029949136e-05,
      "loss": 1.6323,
      "step": 25640
    },
    {
      "epoch": 4.238793637678166,
      "grad_norm": 14.739234924316406,
      "learning_rate": 3.200572908059274e-05,
      "loss": 1.5555,
      "step": 25650
    },
    {
      "epoch": 4.240446188803966,
      "grad_norm": 9.28357982635498,
      "learning_rate": 3.199654786169412e-05,
      "loss": 1.5533,
      "step": 25660
    },
    {
      "epoch": 4.242098739929767,
      "grad_norm": 9.304981231689453,
      "learning_rate": 3.1987366642795494e-05,
      "loss": 1.5796,
      "step": 25670
    },
    {
      "epoch": 4.243751291055567,
      "grad_norm": 17.264741897583008,
      "learning_rate": 3.197818542389688e-05,
      "loss": 1.3938,
      "step": 25680
    },
    {
      "epoch": 4.245403842181368,
      "grad_norm": 6.289753437042236,
      "learning_rate": 3.196900420499826e-05,
      "loss": 1.3865,
      "step": 25690
    },
    {
      "epoch": 4.247056393307168,
      "grad_norm": 7.785802364349365,
      "learning_rate": 3.1959822986099635e-05,
      "loss": 1.4776,
      "step": 25700
    },
    {
      "epoch": 4.248708944432968,
      "grad_norm": 39.08346939086914,
      "learning_rate": 3.195064176720102e-05,
      "loss": 1.5611,
      "step": 25710
    },
    {
      "epoch": 4.250361495558769,
      "grad_norm": 29.96815299987793,
      "learning_rate": 3.1941460548302394e-05,
      "loss": 1.5694,
      "step": 25720
    },
    {
      "epoch": 4.252014046684569,
      "grad_norm": 17.154983520507812,
      "learning_rate": 3.1932279329403776e-05,
      "loss": 1.5357,
      "step": 25730
    },
    {
      "epoch": 4.25366659781037,
      "grad_norm": 11.302206993103027,
      "learning_rate": 3.192309811050515e-05,
      "loss": 1.6063,
      "step": 25740
    },
    {
      "epoch": 4.25531914893617,
      "grad_norm": 9.660160064697266,
      "learning_rate": 3.1913916891606535e-05,
      "loss": 1.5413,
      "step": 25750
    },
    {
      "epoch": 4.256971700061971,
      "grad_norm": 7.870529651641846,
      "learning_rate": 3.190473567270791e-05,
      "loss": 1.5043,
      "step": 25760
    },
    {
      "epoch": 4.258624251187771,
      "grad_norm": 16.945133209228516,
      "learning_rate": 3.1895554453809286e-05,
      "loss": 1.5214,
      "step": 25770
    },
    {
      "epoch": 4.260276802313571,
      "grad_norm": 11.31995677947998,
      "learning_rate": 3.188637323491067e-05,
      "loss": 1.4101,
      "step": 25780
    },
    {
      "epoch": 4.261929353439372,
      "grad_norm": 16.450275421142578,
      "learning_rate": 3.1877192016012045e-05,
      "loss": 1.4787,
      "step": 25790
    },
    {
      "epoch": 4.263581904565172,
      "grad_norm": 10.560013771057129,
      "learning_rate": 3.186801079711343e-05,
      "loss": 1.49,
      "step": 25800
    },
    {
      "epoch": 4.2652344556909725,
      "grad_norm": 27.106367111206055,
      "learning_rate": 3.1858829578214803e-05,
      "loss": 1.5307,
      "step": 25810
    },
    {
      "epoch": 4.2668870068167735,
      "grad_norm": 9.301347732543945,
      "learning_rate": 3.1849648359316186e-05,
      "loss": 1.4364,
      "step": 25820
    },
    {
      "epoch": 4.268539557942574,
      "grad_norm": 13.334724426269531,
      "learning_rate": 3.184046714041756e-05,
      "loss": 1.6001,
      "step": 25830
    },
    {
      "epoch": 4.2701921090683745,
      "grad_norm": 9.709102630615234,
      "learning_rate": 3.1831285921518945e-05,
      "loss": 1.4029,
      "step": 25840
    },
    {
      "epoch": 4.271844660194175,
      "grad_norm": 28.314720153808594,
      "learning_rate": 3.182210470262032e-05,
      "loss": 1.4576,
      "step": 25850
    },
    {
      "epoch": 4.273497211319976,
      "grad_norm": 9.064886093139648,
      "learning_rate": 3.18129234837217e-05,
      "loss": 1.5087,
      "step": 25860
    },
    {
      "epoch": 4.275149762445776,
      "grad_norm": 8.588004112243652,
      "learning_rate": 3.180374226482308e-05,
      "loss": 1.4572,
      "step": 25870
    },
    {
      "epoch": 4.276802313571576,
      "grad_norm": 9.804033279418945,
      "learning_rate": 3.179456104592446e-05,
      "loss": 1.4764,
      "step": 25880
    },
    {
      "epoch": 4.278454864697377,
      "grad_norm": 65.4351806640625,
      "learning_rate": 3.178537982702584e-05,
      "loss": 1.5435,
      "step": 25890
    },
    {
      "epoch": 4.280107415823177,
      "grad_norm": 6.215262413024902,
      "learning_rate": 3.177619860812722e-05,
      "loss": 1.6816,
      "step": 25900
    },
    {
      "epoch": 4.281759966948978,
      "grad_norm": 105.8828353881836,
      "learning_rate": 3.1767017389228596e-05,
      "loss": 1.5335,
      "step": 25910
    },
    {
      "epoch": 4.283412518074778,
      "grad_norm": 13.918194770812988,
      "learning_rate": 3.175783617032997e-05,
      "loss": 1.5267,
      "step": 25920
    },
    {
      "epoch": 4.285065069200578,
      "grad_norm": 8.10260009765625,
      "learning_rate": 3.1748654951431354e-05,
      "loss": 1.4447,
      "step": 25930
    },
    {
      "epoch": 4.286717620326379,
      "grad_norm": 16.183929443359375,
      "learning_rate": 3.173947373253273e-05,
      "loss": 1.4827,
      "step": 25940
    },
    {
      "epoch": 4.288370171452179,
      "grad_norm": 9.055156707763672,
      "learning_rate": 3.173029251363411e-05,
      "loss": 1.506,
      "step": 25950
    },
    {
      "epoch": 4.29002272257798,
      "grad_norm": 7.047090530395508,
      "learning_rate": 3.172111129473549e-05,
      "loss": 1.5904,
      "step": 25960
    },
    {
      "epoch": 4.29167527370378,
      "grad_norm": 6.232375144958496,
      "learning_rate": 3.171193007583687e-05,
      "loss": 1.5418,
      "step": 25970
    },
    {
      "epoch": 4.29332782482958,
      "grad_norm": 12.954999923706055,
      "learning_rate": 3.170274885693825e-05,
      "loss": 1.5424,
      "step": 25980
    },
    {
      "epoch": 4.294980375955381,
      "grad_norm": 15.10276985168457,
      "learning_rate": 3.169356763803963e-05,
      "loss": 1.5881,
      "step": 25990
    },
    {
      "epoch": 4.296632927081181,
      "grad_norm": 14.072202682495117,
      "learning_rate": 3.1684386419141006e-05,
      "loss": 1.6264,
      "step": 26000
    },
    {
      "epoch": 4.298285478206982,
      "grad_norm": 11.788676261901855,
      "learning_rate": 3.167520520024239e-05,
      "loss": 1.4439,
      "step": 26010
    },
    {
      "epoch": 4.2999380293327825,
      "grad_norm": 9.451334953308105,
      "learning_rate": 3.1666023981343764e-05,
      "loss": 1.5252,
      "step": 26020
    },
    {
      "epoch": 4.301590580458583,
      "grad_norm": 22.73940086364746,
      "learning_rate": 3.165684276244515e-05,
      "loss": 1.4381,
      "step": 26030
    },
    {
      "epoch": 4.3032431315843835,
      "grad_norm": 10.139789581298828,
      "learning_rate": 3.164766154354652e-05,
      "loss": 1.4498,
      "step": 26040
    },
    {
      "epoch": 4.304895682710184,
      "grad_norm": 8.749555587768555,
      "learning_rate": 3.16384803246479e-05,
      "loss": 1.4908,
      "step": 26050
    },
    {
      "epoch": 4.306548233835985,
      "grad_norm": 4.8961405754089355,
      "learning_rate": 3.162929910574928e-05,
      "loss": 1.5442,
      "step": 26060
    },
    {
      "epoch": 4.308200784961785,
      "grad_norm": 25.065303802490234,
      "learning_rate": 3.162011788685066e-05,
      "loss": 1.5435,
      "step": 26070
    },
    {
      "epoch": 4.309853336087585,
      "grad_norm": 8.996482849121094,
      "learning_rate": 3.161093666795204e-05,
      "loss": 1.5444,
      "step": 26080
    },
    {
      "epoch": 4.311505887213386,
      "grad_norm": 7.40595817565918,
      "learning_rate": 3.1601755449053416e-05,
      "loss": 1.6196,
      "step": 26090
    },
    {
      "epoch": 4.313158438339186,
      "grad_norm": 10.102790832519531,
      "learning_rate": 3.15925742301548e-05,
      "loss": 1.6037,
      "step": 26100
    },
    {
      "epoch": 4.314810989464987,
      "grad_norm": 7.849573612213135,
      "learning_rate": 3.1583393011256174e-05,
      "loss": 1.5106,
      "step": 26110
    },
    {
      "epoch": 4.316463540590787,
      "grad_norm": 8.720571517944336,
      "learning_rate": 3.157421179235756e-05,
      "loss": 1.5115,
      "step": 26120
    },
    {
      "epoch": 4.318116091716588,
      "grad_norm": 25.104869842529297,
      "learning_rate": 3.156503057345893e-05,
      "loss": 1.5522,
      "step": 26130
    },
    {
      "epoch": 4.319768642842388,
      "grad_norm": 8.44864559173584,
      "learning_rate": 3.1555849354560315e-05,
      "loss": 1.5257,
      "step": 26140
    },
    {
      "epoch": 4.321421193968188,
      "grad_norm": 9.844928741455078,
      "learning_rate": 3.154666813566169e-05,
      "loss": 1.5298,
      "step": 26150
    },
    {
      "epoch": 4.323073745093989,
      "grad_norm": 7.365574836730957,
      "learning_rate": 3.1537486916763074e-05,
      "loss": 1.4086,
      "step": 26160
    },
    {
      "epoch": 4.324726296219789,
      "grad_norm": 5.956984043121338,
      "learning_rate": 3.152830569786445e-05,
      "loss": 1.5345,
      "step": 26170
    },
    {
      "epoch": 4.326378847345589,
      "grad_norm": 18.497196197509766,
      "learning_rate": 3.1519124478965825e-05,
      "loss": 1.6195,
      "step": 26180
    },
    {
      "epoch": 4.32803139847139,
      "grad_norm": 9.667485237121582,
      "learning_rate": 3.150994326006721e-05,
      "loss": 1.5314,
      "step": 26190
    },
    {
      "epoch": 4.32968394959719,
      "grad_norm": 10.36966609954834,
      "learning_rate": 3.1500762041168584e-05,
      "loss": 1.5571,
      "step": 26200
    },
    {
      "epoch": 4.331336500722991,
      "grad_norm": 10.898240089416504,
      "learning_rate": 3.1491580822269967e-05,
      "loss": 1.4782,
      "step": 26210
    },
    {
      "epoch": 4.3329890518487915,
      "grad_norm": 19.933618545532227,
      "learning_rate": 3.148239960337134e-05,
      "loss": 1.4672,
      "step": 26220
    },
    {
      "epoch": 4.3346416029745924,
      "grad_norm": 16.246917724609375,
      "learning_rate": 3.1473218384472725e-05,
      "loss": 1.4771,
      "step": 26230
    },
    {
      "epoch": 4.3362941541003925,
      "grad_norm": 25.84589385986328,
      "learning_rate": 3.14640371655741e-05,
      "loss": 1.4655,
      "step": 26240
    },
    {
      "epoch": 4.337946705226193,
      "grad_norm": 34.56238555908203,
      "learning_rate": 3.1454855946675484e-05,
      "loss": 1.616,
      "step": 26250
    },
    {
      "epoch": 4.339599256351994,
      "grad_norm": 15.874077796936035,
      "learning_rate": 3.1445674727776866e-05,
      "loss": 1.5548,
      "step": 26260
    },
    {
      "epoch": 4.341251807477794,
      "grad_norm": 20.784616470336914,
      "learning_rate": 3.143649350887824e-05,
      "loss": 1.5053,
      "step": 26270
    },
    {
      "epoch": 4.342904358603595,
      "grad_norm": 8.26366138458252,
      "learning_rate": 3.142731228997962e-05,
      "loss": 1.5606,
      "step": 26280
    },
    {
      "epoch": 4.344556909729395,
      "grad_norm": 13.238356590270996,
      "learning_rate": 3.1418131071081e-05,
      "loss": 1.5958,
      "step": 26290
    },
    {
      "epoch": 4.346209460855195,
      "grad_norm": 14.99146556854248,
      "learning_rate": 3.1408949852182376e-05,
      "loss": 1.5942,
      "step": 26300
    },
    {
      "epoch": 4.347862011980996,
      "grad_norm": 20.563129425048828,
      "learning_rate": 3.139976863328375e-05,
      "loss": 1.5517,
      "step": 26310
    },
    {
      "epoch": 4.349514563106796,
      "grad_norm": 8.888270378112793,
      "learning_rate": 3.1390587414385135e-05,
      "loss": 1.48,
      "step": 26320
    },
    {
      "epoch": 4.351167114232597,
      "grad_norm": 12.94379997253418,
      "learning_rate": 3.138140619548651e-05,
      "loss": 1.473,
      "step": 26330
    },
    {
      "epoch": 4.352819665358397,
      "grad_norm": 17.277467727661133,
      "learning_rate": 3.1372224976587893e-05,
      "loss": 1.4717,
      "step": 26340
    },
    {
      "epoch": 4.354472216484197,
      "grad_norm": 5.993367671966553,
      "learning_rate": 3.136304375768927e-05,
      "loss": 1.4952,
      "step": 26350
    },
    {
      "epoch": 4.356124767609998,
      "grad_norm": 30.13902473449707,
      "learning_rate": 3.135386253879065e-05,
      "loss": 1.5322,
      "step": 26360
    },
    {
      "epoch": 4.357777318735798,
      "grad_norm": 18.45423126220703,
      "learning_rate": 3.1344681319892035e-05,
      "loss": 1.4632,
      "step": 26370
    },
    {
      "epoch": 4.359429869861599,
      "grad_norm": 13.011897087097168,
      "learning_rate": 3.133550010099341e-05,
      "loss": 1.5353,
      "step": 26380
    },
    {
      "epoch": 4.361082420987399,
      "grad_norm": 31.791933059692383,
      "learning_rate": 3.132631888209479e-05,
      "loss": 1.3787,
      "step": 26390
    },
    {
      "epoch": 4.362734972113199,
      "grad_norm": 10.626348495483398,
      "learning_rate": 3.131713766319617e-05,
      "loss": 1.4245,
      "step": 26400
    },
    {
      "epoch": 4.364387523239,
      "grad_norm": 8.767387390136719,
      "learning_rate": 3.1307956444297545e-05,
      "loss": 1.5493,
      "step": 26410
    },
    {
      "epoch": 4.3660400743648005,
      "grad_norm": 9.301674842834473,
      "learning_rate": 3.129877522539893e-05,
      "loss": 1.5448,
      "step": 26420
    },
    {
      "epoch": 4.367692625490601,
      "grad_norm": 7.154281139373779,
      "learning_rate": 3.12895940065003e-05,
      "loss": 1.6527,
      "step": 26430
    },
    {
      "epoch": 4.3693451766164015,
      "grad_norm": 10.634243965148926,
      "learning_rate": 3.128041278760168e-05,
      "loss": 1.4756,
      "step": 26440
    },
    {
      "epoch": 4.370997727742202,
      "grad_norm": 23.75998878479004,
      "learning_rate": 3.127123156870306e-05,
      "loss": 1.4276,
      "step": 26450
    },
    {
      "epoch": 4.372650278868003,
      "grad_norm": 11.77112102508545,
      "learning_rate": 3.126205034980444e-05,
      "loss": 1.3796,
      "step": 26460
    },
    {
      "epoch": 4.374302829993803,
      "grad_norm": 15.972389221191406,
      "learning_rate": 3.125286913090582e-05,
      "loss": 1.4398,
      "step": 26470
    },
    {
      "epoch": 4.375955381119604,
      "grad_norm": 16.23674964904785,
      "learning_rate": 3.1243687912007196e-05,
      "loss": 1.5868,
      "step": 26480
    },
    {
      "epoch": 4.377607932245404,
      "grad_norm": 9.995187759399414,
      "learning_rate": 3.123450669310858e-05,
      "loss": 1.5567,
      "step": 26490
    },
    {
      "epoch": 4.379260483371204,
      "grad_norm": 14.570067405700684,
      "learning_rate": 3.122532547420996e-05,
      "loss": 1.5392,
      "step": 26500
    },
    {
      "epoch": 4.380913034497005,
      "grad_norm": 15.810830116271973,
      "learning_rate": 3.121614425531134e-05,
      "loss": 1.5493,
      "step": 26510
    },
    {
      "epoch": 4.382565585622805,
      "grad_norm": 11.888526916503906,
      "learning_rate": 3.120696303641272e-05,
      "loss": 1.4854,
      "step": 26520
    },
    {
      "epoch": 4.384218136748606,
      "grad_norm": 6.8027663230896,
      "learning_rate": 3.1197781817514096e-05,
      "loss": 1.5692,
      "step": 26530
    },
    {
      "epoch": 4.385870687874406,
      "grad_norm": 12.093624114990234,
      "learning_rate": 3.118860059861548e-05,
      "loss": 1.409,
      "step": 26540
    },
    {
      "epoch": 4.387523239000206,
      "grad_norm": 28.978029251098633,
      "learning_rate": 3.1179419379716854e-05,
      "loss": 1.5128,
      "step": 26550
    },
    {
      "epoch": 4.389175790126007,
      "grad_norm": 7.327620506286621,
      "learning_rate": 3.117023816081823e-05,
      "loss": 1.5797,
      "step": 26560
    },
    {
      "epoch": 4.390828341251807,
      "grad_norm": 10.677287101745605,
      "learning_rate": 3.1161056941919606e-05,
      "loss": 1.5038,
      "step": 26570
    },
    {
      "epoch": 4.392480892377608,
      "grad_norm": 9.653717994689941,
      "learning_rate": 3.115187572302099e-05,
      "loss": 1.6638,
      "step": 26580
    },
    {
      "epoch": 4.394133443503408,
      "grad_norm": 6.950878620147705,
      "learning_rate": 3.1142694504122364e-05,
      "loss": 1.4129,
      "step": 26590
    },
    {
      "epoch": 4.395785994629209,
      "grad_norm": 11.13272476196289,
      "learning_rate": 3.113351328522375e-05,
      "loss": 1.5374,
      "step": 26600
    },
    {
      "epoch": 4.397438545755009,
      "grad_norm": 9.883722305297852,
      "learning_rate": 3.112433206632513e-05,
      "loss": 1.6651,
      "step": 26610
    },
    {
      "epoch": 4.3990910968808095,
      "grad_norm": 9.824180603027344,
      "learning_rate": 3.1115150847426506e-05,
      "loss": 1.5053,
      "step": 26620
    },
    {
      "epoch": 4.40074364800661,
      "grad_norm": 11.171745300292969,
      "learning_rate": 3.110596962852789e-05,
      "loss": 1.4877,
      "step": 26630
    },
    {
      "epoch": 4.4023961991324105,
      "grad_norm": 12.247041702270508,
      "learning_rate": 3.1096788409629264e-05,
      "loss": 1.6003,
      "step": 26640
    },
    {
      "epoch": 4.4040487502582115,
      "grad_norm": 7.303963661193848,
      "learning_rate": 3.108760719073065e-05,
      "loss": 1.442,
      "step": 26650
    },
    {
      "epoch": 4.405701301384012,
      "grad_norm": 20.831327438354492,
      "learning_rate": 3.107842597183202e-05,
      "loss": 1.5416,
      "step": 26660
    },
    {
      "epoch": 4.407353852509812,
      "grad_norm": 7.863111972808838,
      "learning_rate": 3.1069244752933405e-05,
      "loss": 1.6196,
      "step": 26670
    },
    {
      "epoch": 4.409006403635613,
      "grad_norm": 9.105256080627441,
      "learning_rate": 3.106006353403478e-05,
      "loss": 1.4094,
      "step": 26680
    },
    {
      "epoch": 4.410658954761413,
      "grad_norm": 54.83008575439453,
      "learning_rate": 3.105088231513616e-05,
      "loss": 1.528,
      "step": 26690
    },
    {
      "epoch": 4.412311505887214,
      "grad_norm": 10.44033432006836,
      "learning_rate": 3.104170109623753e-05,
      "loss": 1.5623,
      "step": 26700
    },
    {
      "epoch": 4.413964057013014,
      "grad_norm": 8.483716011047363,
      "learning_rate": 3.1032519877338915e-05,
      "loss": 1.4732,
      "step": 26710
    },
    {
      "epoch": 4.415616608138814,
      "grad_norm": 10.583048820495605,
      "learning_rate": 3.10233386584403e-05,
      "loss": 1.5956,
      "step": 26720
    },
    {
      "epoch": 4.417269159264615,
      "grad_norm": 12.964887619018555,
      "learning_rate": 3.1014157439541674e-05,
      "loss": 1.3888,
      "step": 26730
    },
    {
      "epoch": 4.418921710390415,
      "grad_norm": 7.349795818328857,
      "learning_rate": 3.1004976220643057e-05,
      "loss": 1.4753,
      "step": 26740
    },
    {
      "epoch": 4.420574261516216,
      "grad_norm": 17.131244659423828,
      "learning_rate": 3.099579500174443e-05,
      "loss": 1.4025,
      "step": 26750
    },
    {
      "epoch": 4.422226812642016,
      "grad_norm": 7.006505012512207,
      "learning_rate": 3.0986613782845815e-05,
      "loss": 1.5483,
      "step": 26760
    },
    {
      "epoch": 4.423879363767816,
      "grad_norm": 7.9716620445251465,
      "learning_rate": 3.097743256394719e-05,
      "loss": 1.5293,
      "step": 26770
    },
    {
      "epoch": 4.425531914893617,
      "grad_norm": 10.431143760681152,
      "learning_rate": 3.0968251345048574e-05,
      "loss": 1.4546,
      "step": 26780
    },
    {
      "epoch": 4.427184466019417,
      "grad_norm": 15.854337692260742,
      "learning_rate": 3.095907012614995e-05,
      "loss": 1.3846,
      "step": 26790
    },
    {
      "epoch": 4.428837017145218,
      "grad_norm": 8.748208999633789,
      "learning_rate": 3.094988890725133e-05,
      "loss": 1.4832,
      "step": 26800
    },
    {
      "epoch": 4.430489568271018,
      "grad_norm": 10.268428802490234,
      "learning_rate": 3.094070768835271e-05,
      "loss": 1.4754,
      "step": 26810
    },
    {
      "epoch": 4.4321421193968185,
      "grad_norm": 8.496047973632812,
      "learning_rate": 3.0931526469454084e-05,
      "loss": 1.5358,
      "step": 26820
    },
    {
      "epoch": 4.433794670522619,
      "grad_norm": 6.902714729309082,
      "learning_rate": 3.0922345250555466e-05,
      "loss": 1.5926,
      "step": 26830
    },
    {
      "epoch": 4.4354472216484195,
      "grad_norm": 8.306710243225098,
      "learning_rate": 3.091316403165684e-05,
      "loss": 1.5422,
      "step": 26840
    },
    {
      "epoch": 4.4370997727742205,
      "grad_norm": 8.802634239196777,
      "learning_rate": 3.0903982812758225e-05,
      "loss": 1.4948,
      "step": 26850
    },
    {
      "epoch": 4.438752323900021,
      "grad_norm": 13.463654518127441,
      "learning_rate": 3.08948015938596e-05,
      "loss": 1.598,
      "step": 26860
    },
    {
      "epoch": 4.440404875025821,
      "grad_norm": 16.30317497253418,
      "learning_rate": 3.0885620374960983e-05,
      "loss": 1.4191,
      "step": 26870
    },
    {
      "epoch": 4.442057426151622,
      "grad_norm": 19.12843894958496,
      "learning_rate": 3.087643915606236e-05,
      "loss": 1.5168,
      "step": 26880
    },
    {
      "epoch": 4.443709977277422,
      "grad_norm": 8.68060302734375,
      "learning_rate": 3.086725793716374e-05,
      "loss": 1.4416,
      "step": 26890
    },
    {
      "epoch": 4.445362528403223,
      "grad_norm": 26.926401138305664,
      "learning_rate": 3.085807671826512e-05,
      "loss": 1.5008,
      "step": 26900
    },
    {
      "epoch": 4.447015079529023,
      "grad_norm": 12.683781623840332,
      "learning_rate": 3.08488954993665e-05,
      "loss": 1.6921,
      "step": 26910
    },
    {
      "epoch": 4.448667630654823,
      "grad_norm": 6.749821186065674,
      "learning_rate": 3.0839714280467876e-05,
      "loss": 1.5364,
      "step": 26920
    },
    {
      "epoch": 4.450320181780624,
      "grad_norm": 7.689281463623047,
      "learning_rate": 3.083053306156926e-05,
      "loss": 1.5276,
      "step": 26930
    },
    {
      "epoch": 4.451972732906424,
      "grad_norm": 10.439373016357422,
      "learning_rate": 3.0821351842670635e-05,
      "loss": 1.4987,
      "step": 26940
    },
    {
      "epoch": 4.453625284032225,
      "grad_norm": 15.080227851867676,
      "learning_rate": 3.081217062377201e-05,
      "loss": 1.5277,
      "step": 26950
    },
    {
      "epoch": 4.455277835158025,
      "grad_norm": 7.284642219543457,
      "learning_rate": 3.080298940487339e-05,
      "loss": 1.6219,
      "step": 26960
    },
    {
      "epoch": 4.456930386283826,
      "grad_norm": 26.771888732910156,
      "learning_rate": 3.079380818597477e-05,
      "loss": 1.3984,
      "step": 26970
    },
    {
      "epoch": 4.458582937409626,
      "grad_norm": 11.829761505126953,
      "learning_rate": 3.078462696707615e-05,
      "loss": 1.4057,
      "step": 26980
    },
    {
      "epoch": 4.460235488535426,
      "grad_norm": 7.455525875091553,
      "learning_rate": 3.077544574817753e-05,
      "loss": 1.5112,
      "step": 26990
    },
    {
      "epoch": 4.461888039661227,
      "grad_norm": 18.92814826965332,
      "learning_rate": 3.076626452927891e-05,
      "loss": 1.5195,
      "step": 27000
    },
    {
      "epoch": 4.463540590787027,
      "grad_norm": 10.309431076049805,
      "learning_rate": 3.0757083310380286e-05,
      "loss": 1.5205,
      "step": 27010
    },
    {
      "epoch": 4.4651931419128275,
      "grad_norm": 6.530293941497803,
      "learning_rate": 3.074790209148167e-05,
      "loss": 1.5346,
      "step": 27020
    },
    {
      "epoch": 4.466845693038628,
      "grad_norm": 13.004956245422363,
      "learning_rate": 3.0738720872583045e-05,
      "loss": 1.5465,
      "step": 27030
    },
    {
      "epoch": 4.4684982441644285,
      "grad_norm": 9.708096504211426,
      "learning_rate": 3.072953965368443e-05,
      "loss": 1.5709,
      "step": 27040
    },
    {
      "epoch": 4.4701507952902295,
      "grad_norm": 32.74457931518555,
      "learning_rate": 3.07203584347858e-05,
      "loss": 1.6366,
      "step": 27050
    },
    {
      "epoch": 4.47180334641603,
      "grad_norm": 16.446645736694336,
      "learning_rate": 3.0711177215887186e-05,
      "loss": 1.456,
      "step": 27060
    },
    {
      "epoch": 4.473455897541831,
      "grad_norm": 11.239441871643066,
      "learning_rate": 3.070199599698856e-05,
      "loss": 1.4364,
      "step": 27070
    },
    {
      "epoch": 4.475108448667631,
      "grad_norm": 12.089319229125977,
      "learning_rate": 3.069281477808994e-05,
      "loss": 1.4886,
      "step": 27080
    },
    {
      "epoch": 4.476760999793431,
      "grad_norm": 8.01862621307373,
      "learning_rate": 3.068363355919132e-05,
      "loss": 1.5668,
      "step": 27090
    },
    {
      "epoch": 4.478413550919232,
      "grad_norm": 10.913430213928223,
      "learning_rate": 3.0674452340292696e-05,
      "loss": 1.5688,
      "step": 27100
    },
    {
      "epoch": 4.480066102045032,
      "grad_norm": 50.19123458862305,
      "learning_rate": 3.066527112139408e-05,
      "loss": 1.5296,
      "step": 27110
    },
    {
      "epoch": 4.481718653170833,
      "grad_norm": 7.208930015563965,
      "learning_rate": 3.0656089902495454e-05,
      "loss": 1.5731,
      "step": 27120
    },
    {
      "epoch": 4.483371204296633,
      "grad_norm": 8.685304641723633,
      "learning_rate": 3.064690868359684e-05,
      "loss": 1.4869,
      "step": 27130
    },
    {
      "epoch": 4.485023755422433,
      "grad_norm": 10.306434631347656,
      "learning_rate": 3.063772746469821e-05,
      "loss": 1.5608,
      "step": 27140
    },
    {
      "epoch": 4.486676306548234,
      "grad_norm": 15.503512382507324,
      "learning_rate": 3.0628546245799596e-05,
      "loss": 1.5302,
      "step": 27150
    },
    {
      "epoch": 4.488328857674034,
      "grad_norm": 11.12451171875,
      "learning_rate": 3.061936502690097e-05,
      "loss": 1.3706,
      "step": 27160
    },
    {
      "epoch": 4.489981408799835,
      "grad_norm": 8.692049026489258,
      "learning_rate": 3.0610183808002354e-05,
      "loss": 1.4919,
      "step": 27170
    },
    {
      "epoch": 4.491633959925635,
      "grad_norm": 12.720965385437012,
      "learning_rate": 3.060100258910373e-05,
      "loss": 1.63,
      "step": 27180
    },
    {
      "epoch": 4.493286511051435,
      "grad_norm": 7.457226753234863,
      "learning_rate": 3.059182137020511e-05,
      "loss": 1.7286,
      "step": 27190
    },
    {
      "epoch": 4.494939062177236,
      "grad_norm": 10.036781311035156,
      "learning_rate": 3.058264015130649e-05,
      "loss": 1.4945,
      "step": 27200
    },
    {
      "epoch": 4.496591613303036,
      "grad_norm": 9.686598777770996,
      "learning_rate": 3.0573458932407864e-05,
      "loss": 1.4577,
      "step": 27210
    },
    {
      "epoch": 4.498244164428837,
      "grad_norm": 10.819511413574219,
      "learning_rate": 3.056427771350925e-05,
      "loss": 1.4983,
      "step": 27220
    },
    {
      "epoch": 4.499896715554637,
      "grad_norm": 6.3481879234313965,
      "learning_rate": 3.055509649461062e-05,
      "loss": 1.5714,
      "step": 27230
    },
    {
      "epoch": 4.501549266680438,
      "grad_norm": 6.407865524291992,
      "learning_rate": 3.0545915275712005e-05,
      "loss": 1.4642,
      "step": 27240
    },
    {
      "epoch": 4.5032018178062385,
      "grad_norm": 22.46175193786621,
      "learning_rate": 3.053673405681338e-05,
      "loss": 1.5947,
      "step": 27250
    },
    {
      "epoch": 4.504854368932039,
      "grad_norm": 12.746009826660156,
      "learning_rate": 3.0527552837914764e-05,
      "loss": 1.4251,
      "step": 27260
    },
    {
      "epoch": 4.50650692005784,
      "grad_norm": 19.7728328704834,
      "learning_rate": 3.051837161901614e-05,
      "loss": 1.5405,
      "step": 27270
    },
    {
      "epoch": 4.50815947118364,
      "grad_norm": 15.723058700561523,
      "learning_rate": 3.0509190400117522e-05,
      "loss": 1.6367,
      "step": 27280
    },
    {
      "epoch": 4.50981202230944,
      "grad_norm": 8.419683456420898,
      "learning_rate": 3.05000091812189e-05,
      "loss": 1.5474,
      "step": 27290
    },
    {
      "epoch": 4.511464573435241,
      "grad_norm": 13.057724952697754,
      "learning_rate": 3.0490827962320277e-05,
      "loss": 1.4442,
      "step": 27300
    },
    {
      "epoch": 4.513117124561041,
      "grad_norm": 12.913500785827637,
      "learning_rate": 3.048164674342166e-05,
      "loss": 1.5384,
      "step": 27310
    },
    {
      "epoch": 4.514769675686842,
      "grad_norm": 7.9755659103393555,
      "learning_rate": 3.0472465524523036e-05,
      "loss": 1.5771,
      "step": 27320
    },
    {
      "epoch": 4.516422226812642,
      "grad_norm": 8.68862247467041,
      "learning_rate": 3.046328430562442e-05,
      "loss": 1.3936,
      "step": 27330
    },
    {
      "epoch": 4.518074777938443,
      "grad_norm": 10.90861701965332,
      "learning_rate": 3.0454103086725794e-05,
      "loss": 1.4325,
      "step": 27340
    },
    {
      "epoch": 4.519727329064243,
      "grad_norm": 10.610472679138184,
      "learning_rate": 3.0444921867827174e-05,
      "loss": 1.4947,
      "step": 27350
    },
    {
      "epoch": 4.521379880190043,
      "grad_norm": 14.162581443786621,
      "learning_rate": 3.043574064892855e-05,
      "loss": 1.4768,
      "step": 27360
    },
    {
      "epoch": 4.523032431315844,
      "grad_norm": 7.975509166717529,
      "learning_rate": 3.0426559430029932e-05,
      "loss": 1.3613,
      "step": 27370
    },
    {
      "epoch": 4.524684982441644,
      "grad_norm": 11.789381980895996,
      "learning_rate": 3.0417378211131308e-05,
      "loss": 1.5751,
      "step": 27380
    },
    {
      "epoch": 4.526337533567444,
      "grad_norm": 8.49382495880127,
      "learning_rate": 3.040819699223269e-05,
      "loss": 1.565,
      "step": 27390
    },
    {
      "epoch": 4.527990084693245,
      "grad_norm": 11.56302547454834,
      "learning_rate": 3.0399015773334073e-05,
      "loss": 1.6405,
      "step": 27400
    },
    {
      "epoch": 4.529642635819045,
      "grad_norm": 7.2713494300842285,
      "learning_rate": 3.038983455443545e-05,
      "loss": 1.5445,
      "step": 27410
    },
    {
      "epoch": 4.531295186944846,
      "grad_norm": 40.96128463745117,
      "learning_rate": 3.038065333553683e-05,
      "loss": 1.5608,
      "step": 27420
    },
    {
      "epoch": 4.532947738070646,
      "grad_norm": 6.933916091918945,
      "learning_rate": 3.0371472116638204e-05,
      "loss": 1.7051,
      "step": 27430
    },
    {
      "epoch": 4.534600289196447,
      "grad_norm": 9.174333572387695,
      "learning_rate": 3.0362290897739587e-05,
      "loss": 1.6537,
      "step": 27440
    },
    {
      "epoch": 4.5362528403222475,
      "grad_norm": 11.921561241149902,
      "learning_rate": 3.0353109678840963e-05,
      "loss": 1.458,
      "step": 27450
    },
    {
      "epoch": 4.537905391448048,
      "grad_norm": 11.688679695129395,
      "learning_rate": 3.0343928459942345e-05,
      "loss": 1.4882,
      "step": 27460
    },
    {
      "epoch": 4.539557942573849,
      "grad_norm": 9.94924545288086,
      "learning_rate": 3.033474724104372e-05,
      "loss": 1.3484,
      "step": 27470
    },
    {
      "epoch": 4.541210493699649,
      "grad_norm": 14.807235717773438,
      "learning_rate": 3.03255660221451e-05,
      "loss": 1.5368,
      "step": 27480
    },
    {
      "epoch": 4.54286304482545,
      "grad_norm": 7.9106597900390625,
      "learning_rate": 3.0316384803246476e-05,
      "loss": 1.5756,
      "step": 27490
    },
    {
      "epoch": 4.54451559595125,
      "grad_norm": 17.72800636291504,
      "learning_rate": 3.030720358434786e-05,
      "loss": 1.6303,
      "step": 27500
    },
    {
      "epoch": 4.54616814707705,
      "grad_norm": 14.3507080078125,
      "learning_rate": 3.0298022365449242e-05,
      "loss": 1.488,
      "step": 27510
    },
    {
      "epoch": 4.547820698202851,
      "grad_norm": 11.83436107635498,
      "learning_rate": 3.0288841146550618e-05,
      "loss": 1.4159,
      "step": 27520
    },
    {
      "epoch": 4.549473249328651,
      "grad_norm": 32.23595428466797,
      "learning_rate": 3.0279659927652e-05,
      "loss": 1.4765,
      "step": 27530
    },
    {
      "epoch": 4.551125800454452,
      "grad_norm": 7.518164157867432,
      "learning_rate": 3.0270478708753376e-05,
      "loss": 1.5396,
      "step": 27540
    },
    {
      "epoch": 4.552778351580252,
      "grad_norm": 12.867490768432617,
      "learning_rate": 3.0261297489854755e-05,
      "loss": 1.6184,
      "step": 27550
    },
    {
      "epoch": 4.554430902706052,
      "grad_norm": 59.21015548706055,
      "learning_rate": 3.025211627095613e-05,
      "loss": 1.3638,
      "step": 27560
    },
    {
      "epoch": 4.556083453831853,
      "grad_norm": 8.144857406616211,
      "learning_rate": 3.0242935052057514e-05,
      "loss": 1.6169,
      "step": 27570
    },
    {
      "epoch": 4.557736004957653,
      "grad_norm": 13.78732681274414,
      "learning_rate": 3.023375383315889e-05,
      "loss": 1.5019,
      "step": 27580
    },
    {
      "epoch": 4.559388556083454,
      "grad_norm": 12.740043640136719,
      "learning_rate": 3.0224572614260272e-05,
      "loss": 1.5669,
      "step": 27590
    },
    {
      "epoch": 4.561041107209254,
      "grad_norm": 7.079008102416992,
      "learning_rate": 3.0215391395361648e-05,
      "loss": 1.5509,
      "step": 27600
    },
    {
      "epoch": 4.562693658335054,
      "grad_norm": 13.054255485534668,
      "learning_rate": 3.0206210176463027e-05,
      "loss": 1.443,
      "step": 27610
    },
    {
      "epoch": 4.564346209460855,
      "grad_norm": 8.606029510498047,
      "learning_rate": 3.0197028957564407e-05,
      "loss": 1.5148,
      "step": 27620
    },
    {
      "epoch": 4.565998760586655,
      "grad_norm": 6.800489902496338,
      "learning_rate": 3.0187847738665786e-05,
      "loss": 1.5745,
      "step": 27630
    },
    {
      "epoch": 4.567651311712456,
      "grad_norm": 10.607884407043457,
      "learning_rate": 3.017866651976717e-05,
      "loss": 1.5926,
      "step": 27640
    },
    {
      "epoch": 4.5693038628382565,
      "grad_norm": 9.12733268737793,
      "learning_rate": 3.0169485300868544e-05,
      "loss": 1.5111,
      "step": 27650
    },
    {
      "epoch": 4.570956413964057,
      "grad_norm": 7.448328971862793,
      "learning_rate": 3.0160304081969927e-05,
      "loss": 1.6386,
      "step": 27660
    },
    {
      "epoch": 4.572608965089858,
      "grad_norm": 21.30801010131836,
      "learning_rate": 3.0151122863071303e-05,
      "loss": 1.4581,
      "step": 27670
    },
    {
      "epoch": 4.574261516215658,
      "grad_norm": 19.67100715637207,
      "learning_rate": 3.0141941644172682e-05,
      "loss": 1.4829,
      "step": 27680
    },
    {
      "epoch": 4.575914067341459,
      "grad_norm": 14.657584190368652,
      "learning_rate": 3.0132760425274058e-05,
      "loss": 1.5703,
      "step": 27690
    },
    {
      "epoch": 4.577566618467259,
      "grad_norm": 7.990481853485107,
      "learning_rate": 3.012357920637544e-05,
      "loss": 1.4242,
      "step": 27700
    },
    {
      "epoch": 4.57921916959306,
      "grad_norm": 8.869722366333008,
      "learning_rate": 3.0114397987476816e-05,
      "loss": 1.5055,
      "step": 27710
    },
    {
      "epoch": 4.58087172071886,
      "grad_norm": 11.212498664855957,
      "learning_rate": 3.01052167685782e-05,
      "loss": 1.4827,
      "step": 27720
    },
    {
      "epoch": 4.58252427184466,
      "grad_norm": 10.676173210144043,
      "learning_rate": 3.0096035549679575e-05,
      "loss": 1.5627,
      "step": 27730
    },
    {
      "epoch": 4.584176822970461,
      "grad_norm": 25.556760787963867,
      "learning_rate": 3.0086854330780954e-05,
      "loss": 1.5832,
      "step": 27740
    },
    {
      "epoch": 4.585829374096261,
      "grad_norm": 8.269768714904785,
      "learning_rate": 3.0077673111882337e-05,
      "loss": 1.5174,
      "step": 27750
    },
    {
      "epoch": 4.587481925222061,
      "grad_norm": 11.795088768005371,
      "learning_rate": 3.0068491892983713e-05,
      "loss": 1.5345,
      "step": 27760
    },
    {
      "epoch": 4.589134476347862,
      "grad_norm": 12.628785133361816,
      "learning_rate": 3.0059310674085095e-05,
      "loss": 1.4388,
      "step": 27770
    },
    {
      "epoch": 4.590787027473662,
      "grad_norm": 8.711159706115723,
      "learning_rate": 3.005012945518647e-05,
      "loss": 1.4389,
      "step": 27780
    },
    {
      "epoch": 4.592439578599463,
      "grad_norm": 5.210586071014404,
      "learning_rate": 3.0040948236287854e-05,
      "loss": 1.4444,
      "step": 27790
    },
    {
      "epoch": 4.594092129725263,
      "grad_norm": 11.018863677978516,
      "learning_rate": 3.003176701738923e-05,
      "loss": 1.4392,
      "step": 27800
    },
    {
      "epoch": 4.595744680851064,
      "grad_norm": 8.213711738586426,
      "learning_rate": 3.002258579849061e-05,
      "loss": 1.4233,
      "step": 27810
    },
    {
      "epoch": 4.597397231976864,
      "grad_norm": 22.795164108276367,
      "learning_rate": 3.0013404579591985e-05,
      "loss": 1.5016,
      "step": 27820
    },
    {
      "epoch": 4.599049783102664,
      "grad_norm": 10.903739929199219,
      "learning_rate": 3.0004223360693367e-05,
      "loss": 1.456,
      "step": 27830
    },
    {
      "epoch": 4.600702334228465,
      "grad_norm": 13.617247581481934,
      "learning_rate": 2.9995042141794743e-05,
      "loss": 1.6139,
      "step": 27840
    },
    {
      "epoch": 4.6023548853542655,
      "grad_norm": 11.749178886413574,
      "learning_rate": 2.9985860922896126e-05,
      "loss": 1.4379,
      "step": 27850
    },
    {
      "epoch": 4.604007436480066,
      "grad_norm": 11.892571449279785,
      "learning_rate": 2.9976679703997505e-05,
      "loss": 1.5134,
      "step": 27860
    },
    {
      "epoch": 4.605659987605867,
      "grad_norm": 16.06785011291504,
      "learning_rate": 2.996749848509888e-05,
      "loss": 1.5021,
      "step": 27870
    },
    {
      "epoch": 4.607312538731667,
      "grad_norm": 11.480622291564941,
      "learning_rate": 2.9958317266200264e-05,
      "loss": 1.5267,
      "step": 27880
    },
    {
      "epoch": 4.608965089857468,
      "grad_norm": 11.960569381713867,
      "learning_rate": 2.994913604730164e-05,
      "loss": 1.5191,
      "step": 27890
    },
    {
      "epoch": 4.610617640983268,
      "grad_norm": 10.92312240600586,
      "learning_rate": 2.9939954828403022e-05,
      "loss": 1.6442,
      "step": 27900
    },
    {
      "epoch": 4.612270192109069,
      "grad_norm": 8.785314559936523,
      "learning_rate": 2.9930773609504398e-05,
      "loss": 1.6221,
      "step": 27910
    },
    {
      "epoch": 4.613922743234869,
      "grad_norm": 10.37857723236084,
      "learning_rate": 2.992159239060578e-05,
      "loss": 1.4723,
      "step": 27920
    },
    {
      "epoch": 4.615575294360669,
      "grad_norm": 47.27583694458008,
      "learning_rate": 2.9912411171707157e-05,
      "loss": 1.5619,
      "step": 27930
    },
    {
      "epoch": 4.61722784548647,
      "grad_norm": 28.969919204711914,
      "learning_rate": 2.9903229952808536e-05,
      "loss": 1.3872,
      "step": 27940
    },
    {
      "epoch": 4.61888039661227,
      "grad_norm": 16.729145050048828,
      "learning_rate": 2.989404873390991e-05,
      "loss": 1.5894,
      "step": 27950
    },
    {
      "epoch": 4.620532947738071,
      "grad_norm": 6.659755229949951,
      "learning_rate": 2.9884867515011294e-05,
      "loss": 1.6566,
      "step": 27960
    },
    {
      "epoch": 4.622185498863871,
      "grad_norm": 15.295912742614746,
      "learning_rate": 2.9875686296112677e-05,
      "loss": 1.4874,
      "step": 27970
    },
    {
      "epoch": 4.623838049989671,
      "grad_norm": 9.182865142822266,
      "learning_rate": 2.9866505077214053e-05,
      "loss": 1.6185,
      "step": 27980
    },
    {
      "epoch": 4.625490601115472,
      "grad_norm": 12.22716999053955,
      "learning_rate": 2.9857323858315432e-05,
      "loss": 1.4013,
      "step": 27990
    },
    {
      "epoch": 4.627143152241272,
      "grad_norm": 14.52023983001709,
      "learning_rate": 2.9848142639416808e-05,
      "loss": 1.6018,
      "step": 28000
    },
    {
      "epoch": 4.628795703367073,
      "grad_norm": 9.691109657287598,
      "learning_rate": 2.983896142051819e-05,
      "loss": 1.4829,
      "step": 28010
    },
    {
      "epoch": 4.630448254492873,
      "grad_norm": 7.735055446624756,
      "learning_rate": 2.9829780201619566e-05,
      "loss": 1.5496,
      "step": 28020
    },
    {
      "epoch": 4.632100805618673,
      "grad_norm": 9.349820137023926,
      "learning_rate": 2.982059898272095e-05,
      "loss": 1.5716,
      "step": 28030
    },
    {
      "epoch": 4.633753356744474,
      "grad_norm": 47.04668426513672,
      "learning_rate": 2.9811417763822325e-05,
      "loss": 1.5508,
      "step": 28040
    },
    {
      "epoch": 4.6354059078702745,
      "grad_norm": 11.281471252441406,
      "learning_rate": 2.9802236544923708e-05,
      "loss": 1.5374,
      "step": 28050
    },
    {
      "epoch": 4.6370584589960755,
      "grad_norm": 10.420440673828125,
      "learning_rate": 2.9793055326025083e-05,
      "loss": 1.4952,
      "step": 28060
    },
    {
      "epoch": 4.638711010121876,
      "grad_norm": 7.602105140686035,
      "learning_rate": 2.9783874107126463e-05,
      "loss": 1.6077,
      "step": 28070
    },
    {
      "epoch": 4.640363561247677,
      "grad_norm": 11.72873306274414,
      "learning_rate": 2.9774692888227845e-05,
      "loss": 1.517,
      "step": 28080
    },
    {
      "epoch": 4.642016112373477,
      "grad_norm": 5.891235828399658,
      "learning_rate": 2.976551166932922e-05,
      "loss": 1.5004,
      "step": 28090
    },
    {
      "epoch": 4.643668663499277,
      "grad_norm": 8.504074096679688,
      "learning_rate": 2.9756330450430604e-05,
      "loss": 1.4433,
      "step": 28100
    },
    {
      "epoch": 4.645321214625078,
      "grad_norm": 11.097142219543457,
      "learning_rate": 2.974714923153198e-05,
      "loss": 1.517,
      "step": 28110
    },
    {
      "epoch": 4.646973765750878,
      "grad_norm": 25.318063735961914,
      "learning_rate": 2.973796801263336e-05,
      "loss": 1.475,
      "step": 28120
    },
    {
      "epoch": 4.648626316876678,
      "grad_norm": 12.631922721862793,
      "learning_rate": 2.9728786793734735e-05,
      "loss": 1.6466,
      "step": 28130
    },
    {
      "epoch": 4.650278868002479,
      "grad_norm": 18.910295486450195,
      "learning_rate": 2.9719605574836117e-05,
      "loss": 1.4223,
      "step": 28140
    },
    {
      "epoch": 4.651931419128279,
      "grad_norm": 10.944915771484375,
      "learning_rate": 2.9710424355937493e-05,
      "loss": 1.4081,
      "step": 28150
    },
    {
      "epoch": 4.65358397025408,
      "grad_norm": 36.19316482543945,
      "learning_rate": 2.9701243137038876e-05,
      "loss": 1.5777,
      "step": 28160
    },
    {
      "epoch": 4.65523652137988,
      "grad_norm": 11.189594268798828,
      "learning_rate": 2.9692061918140252e-05,
      "loss": 1.6023,
      "step": 28170
    },
    {
      "epoch": 4.656889072505681,
      "grad_norm": 8.958883285522461,
      "learning_rate": 2.9682880699241634e-05,
      "loss": 1.4499,
      "step": 28180
    },
    {
      "epoch": 4.658541623631481,
      "grad_norm": 21.078588485717773,
      "learning_rate": 2.967369948034301e-05,
      "loss": 1.5326,
      "step": 28190
    },
    {
      "epoch": 4.660194174757281,
      "grad_norm": 9.72829818725586,
      "learning_rate": 2.966451826144439e-05,
      "loss": 1.665,
      "step": 28200
    },
    {
      "epoch": 4.661846725883082,
      "grad_norm": 6.814817905426025,
      "learning_rate": 2.9655337042545772e-05,
      "loss": 1.4978,
      "step": 28210
    },
    {
      "epoch": 4.663499277008882,
      "grad_norm": 12.29881763458252,
      "learning_rate": 2.9646155823647148e-05,
      "loss": 1.5255,
      "step": 28220
    },
    {
      "epoch": 4.665151828134682,
      "grad_norm": 9.907049179077148,
      "learning_rate": 2.963697460474853e-05,
      "loss": 1.5061,
      "step": 28230
    },
    {
      "epoch": 4.666804379260483,
      "grad_norm": 8.112325668334961,
      "learning_rate": 2.9627793385849906e-05,
      "loss": 1.5927,
      "step": 28240
    },
    {
      "epoch": 4.6684569303862835,
      "grad_norm": 31.478939056396484,
      "learning_rate": 2.9618612166951286e-05,
      "loss": 1.4132,
      "step": 28250
    },
    {
      "epoch": 4.6701094815120845,
      "grad_norm": 8.31009292602539,
      "learning_rate": 2.9609430948052665e-05,
      "loss": 1.409,
      "step": 28260
    },
    {
      "epoch": 4.671762032637885,
      "grad_norm": 69.70064544677734,
      "learning_rate": 2.9600249729154044e-05,
      "loss": 1.646,
      "step": 28270
    },
    {
      "epoch": 4.673414583763686,
      "grad_norm": 8.266491889953613,
      "learning_rate": 2.959106851025542e-05,
      "loss": 1.5419,
      "step": 28280
    },
    {
      "epoch": 4.675067134889486,
      "grad_norm": 7.468424320220947,
      "learning_rate": 2.9581887291356803e-05,
      "loss": 1.4903,
      "step": 28290
    },
    {
      "epoch": 4.676719686015286,
      "grad_norm": 8.034427642822266,
      "learning_rate": 2.957270607245818e-05,
      "loss": 1.5049,
      "step": 28300
    },
    {
      "epoch": 4.678372237141087,
      "grad_norm": 6.364953517913818,
      "learning_rate": 2.956352485355956e-05,
      "loss": 1.4946,
      "step": 28310
    },
    {
      "epoch": 4.680024788266887,
      "grad_norm": 6.952188014984131,
      "learning_rate": 2.955434363466094e-05,
      "loss": 1.585,
      "step": 28320
    },
    {
      "epoch": 4.681677339392688,
      "grad_norm": 6.363594055175781,
      "learning_rate": 2.9545162415762316e-05,
      "loss": 1.4206,
      "step": 28330
    },
    {
      "epoch": 4.683329890518488,
      "grad_norm": 12.087472915649414,
      "learning_rate": 2.95359811968637e-05,
      "loss": 1.5401,
      "step": 28340
    },
    {
      "epoch": 4.684982441644288,
      "grad_norm": 7.647066593170166,
      "learning_rate": 2.9526799977965075e-05,
      "loss": 1.626,
      "step": 28350
    },
    {
      "epoch": 4.686634992770089,
      "grad_norm": 10.387514114379883,
      "learning_rate": 2.9517618759066457e-05,
      "loss": 1.5102,
      "step": 28360
    },
    {
      "epoch": 4.688287543895889,
      "grad_norm": 10.926636695861816,
      "learning_rate": 2.9508437540167833e-05,
      "loss": 1.511,
      "step": 28370
    },
    {
      "epoch": 4.68994009502169,
      "grad_norm": 10.67368221282959,
      "learning_rate": 2.9499256321269213e-05,
      "loss": 1.4165,
      "step": 28380
    },
    {
      "epoch": 4.69159264614749,
      "grad_norm": 9.141593933105469,
      "learning_rate": 2.9490075102370592e-05,
      "loss": 1.5694,
      "step": 28390
    },
    {
      "epoch": 4.69324519727329,
      "grad_norm": 6.583410739898682,
      "learning_rate": 2.948089388347197e-05,
      "loss": 1.5357,
      "step": 28400
    },
    {
      "epoch": 4.694897748399091,
      "grad_norm": 10.10991382598877,
      "learning_rate": 2.9471712664573347e-05,
      "loss": 1.5769,
      "step": 28410
    },
    {
      "epoch": 4.696550299524891,
      "grad_norm": 7.276322841644287,
      "learning_rate": 2.946253144567473e-05,
      "loss": 1.6132,
      "step": 28420
    },
    {
      "epoch": 4.698202850650692,
      "grad_norm": 6.623241901397705,
      "learning_rate": 2.9453350226776112e-05,
      "loss": 1.449,
      "step": 28430
    },
    {
      "epoch": 4.699855401776492,
      "grad_norm": 8.261223793029785,
      "learning_rate": 2.9444169007877488e-05,
      "loss": 1.4652,
      "step": 28440
    },
    {
      "epoch": 4.701507952902293,
      "grad_norm": 38.858604431152344,
      "learning_rate": 2.9434987788978867e-05,
      "loss": 1.5068,
      "step": 28450
    },
    {
      "epoch": 4.7031605040280935,
      "grad_norm": 15.038886070251465,
      "learning_rate": 2.9425806570080243e-05,
      "loss": 1.636,
      "step": 28460
    },
    {
      "epoch": 4.704813055153894,
      "grad_norm": 11.770708084106445,
      "learning_rate": 2.9416625351181626e-05,
      "loss": 1.5393,
      "step": 28470
    },
    {
      "epoch": 4.706465606279695,
      "grad_norm": 24.59950065612793,
      "learning_rate": 2.9407444132283e-05,
      "loss": 1.508,
      "step": 28480
    },
    {
      "epoch": 4.708118157405495,
      "grad_norm": 8.745185852050781,
      "learning_rate": 2.9398262913384384e-05,
      "loss": 1.6451,
      "step": 28490
    },
    {
      "epoch": 4.709770708531295,
      "grad_norm": 12.335895538330078,
      "learning_rate": 2.938908169448576e-05,
      "loss": 1.487,
      "step": 28500
    },
    {
      "epoch": 4.711423259657096,
      "grad_norm": 38.088829040527344,
      "learning_rate": 2.937990047558714e-05,
      "loss": 1.4954,
      "step": 28510
    },
    {
      "epoch": 4.713075810782896,
      "grad_norm": 12.632776260375977,
      "learning_rate": 2.937071925668852e-05,
      "loss": 1.5633,
      "step": 28520
    },
    {
      "epoch": 4.714728361908697,
      "grad_norm": 11.09786319732666,
      "learning_rate": 2.9361538037789898e-05,
      "loss": 1.4451,
      "step": 28530
    },
    {
      "epoch": 4.716380913034497,
      "grad_norm": 11.804417610168457,
      "learning_rate": 2.935235681889128e-05,
      "loss": 1.596,
      "step": 28540
    },
    {
      "epoch": 4.718033464160298,
      "grad_norm": 10.606738090515137,
      "learning_rate": 2.9343175599992656e-05,
      "loss": 1.4794,
      "step": 28550
    },
    {
      "epoch": 4.719686015286098,
      "grad_norm": 8.68057632446289,
      "learning_rate": 2.933399438109404e-05,
      "loss": 1.5829,
      "step": 28560
    },
    {
      "epoch": 4.721338566411898,
      "grad_norm": 9.055428504943848,
      "learning_rate": 2.9324813162195415e-05,
      "loss": 1.5983,
      "step": 28570
    },
    {
      "epoch": 4.722991117537699,
      "grad_norm": 18.714929580688477,
      "learning_rate": 2.9315631943296794e-05,
      "loss": 1.5333,
      "step": 28580
    },
    {
      "epoch": 4.724643668663499,
      "grad_norm": 11.138300895690918,
      "learning_rate": 2.930645072439817e-05,
      "loss": 1.5573,
      "step": 28590
    },
    {
      "epoch": 4.726296219789299,
      "grad_norm": 11.27102279663086,
      "learning_rate": 2.9297269505499553e-05,
      "loss": 1.4813,
      "step": 28600
    },
    {
      "epoch": 4.7279487709151,
      "grad_norm": 9.76452350616455,
      "learning_rate": 2.928808828660093e-05,
      "loss": 1.6399,
      "step": 28610
    },
    {
      "epoch": 4.7296013220409,
      "grad_norm": 9.003868103027344,
      "learning_rate": 2.927890706770231e-05,
      "loss": 1.5414,
      "step": 28620
    },
    {
      "epoch": 4.731253873166701,
      "grad_norm": 8.384162902832031,
      "learning_rate": 2.9269725848803687e-05,
      "loss": 1.4369,
      "step": 28630
    },
    {
      "epoch": 4.732906424292501,
      "grad_norm": 13.919119834899902,
      "learning_rate": 2.9260544629905066e-05,
      "loss": 1.4788,
      "step": 28640
    },
    {
      "epoch": 4.734558975418302,
      "grad_norm": 9.051968574523926,
      "learning_rate": 2.925136341100645e-05,
      "loss": 1.6114,
      "step": 28650
    },
    {
      "epoch": 4.7362115265441025,
      "grad_norm": 12.999975204467773,
      "learning_rate": 2.9242182192107825e-05,
      "loss": 1.4965,
      "step": 28660
    },
    {
      "epoch": 4.737864077669903,
      "grad_norm": 9.09343147277832,
      "learning_rate": 2.9233000973209207e-05,
      "loss": 1.5995,
      "step": 28670
    },
    {
      "epoch": 4.739516628795704,
      "grad_norm": 9.132031440734863,
      "learning_rate": 2.9223819754310583e-05,
      "loss": 1.4807,
      "step": 28680
    },
    {
      "epoch": 4.741169179921504,
      "grad_norm": 7.520742416381836,
      "learning_rate": 2.9214638535411966e-05,
      "loss": 1.5562,
      "step": 28690
    },
    {
      "epoch": 4.742821731047305,
      "grad_norm": 8.723919868469238,
      "learning_rate": 2.920545731651334e-05,
      "loss": 1.4716,
      "step": 28700
    },
    {
      "epoch": 4.744474282173105,
      "grad_norm": 22.840511322021484,
      "learning_rate": 2.919627609761472e-05,
      "loss": 1.5385,
      "step": 28710
    },
    {
      "epoch": 4.746126833298905,
      "grad_norm": 8.98953628540039,
      "learning_rate": 2.9187094878716097e-05,
      "loss": 1.4941,
      "step": 28720
    },
    {
      "epoch": 4.747779384424706,
      "grad_norm": 8.429163932800293,
      "learning_rate": 2.917791365981748e-05,
      "loss": 1.5198,
      "step": 28730
    },
    {
      "epoch": 4.749431935550506,
      "grad_norm": 12.46285629272461,
      "learning_rate": 2.9168732440918855e-05,
      "loss": 1.4711,
      "step": 28740
    },
    {
      "epoch": 4.751084486676307,
      "grad_norm": 10.32840633392334,
      "learning_rate": 2.9159551222020238e-05,
      "loss": 1.4423,
      "step": 28750
    },
    {
      "epoch": 4.752737037802107,
      "grad_norm": 5.583861827850342,
      "learning_rate": 2.9150370003121614e-05,
      "loss": 1.6284,
      "step": 28760
    },
    {
      "epoch": 4.754389588927907,
      "grad_norm": 9.908193588256836,
      "learning_rate": 2.9141188784222993e-05,
      "loss": 1.5041,
      "step": 28770
    },
    {
      "epoch": 4.756042140053708,
      "grad_norm": 11.036250114440918,
      "learning_rate": 2.9132007565324376e-05,
      "loss": 1.529,
      "step": 28780
    },
    {
      "epoch": 4.757694691179508,
      "grad_norm": 13.065105438232422,
      "learning_rate": 2.912282634642575e-05,
      "loss": 1.5542,
      "step": 28790
    },
    {
      "epoch": 4.759347242305309,
      "grad_norm": 11.913582801818848,
      "learning_rate": 2.9113645127527134e-05,
      "loss": 1.6069,
      "step": 28800
    },
    {
      "epoch": 4.760999793431109,
      "grad_norm": 9.790139198303223,
      "learning_rate": 2.910446390862851e-05,
      "loss": 1.5313,
      "step": 28810
    },
    {
      "epoch": 4.76265234455691,
      "grad_norm": 10.083866119384766,
      "learning_rate": 2.9095282689729893e-05,
      "loss": 1.5612,
      "step": 28820
    },
    {
      "epoch": 4.76430489568271,
      "grad_norm": 17.150331497192383,
      "learning_rate": 2.908610147083127e-05,
      "loss": 1.5995,
      "step": 28830
    },
    {
      "epoch": 4.76595744680851,
      "grad_norm": 13.593814849853516,
      "learning_rate": 2.9076920251932648e-05,
      "loss": 1.4406,
      "step": 28840
    },
    {
      "epoch": 4.767609997934311,
      "grad_norm": 8.73280143737793,
      "learning_rate": 2.9067739033034024e-05,
      "loss": 1.6393,
      "step": 28850
    },
    {
      "epoch": 4.7692625490601115,
      "grad_norm": 11.49311351776123,
      "learning_rate": 2.9058557814135406e-05,
      "loss": 1.5007,
      "step": 28860
    },
    {
      "epoch": 4.770915100185912,
      "grad_norm": 18.270648956298828,
      "learning_rate": 2.9049376595236782e-05,
      "loss": 1.4804,
      "step": 28870
    },
    {
      "epoch": 4.772567651311713,
      "grad_norm": 7.664954662322998,
      "learning_rate": 2.9040195376338165e-05,
      "loss": 1.5895,
      "step": 28880
    },
    {
      "epoch": 4.774220202437513,
      "grad_norm": 7.60715389251709,
      "learning_rate": 2.9031014157439544e-05,
      "loss": 1.5162,
      "step": 28890
    },
    {
      "epoch": 4.775872753563314,
      "grad_norm": 9.755659103393555,
      "learning_rate": 2.9021832938540923e-05,
      "loss": 1.4321,
      "step": 28900
    },
    {
      "epoch": 4.777525304689114,
      "grad_norm": 15.464385032653809,
      "learning_rate": 2.9012651719642302e-05,
      "loss": 1.4708,
      "step": 28910
    },
    {
      "epoch": 4.779177855814915,
      "grad_norm": 24.310163497924805,
      "learning_rate": 2.900347050074368e-05,
      "loss": 1.4532,
      "step": 28920
    },
    {
      "epoch": 4.780830406940715,
      "grad_norm": 15.663667678833008,
      "learning_rate": 2.899428928184506e-05,
      "loss": 1.4912,
      "step": 28930
    },
    {
      "epoch": 4.782482958066515,
      "grad_norm": 12.316214561462402,
      "learning_rate": 2.8985108062946437e-05,
      "loss": 1.5429,
      "step": 28940
    },
    {
      "epoch": 4.784135509192316,
      "grad_norm": 14.147255897521973,
      "learning_rate": 2.897592684404782e-05,
      "loss": 1.4476,
      "step": 28950
    },
    {
      "epoch": 4.785788060318116,
      "grad_norm": 9.104997634887695,
      "learning_rate": 2.8966745625149195e-05,
      "loss": 1.4273,
      "step": 28960
    },
    {
      "epoch": 4.787440611443916,
      "grad_norm": 11.243693351745605,
      "learning_rate": 2.8957564406250575e-05,
      "loss": 1.5976,
      "step": 28970
    },
    {
      "epoch": 4.789093162569717,
      "grad_norm": 15.91396427154541,
      "learning_rate": 2.894838318735195e-05,
      "loss": 1.5532,
      "step": 28980
    },
    {
      "epoch": 4.790745713695517,
      "grad_norm": 23.990699768066406,
      "learning_rate": 2.8939201968453333e-05,
      "loss": 1.5387,
      "step": 28990
    },
    {
      "epoch": 4.792398264821318,
      "grad_norm": 13.71623706817627,
      "learning_rate": 2.8930020749554716e-05,
      "loss": 1.5653,
      "step": 29000
    },
    {
      "epoch": 4.794050815947118,
      "grad_norm": 7.524801731109619,
      "learning_rate": 2.892083953065609e-05,
      "loss": 1.5804,
      "step": 29010
    },
    {
      "epoch": 4.795703367072919,
      "grad_norm": 31.6929874420166,
      "learning_rate": 2.891165831175747e-05,
      "loss": 1.5599,
      "step": 29020
    },
    {
      "epoch": 4.797355918198719,
      "grad_norm": 10.225500106811523,
      "learning_rate": 2.890247709285885e-05,
      "loss": 1.3953,
      "step": 29030
    },
    {
      "epoch": 4.799008469324519,
      "grad_norm": 11.53248405456543,
      "learning_rate": 2.889329587396023e-05,
      "loss": 1.5633,
      "step": 29040
    },
    {
      "epoch": 4.80066102045032,
      "grad_norm": 9.548233985900879,
      "learning_rate": 2.8884114655061605e-05,
      "loss": 1.4989,
      "step": 29050
    },
    {
      "epoch": 4.8023135715761205,
      "grad_norm": 5.22447395324707,
      "learning_rate": 2.8874933436162988e-05,
      "loss": 1.4038,
      "step": 29060
    },
    {
      "epoch": 4.8039661227019215,
      "grad_norm": 15.642428398132324,
      "learning_rate": 2.8865752217264364e-05,
      "loss": 1.4773,
      "step": 29070
    },
    {
      "epoch": 4.805618673827722,
      "grad_norm": 7.29582405090332,
      "learning_rate": 2.8856570998365746e-05,
      "loss": 1.5097,
      "step": 29080
    },
    {
      "epoch": 4.807271224953522,
      "grad_norm": 14.306315422058105,
      "learning_rate": 2.8847389779467122e-05,
      "loss": 1.4798,
      "step": 29090
    },
    {
      "epoch": 4.808923776079323,
      "grad_norm": 9.41856861114502,
      "learning_rate": 2.88382085605685e-05,
      "loss": 1.5179,
      "step": 29100
    },
    {
      "epoch": 4.810576327205123,
      "grad_norm": 14.0534086227417,
      "learning_rate": 2.8829027341669884e-05,
      "loss": 1.4123,
      "step": 29110
    },
    {
      "epoch": 4.812228878330924,
      "grad_norm": 73.24070739746094,
      "learning_rate": 2.881984612277126e-05,
      "loss": 1.5063,
      "step": 29120
    },
    {
      "epoch": 4.813881429456724,
      "grad_norm": 11.320660591125488,
      "learning_rate": 2.8810664903872643e-05,
      "loss": 1.5333,
      "step": 29130
    },
    {
      "epoch": 4.815533980582524,
      "grad_norm": 10.174736976623535,
      "learning_rate": 2.880148368497402e-05,
      "loss": 1.4865,
      "step": 29140
    },
    {
      "epoch": 4.817186531708325,
      "grad_norm": 13.212072372436523,
      "learning_rate": 2.8792302466075398e-05,
      "loss": 1.486,
      "step": 29150
    },
    {
      "epoch": 4.818839082834125,
      "grad_norm": 14.246745109558105,
      "learning_rate": 2.8783121247176777e-05,
      "loss": 1.5052,
      "step": 29160
    },
    {
      "epoch": 4.820491633959926,
      "grad_norm": 10.12642765045166,
      "learning_rate": 2.8773940028278156e-05,
      "loss": 1.3828,
      "step": 29170
    },
    {
      "epoch": 4.822144185085726,
      "grad_norm": 8.442862510681152,
      "learning_rate": 2.8764758809379532e-05,
      "loss": 1.5126,
      "step": 29180
    },
    {
      "epoch": 4.823796736211526,
      "grad_norm": 18.026945114135742,
      "learning_rate": 2.8755577590480915e-05,
      "loss": 1.5676,
      "step": 29190
    },
    {
      "epoch": 4.825449287337327,
      "grad_norm": 7.779977321624756,
      "learning_rate": 2.874639637158229e-05,
      "loss": 1.4621,
      "step": 29200
    },
    {
      "epoch": 4.827101838463127,
      "grad_norm": 21.727739334106445,
      "learning_rate": 2.8737215152683673e-05,
      "loss": 1.551,
      "step": 29210
    },
    {
      "epoch": 4.828754389588928,
      "grad_norm": 9.406710624694824,
      "learning_rate": 2.872803393378505e-05,
      "loss": 1.4938,
      "step": 29220
    },
    {
      "epoch": 4.830406940714728,
      "grad_norm": 9.602388381958008,
      "learning_rate": 2.8718852714886428e-05,
      "loss": 1.5477,
      "step": 29230
    },
    {
      "epoch": 4.832059491840528,
      "grad_norm": 13.313919067382812,
      "learning_rate": 2.870967149598781e-05,
      "loss": 1.5438,
      "step": 29240
    },
    {
      "epoch": 4.833712042966329,
      "grad_norm": 6.431525707244873,
      "learning_rate": 2.8700490277089187e-05,
      "loss": 1.5184,
      "step": 29250
    },
    {
      "epoch": 4.8353645940921295,
      "grad_norm": 6.1765313148498535,
      "learning_rate": 2.869130905819057e-05,
      "loss": 1.4919,
      "step": 29260
    },
    {
      "epoch": 4.8370171452179305,
      "grad_norm": 9.743697166442871,
      "learning_rate": 2.8682127839291945e-05,
      "loss": 1.4928,
      "step": 29270
    },
    {
      "epoch": 4.838669696343731,
      "grad_norm": 9.40673542022705,
      "learning_rate": 2.8672946620393324e-05,
      "loss": 1.4567,
      "step": 29280
    },
    {
      "epoch": 4.8403222474695315,
      "grad_norm": 9.90953254699707,
      "learning_rate": 2.8663765401494704e-05,
      "loss": 1.4985,
      "step": 29290
    },
    {
      "epoch": 4.841974798595332,
      "grad_norm": 15.230709075927734,
      "learning_rate": 2.8654584182596083e-05,
      "loss": 1.5292,
      "step": 29300
    },
    {
      "epoch": 4.843627349721132,
      "grad_norm": 8.818098068237305,
      "learning_rate": 2.864540296369746e-05,
      "loss": 1.4527,
      "step": 29310
    },
    {
      "epoch": 4.845279900846933,
      "grad_norm": 7.097944736480713,
      "learning_rate": 2.863622174479884e-05,
      "loss": 1.627,
      "step": 29320
    },
    {
      "epoch": 4.846932451972733,
      "grad_norm": 9.903128623962402,
      "learning_rate": 2.8627040525900217e-05,
      "loss": 1.6714,
      "step": 29330
    },
    {
      "epoch": 4.848585003098533,
      "grad_norm": 11.924121856689453,
      "learning_rate": 2.86178593070016e-05,
      "loss": 1.2808,
      "step": 29340
    },
    {
      "epoch": 4.850237554224334,
      "grad_norm": 15.464925765991211,
      "learning_rate": 2.860867808810298e-05,
      "loss": 1.5392,
      "step": 29350
    },
    {
      "epoch": 4.851890105350134,
      "grad_norm": 9.023946762084961,
      "learning_rate": 2.8599496869204355e-05,
      "loss": 1.5101,
      "step": 29360
    },
    {
      "epoch": 4.853542656475935,
      "grad_norm": 13.63484001159668,
      "learning_rate": 2.8590315650305738e-05,
      "loss": 1.4716,
      "step": 29370
    },
    {
      "epoch": 4.855195207601735,
      "grad_norm": 8.59793472290039,
      "learning_rate": 2.8581134431407114e-05,
      "loss": 1.4911,
      "step": 29380
    },
    {
      "epoch": 4.856847758727536,
      "grad_norm": 10.655295372009277,
      "learning_rate": 2.8571953212508496e-05,
      "loss": 1.4926,
      "step": 29390
    },
    {
      "epoch": 4.858500309853336,
      "grad_norm": 44.5798225402832,
      "learning_rate": 2.8562771993609872e-05,
      "loss": 1.4292,
      "step": 29400
    },
    {
      "epoch": 4.860152860979136,
      "grad_norm": 10.469300270080566,
      "learning_rate": 2.855359077471125e-05,
      "loss": 1.3958,
      "step": 29410
    },
    {
      "epoch": 4.861805412104937,
      "grad_norm": 125.58094024658203,
      "learning_rate": 2.854440955581263e-05,
      "loss": 1.5653,
      "step": 29420
    },
    {
      "epoch": 4.863457963230737,
      "grad_norm": 9.53484058380127,
      "learning_rate": 2.853522833691401e-05,
      "loss": 1.5401,
      "step": 29430
    },
    {
      "epoch": 4.865110514356537,
      "grad_norm": 8.112886428833008,
      "learning_rate": 2.8526047118015386e-05,
      "loss": 1.4984,
      "step": 29440
    },
    {
      "epoch": 4.866763065482338,
      "grad_norm": 19.068845748901367,
      "learning_rate": 2.851686589911677e-05,
      "loss": 1.4704,
      "step": 29450
    },
    {
      "epoch": 4.8684156166081385,
      "grad_norm": 14.235884666442871,
      "learning_rate": 2.850768468021815e-05,
      "loss": 1.4131,
      "step": 29460
    },
    {
      "epoch": 4.8700681677339395,
      "grad_norm": 8.707880973815918,
      "learning_rate": 2.8498503461319527e-05,
      "loss": 1.3683,
      "step": 29470
    },
    {
      "epoch": 4.87172071885974,
      "grad_norm": 8.955572128295898,
      "learning_rate": 2.8489322242420906e-05,
      "loss": 1.4594,
      "step": 29480
    },
    {
      "epoch": 4.8733732699855405,
      "grad_norm": 14.345805168151855,
      "learning_rate": 2.8480141023522282e-05,
      "loss": 1.5525,
      "step": 29490
    },
    {
      "epoch": 4.875025821111341,
      "grad_norm": 20.3609619140625,
      "learning_rate": 2.8470959804623665e-05,
      "loss": 1.5474,
      "step": 29500
    },
    {
      "epoch": 4.876678372237141,
      "grad_norm": 9.365708351135254,
      "learning_rate": 2.846177858572504e-05,
      "loss": 1.5406,
      "step": 29510
    },
    {
      "epoch": 4.878330923362942,
      "grad_norm": 8.881852149963379,
      "learning_rate": 2.8452597366826423e-05,
      "loss": 1.3521,
      "step": 29520
    },
    {
      "epoch": 4.879983474488742,
      "grad_norm": 46.270851135253906,
      "learning_rate": 2.84434161479278e-05,
      "loss": 1.6585,
      "step": 29530
    },
    {
      "epoch": 4.881636025614543,
      "grad_norm": 6.4464240074157715,
      "learning_rate": 2.843423492902918e-05,
      "loss": 1.4865,
      "step": 29540
    },
    {
      "epoch": 4.883288576740343,
      "grad_norm": 12.958675384521484,
      "learning_rate": 2.8425053710130557e-05,
      "loss": 1.5643,
      "step": 29550
    },
    {
      "epoch": 4.884941127866143,
      "grad_norm": 9.830168724060059,
      "learning_rate": 2.8415872491231937e-05,
      "loss": 1.3667,
      "step": 29560
    },
    {
      "epoch": 4.886593678991944,
      "grad_norm": 15.14609146118164,
      "learning_rate": 2.840669127233332e-05,
      "loss": 1.4713,
      "step": 29570
    },
    {
      "epoch": 4.888246230117744,
      "grad_norm": 10.329089164733887,
      "learning_rate": 2.8397510053434695e-05,
      "loss": 1.5543,
      "step": 29580
    },
    {
      "epoch": 4.889898781243545,
      "grad_norm": 6.754423141479492,
      "learning_rate": 2.8388328834536078e-05,
      "loss": 1.3762,
      "step": 29590
    },
    {
      "epoch": 4.891551332369345,
      "grad_norm": 23.90659523010254,
      "learning_rate": 2.8379147615637454e-05,
      "loss": 1.5471,
      "step": 29600
    },
    {
      "epoch": 4.893203883495145,
      "grad_norm": 14.84753131866455,
      "learning_rate": 2.8369966396738833e-05,
      "loss": 1.6663,
      "step": 29610
    },
    {
      "epoch": 4.894856434620946,
      "grad_norm": 8.822469711303711,
      "learning_rate": 2.836078517784021e-05,
      "loss": 1.3787,
      "step": 29620
    },
    {
      "epoch": 4.896508985746746,
      "grad_norm": 7.7165913581848145,
      "learning_rate": 2.835160395894159e-05,
      "loss": 1.4144,
      "step": 29630
    },
    {
      "epoch": 4.898161536872547,
      "grad_norm": 15.842850685119629,
      "learning_rate": 2.8342422740042967e-05,
      "loss": 1.4095,
      "step": 29640
    },
    {
      "epoch": 4.899814087998347,
      "grad_norm": 6.203104019165039,
      "learning_rate": 2.833324152114435e-05,
      "loss": 1.4753,
      "step": 29650
    },
    {
      "epoch": 4.901466639124148,
      "grad_norm": 9.8826904296875,
      "learning_rate": 2.8324060302245726e-05,
      "loss": 1.4384,
      "step": 29660
    },
    {
      "epoch": 4.9031191902499485,
      "grad_norm": 12.302288055419922,
      "learning_rate": 2.831487908334711e-05,
      "loss": 1.5356,
      "step": 29670
    },
    {
      "epoch": 4.904771741375749,
      "grad_norm": 11.829875946044922,
      "learning_rate": 2.8305697864448488e-05,
      "loss": 1.4053,
      "step": 29680
    },
    {
      "epoch": 4.9064242925015495,
      "grad_norm": 91.15592956542969,
      "learning_rate": 2.8296516645549863e-05,
      "loss": 1.5348,
      "step": 29690
    },
    {
      "epoch": 4.90807684362735,
      "grad_norm": 8.566816329956055,
      "learning_rate": 2.8287335426651246e-05,
      "loss": 1.4056,
      "step": 29700
    },
    {
      "epoch": 4.90972939475315,
      "grad_norm": 14.642806053161621,
      "learning_rate": 2.8278154207752622e-05,
      "loss": 1.5218,
      "step": 29710
    },
    {
      "epoch": 4.911381945878951,
      "grad_norm": 11.299127578735352,
      "learning_rate": 2.8268972988854005e-05,
      "loss": 1.5021,
      "step": 29720
    },
    {
      "epoch": 4.913034497004751,
      "grad_norm": 7.661716938018799,
      "learning_rate": 2.825979176995538e-05,
      "loss": 1.4599,
      "step": 29730
    },
    {
      "epoch": 4.914687048130552,
      "grad_norm": 16.636594772338867,
      "learning_rate": 2.825061055105676e-05,
      "loss": 1.4983,
      "step": 29740
    },
    {
      "epoch": 4.916339599256352,
      "grad_norm": 11.943872451782227,
      "learning_rate": 2.8241429332158136e-05,
      "loss": 1.6077,
      "step": 29750
    },
    {
      "epoch": 4.917992150382153,
      "grad_norm": 8.194153785705566,
      "learning_rate": 2.8232248113259518e-05,
      "loss": 1.4847,
      "step": 29760
    },
    {
      "epoch": 4.919644701507953,
      "grad_norm": 9.320284843444824,
      "learning_rate": 2.8223066894360894e-05,
      "loss": 1.5322,
      "step": 29770
    },
    {
      "epoch": 4.921297252633753,
      "grad_norm": 6.6243391036987305,
      "learning_rate": 2.8213885675462277e-05,
      "loss": 1.5005,
      "step": 29780
    },
    {
      "epoch": 4.922949803759554,
      "grad_norm": 6.813876628875732,
      "learning_rate": 2.8204704456563653e-05,
      "loss": 1.6577,
      "step": 29790
    },
    {
      "epoch": 4.924602354885354,
      "grad_norm": 7.4178876876831055,
      "learning_rate": 2.8195523237665035e-05,
      "loss": 1.4614,
      "step": 29800
    },
    {
      "epoch": 4.926254906011154,
      "grad_norm": 5.743941783905029,
      "learning_rate": 2.8186342018766414e-05,
      "loss": 1.448,
      "step": 29810
    },
    {
      "epoch": 4.927907457136955,
      "grad_norm": 17.722078323364258,
      "learning_rate": 2.817716079986779e-05,
      "loss": 1.5352,
      "step": 29820
    },
    {
      "epoch": 4.929560008262755,
      "grad_norm": 8.234187126159668,
      "learning_rate": 2.8167979580969173e-05,
      "loss": 1.4319,
      "step": 29830
    },
    {
      "epoch": 4.931212559388556,
      "grad_norm": 14.171329498291016,
      "learning_rate": 2.815879836207055e-05,
      "loss": 1.6987,
      "step": 29840
    },
    {
      "epoch": 4.932865110514356,
      "grad_norm": 18.609638214111328,
      "learning_rate": 2.814961714317193e-05,
      "loss": 1.4605,
      "step": 29850
    },
    {
      "epoch": 4.934517661640157,
      "grad_norm": 10.389782905578613,
      "learning_rate": 2.8140435924273307e-05,
      "loss": 1.571,
      "step": 29860
    },
    {
      "epoch": 4.9361702127659575,
      "grad_norm": 206.19345092773438,
      "learning_rate": 2.8131254705374687e-05,
      "loss": 1.4592,
      "step": 29870
    },
    {
      "epoch": 4.937822763891758,
      "grad_norm": 8.415513038635254,
      "learning_rate": 2.8122073486476062e-05,
      "loss": 1.4644,
      "step": 29880
    },
    {
      "epoch": 4.9394753150175585,
      "grad_norm": 13.903251647949219,
      "learning_rate": 2.8112892267577445e-05,
      "loss": 1.4461,
      "step": 29890
    },
    {
      "epoch": 4.941127866143359,
      "grad_norm": 7.16209602355957,
      "learning_rate": 2.810371104867882e-05,
      "loss": 1.5743,
      "step": 29900
    },
    {
      "epoch": 4.94278041726916,
      "grad_norm": 11.736125946044922,
      "learning_rate": 2.8094529829780204e-05,
      "loss": 1.6183,
      "step": 29910
    },
    {
      "epoch": 4.94443296839496,
      "grad_norm": 12.971633911132812,
      "learning_rate": 2.8085348610881583e-05,
      "loss": 1.4226,
      "step": 29920
    },
    {
      "epoch": 4.94608551952076,
      "grad_norm": 6.504984378814697,
      "learning_rate": 2.8076167391982962e-05,
      "loss": 1.4956,
      "step": 29930
    },
    {
      "epoch": 4.947738070646561,
      "grad_norm": 7.893176555633545,
      "learning_rate": 2.806698617308434e-05,
      "loss": 1.4615,
      "step": 29940
    },
    {
      "epoch": 4.949390621772361,
      "grad_norm": 8.680290222167969,
      "learning_rate": 2.8057804954185717e-05,
      "loss": 1.4396,
      "step": 29950
    },
    {
      "epoch": 4.951043172898162,
      "grad_norm": 11.371504783630371,
      "learning_rate": 2.80486237352871e-05,
      "loss": 1.4316,
      "step": 29960
    },
    {
      "epoch": 4.952695724023962,
      "grad_norm": 11.4708251953125,
      "learning_rate": 2.8039442516388476e-05,
      "loss": 1.5216,
      "step": 29970
    },
    {
      "epoch": 4.954348275149762,
      "grad_norm": 7.596912860870361,
      "learning_rate": 2.8030261297489858e-05,
      "loss": 1.538,
      "step": 29980
    },
    {
      "epoch": 4.956000826275563,
      "grad_norm": 12.889708518981934,
      "learning_rate": 2.8021080078591234e-05,
      "loss": 1.4701,
      "step": 29990
    },
    {
      "epoch": 4.957653377401363,
      "grad_norm": 23.94351577758789,
      "learning_rate": 2.8011898859692613e-05,
      "loss": 1.4761,
      "step": 30000
    },
    {
      "epoch": 4.959305928527164,
      "grad_norm": 10.514241218566895,
      "learning_rate": 2.800271764079399e-05,
      "loss": 1.5601,
      "step": 30010
    },
    {
      "epoch": 4.960958479652964,
      "grad_norm": 134.59942626953125,
      "learning_rate": 2.7993536421895372e-05,
      "loss": 1.5578,
      "step": 30020
    },
    {
      "epoch": 4.962611030778765,
      "grad_norm": 8.633646965026855,
      "learning_rate": 2.7984355202996755e-05,
      "loss": 1.3932,
      "step": 30030
    },
    {
      "epoch": 4.964263581904565,
      "grad_norm": 8.6370849609375,
      "learning_rate": 2.797517398409813e-05,
      "loss": 1.4556,
      "step": 30040
    },
    {
      "epoch": 4.965916133030365,
      "grad_norm": 7.444171905517578,
      "learning_rate": 2.796599276519951e-05,
      "loss": 1.4341,
      "step": 30050
    },
    {
      "epoch": 4.967568684156166,
      "grad_norm": 14.582514762878418,
      "learning_rate": 2.795681154630089e-05,
      "loss": 1.5428,
      "step": 30060
    },
    {
      "epoch": 4.9692212352819665,
      "grad_norm": 10.481239318847656,
      "learning_rate": 2.7947630327402268e-05,
      "loss": 1.4816,
      "step": 30070
    },
    {
      "epoch": 4.970873786407767,
      "grad_norm": 9.104494094848633,
      "learning_rate": 2.7938449108503644e-05,
      "loss": 1.4667,
      "step": 30080
    },
    {
      "epoch": 4.9725263375335675,
      "grad_norm": 14.128423690795898,
      "learning_rate": 2.7929267889605027e-05,
      "loss": 1.4932,
      "step": 30090
    },
    {
      "epoch": 4.974178888659368,
      "grad_norm": 8.712443351745605,
      "learning_rate": 2.7920086670706402e-05,
      "loss": 1.57,
      "step": 30100
    },
    {
      "epoch": 4.975831439785169,
      "grad_norm": 8.335115432739258,
      "learning_rate": 2.7910905451807785e-05,
      "loss": 1.4954,
      "step": 30110
    },
    {
      "epoch": 4.977483990910969,
      "grad_norm": 47.504981994628906,
      "learning_rate": 2.790172423290916e-05,
      "loss": 1.5175,
      "step": 30120
    },
    {
      "epoch": 4.97913654203677,
      "grad_norm": 8.322532653808594,
      "learning_rate": 2.789254301401054e-05,
      "loss": 1.5074,
      "step": 30130
    },
    {
      "epoch": 4.98078909316257,
      "grad_norm": 7.302855014801025,
      "learning_rate": 2.7883361795111923e-05,
      "loss": 1.6079,
      "step": 30140
    },
    {
      "epoch": 4.98244164428837,
      "grad_norm": 17.811246871948242,
      "learning_rate": 2.78741805762133e-05,
      "loss": 1.4465,
      "step": 30150
    },
    {
      "epoch": 4.984094195414171,
      "grad_norm": 16.895263671875,
      "learning_rate": 2.786499935731468e-05,
      "loss": 1.5104,
      "step": 30160
    },
    {
      "epoch": 4.985746746539971,
      "grad_norm": 31.71302604675293,
      "learning_rate": 2.7855818138416057e-05,
      "loss": 1.5411,
      "step": 30170
    },
    {
      "epoch": 4.987399297665771,
      "grad_norm": 19.637605667114258,
      "learning_rate": 2.7846636919517436e-05,
      "loss": 1.4842,
      "step": 30180
    },
    {
      "epoch": 4.989051848791572,
      "grad_norm": 134.47610473632812,
      "learning_rate": 2.7837455700618816e-05,
      "loss": 1.6898,
      "step": 30190
    },
    {
      "epoch": 4.990704399917372,
      "grad_norm": 18.40015983581543,
      "learning_rate": 2.7828274481720195e-05,
      "loss": 1.4774,
      "step": 30200
    },
    {
      "epoch": 4.992356951043173,
      "grad_norm": 8.392579078674316,
      "learning_rate": 2.781909326282157e-05,
      "loss": 1.3739,
      "step": 30210
    },
    {
      "epoch": 4.994009502168973,
      "grad_norm": 12.9806547164917,
      "learning_rate": 2.7809912043922953e-05,
      "loss": 1.4431,
      "step": 30220
    },
    {
      "epoch": 4.995662053294774,
      "grad_norm": 9.284590721130371,
      "learning_rate": 2.780073082502433e-05,
      "loss": 1.5775,
      "step": 30230
    },
    {
      "epoch": 4.997314604420574,
      "grad_norm": 13.390445709228516,
      "learning_rate": 2.7791549606125712e-05,
      "loss": 1.5908,
      "step": 30240
    },
    {
      "epoch": 4.998967155546374,
      "grad_norm": 9.49555778503418,
      "learning_rate": 2.778236838722709e-05,
      "loss": 1.5371,
      "step": 30250
    },
    {
      "epoch": 4.999958686221855,
      "eval_accuracy": 0.2968151510852603,
      "eval_loss": 2.206881046295166,
      "eval_runtime": 816.4999,
      "eval_samples_per_second": 34.533,
      "eval_steps_per_second": 8.633,
      "step": 30256
    },
    {
      "epoch": 5.000619706672175,
      "grad_norm": 15.309430122375488,
      "learning_rate": 2.7773187168328467e-05,
      "loss": 1.5734,
      "step": 30260
    },
    {
      "epoch": 5.0022722577979755,
      "grad_norm": 11.606006622314453,
      "learning_rate": 2.776400594942985e-05,
      "loss": 1.4502,
      "step": 30270
    },
    {
      "epoch": 5.0039248089237764,
      "grad_norm": 13.329957962036133,
      "learning_rate": 2.7754824730531226e-05,
      "loss": 1.3861,
      "step": 30280
    },
    {
      "epoch": 5.0055773600495765,
      "grad_norm": 35.278804779052734,
      "learning_rate": 2.7745643511632608e-05,
      "loss": 1.468,
      "step": 30290
    },
    {
      "epoch": 5.007229911175377,
      "grad_norm": 27.012672424316406,
      "learning_rate": 2.7736462292733984e-05,
      "loss": 1.6023,
      "step": 30300
    },
    {
      "epoch": 5.008882462301178,
      "grad_norm": 7.429107189178467,
      "learning_rate": 2.7727281073835367e-05,
      "loss": 1.4019,
      "step": 30310
    },
    {
      "epoch": 5.010535013426978,
      "grad_norm": 8.644866943359375,
      "learning_rate": 2.7718099854936743e-05,
      "loss": 1.3994,
      "step": 30320
    },
    {
      "epoch": 5.012187564552779,
      "grad_norm": 8.626925468444824,
      "learning_rate": 2.7708918636038122e-05,
      "loss": 1.6237,
      "step": 30330
    },
    {
      "epoch": 5.013840115678579,
      "grad_norm": 23.41725730895996,
      "learning_rate": 2.7699737417139498e-05,
      "loss": 1.3434,
      "step": 30340
    },
    {
      "epoch": 5.015492666804379,
      "grad_norm": 12.992168426513672,
      "learning_rate": 2.769055619824088e-05,
      "loss": 1.4989,
      "step": 30350
    },
    {
      "epoch": 5.01714521793018,
      "grad_norm": 19.700824737548828,
      "learning_rate": 2.7681374979342256e-05,
      "loss": 1.6157,
      "step": 30360
    },
    {
      "epoch": 5.01879776905598,
      "grad_norm": 35.20869827270508,
      "learning_rate": 2.767219376044364e-05,
      "loss": 1.3213,
      "step": 30370
    },
    {
      "epoch": 5.020450320181781,
      "grad_norm": 10.241279602050781,
      "learning_rate": 2.7663012541545018e-05,
      "loss": 1.5054,
      "step": 30380
    },
    {
      "epoch": 5.022102871307581,
      "grad_norm": 7.736096382141113,
      "learning_rate": 2.7653831322646394e-05,
      "loss": 1.5098,
      "step": 30390
    },
    {
      "epoch": 5.023755422433381,
      "grad_norm": 16.00613784790039,
      "learning_rate": 2.7644650103747777e-05,
      "loss": 1.4232,
      "step": 30400
    },
    {
      "epoch": 5.025407973559182,
      "grad_norm": 9.938050270080566,
      "learning_rate": 2.7635468884849152e-05,
      "loss": 1.472,
      "step": 30410
    },
    {
      "epoch": 5.027060524684982,
      "grad_norm": 7.0826640129089355,
      "learning_rate": 2.7626287665950535e-05,
      "loss": 1.4146,
      "step": 30420
    },
    {
      "epoch": 5.028713075810783,
      "grad_norm": 8.384482383728027,
      "learning_rate": 2.761710644705191e-05,
      "loss": 1.503,
      "step": 30430
    },
    {
      "epoch": 5.030365626936583,
      "grad_norm": 8.713443756103516,
      "learning_rate": 2.7607925228153293e-05,
      "loss": 1.2782,
      "step": 30440
    },
    {
      "epoch": 5.032018178062383,
      "grad_norm": 11.568880081176758,
      "learning_rate": 2.759874400925467e-05,
      "loss": 1.4832,
      "step": 30450
    },
    {
      "epoch": 5.033670729188184,
      "grad_norm": 11.942070960998535,
      "learning_rate": 2.758956279035605e-05,
      "loss": 1.4826,
      "step": 30460
    },
    {
      "epoch": 5.0353232803139845,
      "grad_norm": 8.392718315124512,
      "learning_rate": 2.7580381571457424e-05,
      "loss": 1.5613,
      "step": 30470
    },
    {
      "epoch": 5.0369758314397854,
      "grad_norm": 8.625317573547363,
      "learning_rate": 2.7571200352558807e-05,
      "loss": 1.5039,
      "step": 30480
    },
    {
      "epoch": 5.0386283825655855,
      "grad_norm": 34.13085174560547,
      "learning_rate": 2.756201913366019e-05,
      "loss": 1.5013,
      "step": 30490
    },
    {
      "epoch": 5.0402809336913865,
      "grad_norm": 9.937477111816406,
      "learning_rate": 2.7552837914761566e-05,
      "loss": 1.4878,
      "step": 30500
    },
    {
      "epoch": 5.041933484817187,
      "grad_norm": 10.615894317626953,
      "learning_rate": 2.7543656695862945e-05,
      "loss": 1.5141,
      "step": 30510
    },
    {
      "epoch": 5.043586035942987,
      "grad_norm": 9.882899284362793,
      "learning_rate": 2.753447547696432e-05,
      "loss": 1.4341,
      "step": 30520
    },
    {
      "epoch": 5.045238587068788,
      "grad_norm": 13.298185348510742,
      "learning_rate": 2.7525294258065703e-05,
      "loss": 1.3448,
      "step": 30530
    },
    {
      "epoch": 5.046891138194588,
      "grad_norm": 10.530019760131836,
      "learning_rate": 2.751611303916708e-05,
      "loss": 1.4876,
      "step": 30540
    },
    {
      "epoch": 5.048543689320389,
      "grad_norm": 7.8333892822265625,
      "learning_rate": 2.7506931820268462e-05,
      "loss": 1.3568,
      "step": 30550
    },
    {
      "epoch": 5.050196240446189,
      "grad_norm": 8.093653678894043,
      "learning_rate": 2.7497750601369838e-05,
      "loss": 1.4622,
      "step": 30560
    },
    {
      "epoch": 5.051848791571989,
      "grad_norm": 15.520524978637695,
      "learning_rate": 2.748856938247122e-05,
      "loss": 1.5338,
      "step": 30570
    },
    {
      "epoch": 5.05350134269779,
      "grad_norm": 7.4860382080078125,
      "learning_rate": 2.7479388163572596e-05,
      "loss": 1.3255,
      "step": 30580
    },
    {
      "epoch": 5.05515389382359,
      "grad_norm": 24.090126037597656,
      "learning_rate": 2.7470206944673975e-05,
      "loss": 1.4614,
      "step": 30590
    },
    {
      "epoch": 5.056806444949391,
      "grad_norm": 6.862220287322998,
      "learning_rate": 2.7461025725775358e-05,
      "loss": 1.4854,
      "step": 30600
    },
    {
      "epoch": 5.058458996075191,
      "grad_norm": 28.299789428710938,
      "learning_rate": 2.7451844506876734e-05,
      "loss": 1.4952,
      "step": 30610
    },
    {
      "epoch": 5.060111547200991,
      "grad_norm": 9.547993659973145,
      "learning_rate": 2.7442663287978117e-05,
      "loss": 1.4909,
      "step": 30620
    },
    {
      "epoch": 5.061764098326792,
      "grad_norm": 14.898249626159668,
      "learning_rate": 2.7433482069079492e-05,
      "loss": 1.4672,
      "step": 30630
    },
    {
      "epoch": 5.063416649452592,
      "grad_norm": 8.26840591430664,
      "learning_rate": 2.742430085018087e-05,
      "loss": 1.5186,
      "step": 30640
    },
    {
      "epoch": 5.065069200578393,
      "grad_norm": 7.622718811035156,
      "learning_rate": 2.7415119631282248e-05,
      "loss": 1.3791,
      "step": 30650
    },
    {
      "epoch": 5.066721751704193,
      "grad_norm": 9.911038398742676,
      "learning_rate": 2.740593841238363e-05,
      "loss": 1.3867,
      "step": 30660
    },
    {
      "epoch": 5.0683743028299935,
      "grad_norm": 21.617273330688477,
      "learning_rate": 2.7396757193485006e-05,
      "loss": 1.4964,
      "step": 30670
    },
    {
      "epoch": 5.070026853955794,
      "grad_norm": 10.754417419433594,
      "learning_rate": 2.738757597458639e-05,
      "loss": 1.4902,
      "step": 30680
    },
    {
      "epoch": 5.0716794050815945,
      "grad_norm": 15.153556823730469,
      "learning_rate": 2.7378394755687765e-05,
      "loss": 1.4118,
      "step": 30690
    },
    {
      "epoch": 5.0733319562073955,
      "grad_norm": 10.339966773986816,
      "learning_rate": 2.7369213536789147e-05,
      "loss": 1.5785,
      "step": 30700
    },
    {
      "epoch": 5.074984507333196,
      "grad_norm": 16.77724838256836,
      "learning_rate": 2.7360032317890526e-05,
      "loss": 1.3934,
      "step": 30710
    },
    {
      "epoch": 5.076637058458996,
      "grad_norm": 11.809764862060547,
      "learning_rate": 2.7350851098991902e-05,
      "loss": 1.4605,
      "step": 30720
    },
    {
      "epoch": 5.078289609584797,
      "grad_norm": 16.061250686645508,
      "learning_rate": 2.7341669880093285e-05,
      "loss": 1.5512,
      "step": 30730
    },
    {
      "epoch": 5.079942160710597,
      "grad_norm": 8.934623718261719,
      "learning_rate": 2.733248866119466e-05,
      "loss": 1.4989,
      "step": 30740
    },
    {
      "epoch": 5.081594711836398,
      "grad_norm": 8.88762378692627,
      "learning_rate": 2.7323307442296043e-05,
      "loss": 1.3776,
      "step": 30750
    },
    {
      "epoch": 5.083247262962198,
      "grad_norm": 14.036128044128418,
      "learning_rate": 2.731412622339742e-05,
      "loss": 1.5089,
      "step": 30760
    },
    {
      "epoch": 5.084899814087998,
      "grad_norm": 12.190908432006836,
      "learning_rate": 2.73049450044988e-05,
      "loss": 1.4895,
      "step": 30770
    },
    {
      "epoch": 5.086552365213799,
      "grad_norm": 10.279311180114746,
      "learning_rate": 2.7295763785600174e-05,
      "loss": 1.5135,
      "step": 30780
    },
    {
      "epoch": 5.088204916339599,
      "grad_norm": 7.8421549797058105,
      "learning_rate": 2.7286582566701557e-05,
      "loss": 1.4824,
      "step": 30790
    },
    {
      "epoch": 5.0898574674654,
      "grad_norm": 33.02604293823242,
      "learning_rate": 2.7277401347802933e-05,
      "loss": 1.4672,
      "step": 30800
    },
    {
      "epoch": 5.0915100185912,
      "grad_norm": 9.192763328552246,
      "learning_rate": 2.7268220128904315e-05,
      "loss": 1.5224,
      "step": 30810
    },
    {
      "epoch": 5.093162569717,
      "grad_norm": 14.467486381530762,
      "learning_rate": 2.7259038910005695e-05,
      "loss": 1.5533,
      "step": 30820
    },
    {
      "epoch": 5.094815120842801,
      "grad_norm": 12.854667663574219,
      "learning_rate": 2.7249857691107074e-05,
      "loss": 1.5342,
      "step": 30830
    },
    {
      "epoch": 5.096467671968601,
      "grad_norm": 8.334192276000977,
      "learning_rate": 2.7240676472208453e-05,
      "loss": 1.5695,
      "step": 30840
    },
    {
      "epoch": 5.098120223094402,
      "grad_norm": 7.920324325561523,
      "learning_rate": 2.723149525330983e-05,
      "loss": 1.5243,
      "step": 30850
    },
    {
      "epoch": 5.099772774220202,
      "grad_norm": 15.681721687316895,
      "learning_rate": 2.7222314034411212e-05,
      "loss": 1.5742,
      "step": 30860
    },
    {
      "epoch": 5.1014253253460025,
      "grad_norm": 10.732522010803223,
      "learning_rate": 2.7213132815512588e-05,
      "loss": 1.451,
      "step": 30870
    },
    {
      "epoch": 5.103077876471803,
      "grad_norm": 9.83503532409668,
      "learning_rate": 2.720395159661397e-05,
      "loss": 1.5081,
      "step": 30880
    },
    {
      "epoch": 5.1047304275976035,
      "grad_norm": 15.027039527893066,
      "learning_rate": 2.7194770377715346e-05,
      "loss": 1.5076,
      "step": 30890
    },
    {
      "epoch": 5.1063829787234045,
      "grad_norm": 6.365040302276611,
      "learning_rate": 2.7185589158816725e-05,
      "loss": 1.4941,
      "step": 30900
    },
    {
      "epoch": 5.108035529849205,
      "grad_norm": 14.915818214416504,
      "learning_rate": 2.71764079399181e-05,
      "loss": 1.5407,
      "step": 30910
    },
    {
      "epoch": 5.109688080975006,
      "grad_norm": 6.7997236251831055,
      "learning_rate": 2.7167226721019484e-05,
      "loss": 1.5268,
      "step": 30920
    },
    {
      "epoch": 5.111340632100806,
      "grad_norm": 9.963415145874023,
      "learning_rate": 2.715804550212086e-05,
      "loss": 1.3845,
      "step": 30930
    },
    {
      "epoch": 5.112993183226606,
      "grad_norm": 10.191669464111328,
      "learning_rate": 2.7148864283222242e-05,
      "loss": 1.4634,
      "step": 30940
    },
    {
      "epoch": 5.114645734352407,
      "grad_norm": 12.32084846496582,
      "learning_rate": 2.7139683064323625e-05,
      "loss": 1.4858,
      "step": 30950
    },
    {
      "epoch": 5.116298285478207,
      "grad_norm": 5.869612216949463,
      "learning_rate": 2.7130501845425e-05,
      "loss": 1.4629,
      "step": 30960
    },
    {
      "epoch": 5.117950836604008,
      "grad_norm": 17.75486183166504,
      "learning_rate": 2.712132062652638e-05,
      "loss": 1.4306,
      "step": 30970
    },
    {
      "epoch": 5.119603387729808,
      "grad_norm": 15.648809432983398,
      "learning_rate": 2.7112139407627756e-05,
      "loss": 1.4946,
      "step": 30980
    },
    {
      "epoch": 5.121255938855608,
      "grad_norm": 10.89876651763916,
      "learning_rate": 2.710295818872914e-05,
      "loss": 1.6039,
      "step": 30990
    },
    {
      "epoch": 5.122908489981409,
      "grad_norm": 13.772411346435547,
      "learning_rate": 2.7093776969830514e-05,
      "loss": 1.5348,
      "step": 31000
    },
    {
      "epoch": 5.124561041107209,
      "grad_norm": 11.758411407470703,
      "learning_rate": 2.7084595750931897e-05,
      "loss": 1.4812,
      "step": 31010
    },
    {
      "epoch": 5.12621359223301,
      "grad_norm": 27.372583389282227,
      "learning_rate": 2.7075414532033273e-05,
      "loss": 1.4999,
      "step": 31020
    },
    {
      "epoch": 5.12786614335881,
      "grad_norm": 9.395577430725098,
      "learning_rate": 2.7066233313134652e-05,
      "loss": 1.4854,
      "step": 31030
    },
    {
      "epoch": 5.12951869448461,
      "grad_norm": 24.79270362854004,
      "learning_rate": 2.7057052094236028e-05,
      "loss": 1.4465,
      "step": 31040
    },
    {
      "epoch": 5.131171245610411,
      "grad_norm": 10.64253044128418,
      "learning_rate": 2.704787087533741e-05,
      "loss": 1.5273,
      "step": 31050
    },
    {
      "epoch": 5.132823796736211,
      "grad_norm": 15.544374465942383,
      "learning_rate": 2.7038689656438793e-05,
      "loss": 1.4328,
      "step": 31060
    },
    {
      "epoch": 5.134476347862012,
      "grad_norm": 13.274250030517578,
      "learning_rate": 2.702950843754017e-05,
      "loss": 1.3941,
      "step": 31070
    },
    {
      "epoch": 5.136128898987812,
      "grad_norm": 7.694459438323975,
      "learning_rate": 2.7020327218641552e-05,
      "loss": 1.2814,
      "step": 31080
    },
    {
      "epoch": 5.1377814501136125,
      "grad_norm": 10.96912956237793,
      "learning_rate": 2.7011145999742928e-05,
      "loss": 1.5482,
      "step": 31090
    },
    {
      "epoch": 5.1394340012394135,
      "grad_norm": 107.31693267822266,
      "learning_rate": 2.7001964780844307e-05,
      "loss": 1.6004,
      "step": 31100
    },
    {
      "epoch": 5.141086552365214,
      "grad_norm": 10.470490455627441,
      "learning_rate": 2.6992783561945683e-05,
      "loss": 1.4453,
      "step": 31110
    },
    {
      "epoch": 5.142739103491015,
      "grad_norm": 8.99507999420166,
      "learning_rate": 2.6983602343047065e-05,
      "loss": 1.6031,
      "step": 31120
    },
    {
      "epoch": 5.144391654616815,
      "grad_norm": 7.556997776031494,
      "learning_rate": 2.697442112414844e-05,
      "loss": 1.6775,
      "step": 31130
    },
    {
      "epoch": 5.146044205742615,
      "grad_norm": 6.91307258605957,
      "learning_rate": 2.6965239905249824e-05,
      "loss": 1.4369,
      "step": 31140
    },
    {
      "epoch": 5.147696756868416,
      "grad_norm": 11.001294136047363,
      "learning_rate": 2.69560586863512e-05,
      "loss": 1.5253,
      "step": 31150
    },
    {
      "epoch": 5.149349307994216,
      "grad_norm": 11.226962089538574,
      "learning_rate": 2.694687746745258e-05,
      "loss": 1.4318,
      "step": 31160
    },
    {
      "epoch": 5.151001859120017,
      "grad_norm": 9.484665870666504,
      "learning_rate": 2.693769624855396e-05,
      "loss": 1.6958,
      "step": 31170
    },
    {
      "epoch": 5.152654410245817,
      "grad_norm": 11.47225570678711,
      "learning_rate": 2.6928515029655337e-05,
      "loss": 1.4713,
      "step": 31180
    },
    {
      "epoch": 5.154306961371617,
      "grad_norm": 13.216545104980469,
      "learning_rate": 2.691933381075672e-05,
      "loss": 1.462,
      "step": 31190
    },
    {
      "epoch": 5.155959512497418,
      "grad_norm": 14.571917533874512,
      "learning_rate": 2.6910152591858096e-05,
      "loss": 1.5572,
      "step": 31200
    },
    {
      "epoch": 5.157612063623218,
      "grad_norm": 6.982619285583496,
      "learning_rate": 2.690097137295948e-05,
      "loss": 1.5357,
      "step": 31210
    },
    {
      "epoch": 5.159264614749019,
      "grad_norm": 7.8256988525390625,
      "learning_rate": 2.6891790154060854e-05,
      "loss": 1.4597,
      "step": 31220
    },
    {
      "epoch": 5.160917165874819,
      "grad_norm": 11.033082962036133,
      "learning_rate": 2.6882608935162234e-05,
      "loss": 1.3216,
      "step": 31230
    },
    {
      "epoch": 5.162569717000619,
      "grad_norm": 9.309578895568848,
      "learning_rate": 2.687342771626361e-05,
      "loss": 1.3888,
      "step": 31240
    },
    {
      "epoch": 5.16422226812642,
      "grad_norm": 36.22593307495117,
      "learning_rate": 2.6864246497364992e-05,
      "loss": 1.3437,
      "step": 31250
    },
    {
      "epoch": 5.16587481925222,
      "grad_norm": 24.026330947875977,
      "learning_rate": 2.6855065278466368e-05,
      "loss": 1.4197,
      "step": 31260
    },
    {
      "epoch": 5.167527370378021,
      "grad_norm": 11.367629051208496,
      "learning_rate": 2.684588405956775e-05,
      "loss": 1.4089,
      "step": 31270
    },
    {
      "epoch": 5.169179921503821,
      "grad_norm": 8.361258506774902,
      "learning_rate": 2.683670284066913e-05,
      "loss": 1.4635,
      "step": 31280
    },
    {
      "epoch": 5.1708324726296215,
      "grad_norm": 25.029769897460938,
      "learning_rate": 2.6827521621770506e-05,
      "loss": 1.446,
      "step": 31290
    },
    {
      "epoch": 5.1724850237554225,
      "grad_norm": 11.334177017211914,
      "learning_rate": 2.681834040287189e-05,
      "loss": 1.4217,
      "step": 31300
    },
    {
      "epoch": 5.174137574881223,
      "grad_norm": 8.15962028503418,
      "learning_rate": 2.6809159183973264e-05,
      "loss": 1.4519,
      "step": 31310
    },
    {
      "epoch": 5.175790126007024,
      "grad_norm": 10.570094108581543,
      "learning_rate": 2.6799977965074647e-05,
      "loss": 1.3329,
      "step": 31320
    },
    {
      "epoch": 5.177442677132824,
      "grad_norm": 9.071272850036621,
      "learning_rate": 2.6790796746176023e-05,
      "loss": 1.558,
      "step": 31330
    },
    {
      "epoch": 5.179095228258625,
      "grad_norm": 9.73098373413086,
      "learning_rate": 2.6781615527277405e-05,
      "loss": 1.4698,
      "step": 31340
    },
    {
      "epoch": 5.180747779384425,
      "grad_norm": 13.9347505569458,
      "learning_rate": 2.677243430837878e-05,
      "loss": 1.5003,
      "step": 31350
    },
    {
      "epoch": 5.182400330510225,
      "grad_norm": 10.36252498626709,
      "learning_rate": 2.676325308948016e-05,
      "loss": 1.4587,
      "step": 31360
    },
    {
      "epoch": 5.184052881636026,
      "grad_norm": 16.580341339111328,
      "learning_rate": 2.6754071870581536e-05,
      "loss": 1.4521,
      "step": 31370
    },
    {
      "epoch": 5.185705432761826,
      "grad_norm": 14.264723777770996,
      "learning_rate": 2.674489065168292e-05,
      "loss": 1.3813,
      "step": 31380
    },
    {
      "epoch": 5.187357983887627,
      "grad_norm": 13.60984992980957,
      "learning_rate": 2.6735709432784295e-05,
      "loss": 1.6065,
      "step": 31390
    },
    {
      "epoch": 5.189010535013427,
      "grad_norm": 8.337096214294434,
      "learning_rate": 2.6726528213885678e-05,
      "loss": 1.4814,
      "step": 31400
    },
    {
      "epoch": 5.190663086139227,
      "grad_norm": 8.234798431396484,
      "learning_rate": 2.6717346994987057e-05,
      "loss": 1.5305,
      "step": 31410
    },
    {
      "epoch": 5.192315637265028,
      "grad_norm": 12.562371253967285,
      "learning_rate": 2.6708165776088433e-05,
      "loss": 1.5045,
      "step": 31420
    },
    {
      "epoch": 5.193968188390828,
      "grad_norm": 11.537344932556152,
      "learning_rate": 2.6698984557189815e-05,
      "loss": 1.5556,
      "step": 31430
    },
    {
      "epoch": 5.195620739516629,
      "grad_norm": 34.379127502441406,
      "learning_rate": 2.668980333829119e-05,
      "loss": 1.4086,
      "step": 31440
    },
    {
      "epoch": 5.197273290642429,
      "grad_norm": 10.825907707214355,
      "learning_rate": 2.6680622119392574e-05,
      "loss": 1.6994,
      "step": 31450
    },
    {
      "epoch": 5.198925841768229,
      "grad_norm": 9.560426712036133,
      "learning_rate": 2.667144090049395e-05,
      "loss": 1.606,
      "step": 31460
    },
    {
      "epoch": 5.20057839289403,
      "grad_norm": 7.058825492858887,
      "learning_rate": 2.6662259681595332e-05,
      "loss": 1.5243,
      "step": 31470
    },
    {
      "epoch": 5.20223094401983,
      "grad_norm": 8.447346687316895,
      "learning_rate": 2.6653078462696708e-05,
      "loss": 1.4428,
      "step": 31480
    },
    {
      "epoch": 5.203883495145631,
      "grad_norm": 8.007600784301758,
      "learning_rate": 2.6643897243798087e-05,
      "loss": 1.419,
      "step": 31490
    },
    {
      "epoch": 5.2055360462714315,
      "grad_norm": 10.7009916305542,
      "learning_rate": 2.6634716024899463e-05,
      "loss": 1.5057,
      "step": 31500
    },
    {
      "epoch": 5.207188597397232,
      "grad_norm": 10.528794288635254,
      "learning_rate": 2.6625534806000846e-05,
      "loss": 1.3964,
      "step": 31510
    },
    {
      "epoch": 5.208841148523033,
      "grad_norm": 11.727227210998535,
      "learning_rate": 2.661635358710223e-05,
      "loss": 1.5043,
      "step": 31520
    },
    {
      "epoch": 5.210493699648833,
      "grad_norm": 7.113905429840088,
      "learning_rate": 2.6607172368203604e-05,
      "loss": 1.4591,
      "step": 31530
    },
    {
      "epoch": 5.212146250774634,
      "grad_norm": 8.97098159790039,
      "learning_rate": 2.6597991149304984e-05,
      "loss": 1.5621,
      "step": 31540
    },
    {
      "epoch": 5.213798801900434,
      "grad_norm": 10.305506706237793,
      "learning_rate": 2.658880993040636e-05,
      "loss": 1.4602,
      "step": 31550
    },
    {
      "epoch": 5.215451353026234,
      "grad_norm": 13.641674995422363,
      "learning_rate": 2.6579628711507742e-05,
      "loss": 1.4489,
      "step": 31560
    },
    {
      "epoch": 5.217103904152035,
      "grad_norm": 9.43690299987793,
      "learning_rate": 2.6570447492609118e-05,
      "loss": 1.5889,
      "step": 31570
    },
    {
      "epoch": 5.218756455277835,
      "grad_norm": 6.873445510864258,
      "learning_rate": 2.65612662737105e-05,
      "loss": 1.5353,
      "step": 31580
    },
    {
      "epoch": 5.220409006403636,
      "grad_norm": 28.370594024658203,
      "learning_rate": 2.6552085054811876e-05,
      "loss": 1.5797,
      "step": 31590
    },
    {
      "epoch": 5.222061557529436,
      "grad_norm": 11.743301391601562,
      "learning_rate": 2.654290383591326e-05,
      "loss": 1.379,
      "step": 31600
    },
    {
      "epoch": 5.223714108655236,
      "grad_norm": 8.087701797485352,
      "learning_rate": 2.6533722617014635e-05,
      "loss": 1.5028,
      "step": 31610
    },
    {
      "epoch": 5.225366659781037,
      "grad_norm": 49.10771179199219,
      "learning_rate": 2.6524541398116014e-05,
      "loss": 1.4961,
      "step": 31620
    },
    {
      "epoch": 5.227019210906837,
      "grad_norm": 12.940389633178711,
      "learning_rate": 2.6515360179217397e-05,
      "loss": 1.5188,
      "step": 31630
    },
    {
      "epoch": 5.228671762032638,
      "grad_norm": 9.253148078918457,
      "learning_rate": 2.6506178960318773e-05,
      "loss": 1.5103,
      "step": 31640
    },
    {
      "epoch": 5.230324313158438,
      "grad_norm": 7.253983497619629,
      "learning_rate": 2.6496997741420155e-05,
      "loss": 1.5957,
      "step": 31650
    },
    {
      "epoch": 5.231976864284238,
      "grad_norm": 7.679978370666504,
      "learning_rate": 2.648781652252153e-05,
      "loss": 1.4185,
      "step": 31660
    },
    {
      "epoch": 5.233629415410039,
      "grad_norm": 13.257478713989258,
      "learning_rate": 2.647863530362291e-05,
      "loss": 1.5459,
      "step": 31670
    },
    {
      "epoch": 5.235281966535839,
      "grad_norm": 11.092429161071777,
      "learning_rate": 2.6469454084724286e-05,
      "loss": 1.4213,
      "step": 31680
    },
    {
      "epoch": 5.23693451766164,
      "grad_norm": 13.886821746826172,
      "learning_rate": 2.646027286582567e-05,
      "loss": 1.4144,
      "step": 31690
    },
    {
      "epoch": 5.2385870687874405,
      "grad_norm": 9.925250053405762,
      "learning_rate": 2.6451091646927045e-05,
      "loss": 1.3833,
      "step": 31700
    },
    {
      "epoch": 5.2402396199132415,
      "grad_norm": 8.703375816345215,
      "learning_rate": 2.6441910428028427e-05,
      "loss": 1.4315,
      "step": 31710
    },
    {
      "epoch": 5.241892171039042,
      "grad_norm": 91.34703826904297,
      "learning_rate": 2.6432729209129803e-05,
      "loss": 1.3998,
      "step": 31720
    },
    {
      "epoch": 5.243544722164842,
      "grad_norm": 15.27595043182373,
      "learning_rate": 2.6423547990231186e-05,
      "loss": 1.5865,
      "step": 31730
    },
    {
      "epoch": 5.245197273290643,
      "grad_norm": 11.300504684448242,
      "learning_rate": 2.6414366771332565e-05,
      "loss": 1.4581,
      "step": 31740
    },
    {
      "epoch": 5.246849824416443,
      "grad_norm": 10.756152153015137,
      "learning_rate": 2.640518555243394e-05,
      "loss": 1.5089,
      "step": 31750
    },
    {
      "epoch": 5.248502375542244,
      "grad_norm": 32.1532096862793,
      "learning_rate": 2.6396004333535324e-05,
      "loss": 1.4962,
      "step": 31760
    },
    {
      "epoch": 5.250154926668044,
      "grad_norm": 9.208897590637207,
      "learning_rate": 2.63868231146367e-05,
      "loss": 1.4471,
      "step": 31770
    },
    {
      "epoch": 5.251807477793844,
      "grad_norm": 17.185142517089844,
      "learning_rate": 2.6377641895738082e-05,
      "loss": 1.5346,
      "step": 31780
    },
    {
      "epoch": 5.253460028919645,
      "grad_norm": 15.642415046691895,
      "learning_rate": 2.6368460676839458e-05,
      "loss": 1.4897,
      "step": 31790
    },
    {
      "epoch": 5.255112580045445,
      "grad_norm": 11.783285140991211,
      "learning_rate": 2.6359279457940837e-05,
      "loss": 1.4354,
      "step": 31800
    },
    {
      "epoch": 5.256765131171246,
      "grad_norm": 10.766020774841309,
      "learning_rate": 2.6350098239042213e-05,
      "loss": 1.4611,
      "step": 31810
    },
    {
      "epoch": 5.258417682297046,
      "grad_norm": 9.75991439819336,
      "learning_rate": 2.6340917020143596e-05,
      "loss": 1.397,
      "step": 31820
    },
    {
      "epoch": 5.260070233422846,
      "grad_norm": 9.640849113464355,
      "learning_rate": 2.633173580124497e-05,
      "loss": 1.4211,
      "step": 31830
    },
    {
      "epoch": 5.261722784548647,
      "grad_norm": 49.9780387878418,
      "learning_rate": 2.6322554582346354e-05,
      "loss": 1.4722,
      "step": 31840
    },
    {
      "epoch": 5.263375335674447,
      "grad_norm": 11.358599662780762,
      "learning_rate": 2.6313373363447737e-05,
      "loss": 1.4281,
      "step": 31850
    },
    {
      "epoch": 5.265027886800248,
      "grad_norm": 6.540308475494385,
      "learning_rate": 2.6304192144549113e-05,
      "loss": 1.4257,
      "step": 31860
    },
    {
      "epoch": 5.266680437926048,
      "grad_norm": 11.856473922729492,
      "learning_rate": 2.6295010925650492e-05,
      "loss": 1.4528,
      "step": 31870
    },
    {
      "epoch": 5.268332989051848,
      "grad_norm": 6.613503456115723,
      "learning_rate": 2.6285829706751868e-05,
      "loss": 1.2896,
      "step": 31880
    },
    {
      "epoch": 5.269985540177649,
      "grad_norm": 10.662419319152832,
      "learning_rate": 2.627664848785325e-05,
      "loss": 1.4951,
      "step": 31890
    },
    {
      "epoch": 5.2716380913034495,
      "grad_norm": 13.057232856750488,
      "learning_rate": 2.6267467268954626e-05,
      "loss": 1.5273,
      "step": 31900
    },
    {
      "epoch": 5.2732906424292505,
      "grad_norm": 11.477456092834473,
      "learning_rate": 2.625828605005601e-05,
      "loss": 1.6712,
      "step": 31910
    },
    {
      "epoch": 5.274943193555051,
      "grad_norm": 7.808887481689453,
      "learning_rate": 2.6249104831157385e-05,
      "loss": 1.4828,
      "step": 31920
    },
    {
      "epoch": 5.276595744680851,
      "grad_norm": 8.068138122558594,
      "learning_rate": 2.6239923612258764e-05,
      "loss": 1.5149,
      "step": 31930
    },
    {
      "epoch": 5.278248295806652,
      "grad_norm": 11.689132690429688,
      "learning_rate": 2.623074239336014e-05,
      "loss": 1.4828,
      "step": 31940
    },
    {
      "epoch": 5.279900846932452,
      "grad_norm": 11.739019393920898,
      "learning_rate": 2.6221561174461523e-05,
      "loss": 1.5513,
      "step": 31950
    },
    {
      "epoch": 5.281553398058253,
      "grad_norm": 9.972562789916992,
      "learning_rate": 2.62123799555629e-05,
      "loss": 1.4665,
      "step": 31960
    },
    {
      "epoch": 5.283205949184053,
      "grad_norm": 9.730892181396484,
      "learning_rate": 2.620319873666428e-05,
      "loss": 1.6372,
      "step": 31970
    },
    {
      "epoch": 5.284858500309853,
      "grad_norm": 7.058382034301758,
      "learning_rate": 2.6194017517765664e-05,
      "loss": 1.3208,
      "step": 31980
    },
    {
      "epoch": 5.286511051435654,
      "grad_norm": 21.382532119750977,
      "learning_rate": 2.618483629886704e-05,
      "loss": 1.6161,
      "step": 31990
    },
    {
      "epoch": 5.288163602561454,
      "grad_norm": 12.467560768127441,
      "learning_rate": 2.617565507996842e-05,
      "loss": 1.5576,
      "step": 32000
    },
    {
      "epoch": 5.289816153687255,
      "grad_norm": 8.383124351501465,
      "learning_rate": 2.6166473861069795e-05,
      "loss": 1.5683,
      "step": 32010
    },
    {
      "epoch": 5.291468704813055,
      "grad_norm": 11.486825942993164,
      "learning_rate": 2.6157292642171177e-05,
      "loss": 1.5452,
      "step": 32020
    },
    {
      "epoch": 5.293121255938855,
      "grad_norm": 10.539969444274902,
      "learning_rate": 2.6148111423272553e-05,
      "loss": 1.5899,
      "step": 32030
    },
    {
      "epoch": 5.294773807064656,
      "grad_norm": 6.630085468292236,
      "learning_rate": 2.6138930204373936e-05,
      "loss": 1.6407,
      "step": 32040
    },
    {
      "epoch": 5.296426358190456,
      "grad_norm": 18.89915657043457,
      "learning_rate": 2.6129748985475312e-05,
      "loss": 1.4234,
      "step": 32050
    },
    {
      "epoch": 5.298078909316257,
      "grad_norm": 6.79443883895874,
      "learning_rate": 2.612056776657669e-05,
      "loss": 1.428,
      "step": 32060
    },
    {
      "epoch": 5.299731460442057,
      "grad_norm": 6.143259048461914,
      "learning_rate": 2.6111386547678067e-05,
      "loss": 1.4715,
      "step": 32070
    },
    {
      "epoch": 5.301384011567858,
      "grad_norm": 11.97968864440918,
      "learning_rate": 2.610220532877945e-05,
      "loss": 1.3861,
      "step": 32080
    },
    {
      "epoch": 5.303036562693658,
      "grad_norm": 6.659870624542236,
      "learning_rate": 2.6093024109880832e-05,
      "loss": 1.3774,
      "step": 32090
    },
    {
      "epoch": 5.3046891138194585,
      "grad_norm": 19.500179290771484,
      "learning_rate": 2.6083842890982208e-05,
      "loss": 1.4016,
      "step": 32100
    },
    {
      "epoch": 5.3063416649452595,
      "grad_norm": 10.958991050720215,
      "learning_rate": 2.607466167208359e-05,
      "loss": 1.2945,
      "step": 32110
    },
    {
      "epoch": 5.30799421607106,
      "grad_norm": 12.997007369995117,
      "learning_rate": 2.6065480453184966e-05,
      "loss": 1.4438,
      "step": 32120
    },
    {
      "epoch": 5.30964676719686,
      "grad_norm": 6.068414688110352,
      "learning_rate": 2.6056299234286346e-05,
      "loss": 1.4711,
      "step": 32130
    },
    {
      "epoch": 5.311299318322661,
      "grad_norm": 11.195246696472168,
      "learning_rate": 2.604711801538772e-05,
      "loss": 1.4093,
      "step": 32140
    },
    {
      "epoch": 5.312951869448461,
      "grad_norm": 40.626102447509766,
      "learning_rate": 2.6037936796489104e-05,
      "loss": 1.575,
      "step": 32150
    },
    {
      "epoch": 5.314604420574262,
      "grad_norm": 9.044241905212402,
      "learning_rate": 2.602875557759048e-05,
      "loss": 1.3712,
      "step": 32160
    },
    {
      "epoch": 5.316256971700062,
      "grad_norm": 13.25019359588623,
      "learning_rate": 2.6019574358691863e-05,
      "loss": 1.4445,
      "step": 32170
    },
    {
      "epoch": 5.317909522825863,
      "grad_norm": 20.42643165588379,
      "learning_rate": 2.601039313979324e-05,
      "loss": 1.3569,
      "step": 32180
    },
    {
      "epoch": 5.319562073951663,
      "grad_norm": 9.683372497558594,
      "learning_rate": 2.6001211920894618e-05,
      "loss": 1.5151,
      "step": 32190
    },
    {
      "epoch": 5.321214625077463,
      "grad_norm": 11.241543769836426,
      "learning_rate": 2.5992030701996e-05,
      "loss": 1.4425,
      "step": 32200
    },
    {
      "epoch": 5.322867176203264,
      "grad_norm": 12.246685981750488,
      "learning_rate": 2.5982849483097376e-05,
      "loss": 1.6109,
      "step": 32210
    },
    {
      "epoch": 5.324519727329064,
      "grad_norm": 30.807636260986328,
      "learning_rate": 2.597366826419876e-05,
      "loss": 1.4588,
      "step": 32220
    },
    {
      "epoch": 5.326172278454865,
      "grad_norm": 19.119007110595703,
      "learning_rate": 2.5964487045300135e-05,
      "loss": 1.4311,
      "step": 32230
    },
    {
      "epoch": 5.327824829580665,
      "grad_norm": 12.0513277053833,
      "learning_rate": 2.5955305826401517e-05,
      "loss": 1.5163,
      "step": 32240
    },
    {
      "epoch": 5.329477380706465,
      "grad_norm": 11.274496078491211,
      "learning_rate": 2.5946124607502893e-05,
      "loss": 1.3246,
      "step": 32250
    },
    {
      "epoch": 5.331129931832266,
      "grad_norm": 56.508384704589844,
      "learning_rate": 2.5936943388604273e-05,
      "loss": 1.5643,
      "step": 32260
    },
    {
      "epoch": 5.332782482958066,
      "grad_norm": 14.02528190612793,
      "learning_rate": 2.592776216970565e-05,
      "loss": 1.5164,
      "step": 32270
    },
    {
      "epoch": 5.334435034083867,
      "grad_norm": 16.86151123046875,
      "learning_rate": 2.591858095080703e-05,
      "loss": 1.4939,
      "step": 32280
    },
    {
      "epoch": 5.336087585209667,
      "grad_norm": 8.078414916992188,
      "learning_rate": 2.5909399731908407e-05,
      "loss": 1.4786,
      "step": 32290
    },
    {
      "epoch": 5.3377401363354675,
      "grad_norm": 11.507722854614258,
      "learning_rate": 2.590021851300979e-05,
      "loss": 1.4795,
      "step": 32300
    },
    {
      "epoch": 5.3393926874612685,
      "grad_norm": 20.504240036010742,
      "learning_rate": 2.589103729411117e-05,
      "loss": 1.5318,
      "step": 32310
    },
    {
      "epoch": 5.341045238587069,
      "grad_norm": 8.86858081817627,
      "learning_rate": 2.5881856075212545e-05,
      "loss": 1.5606,
      "step": 32320
    },
    {
      "epoch": 5.34269778971287,
      "grad_norm": 6.756838321685791,
      "learning_rate": 2.5872674856313927e-05,
      "loss": 1.4607,
      "step": 32330
    },
    {
      "epoch": 5.34435034083867,
      "grad_norm": 17.09296989440918,
      "learning_rate": 2.5863493637415303e-05,
      "loss": 1.5001,
      "step": 32340
    },
    {
      "epoch": 5.34600289196447,
      "grad_norm": 7.092482566833496,
      "learning_rate": 2.5854312418516686e-05,
      "loss": 1.5361,
      "step": 32350
    },
    {
      "epoch": 5.347655443090271,
      "grad_norm": 10.106451988220215,
      "learning_rate": 2.584513119961806e-05,
      "loss": 1.4626,
      "step": 32360
    },
    {
      "epoch": 5.349307994216071,
      "grad_norm": 7.64391565322876,
      "learning_rate": 2.5835949980719444e-05,
      "loss": 1.5781,
      "step": 32370
    },
    {
      "epoch": 5.350960545341872,
      "grad_norm": 10.09816837310791,
      "learning_rate": 2.582676876182082e-05,
      "loss": 1.5339,
      "step": 32380
    },
    {
      "epoch": 5.352613096467672,
      "grad_norm": 12.53437328338623,
      "learning_rate": 2.58175875429222e-05,
      "loss": 1.4137,
      "step": 32390
    },
    {
      "epoch": 5.354265647593472,
      "grad_norm": 9.417365074157715,
      "learning_rate": 2.5808406324023575e-05,
      "loss": 1.4312,
      "step": 32400
    },
    {
      "epoch": 5.355918198719273,
      "grad_norm": 8.617480278015137,
      "learning_rate": 2.5799225105124958e-05,
      "loss": 1.527,
      "step": 32410
    },
    {
      "epoch": 5.357570749845073,
      "grad_norm": 13.1574068069458,
      "learning_rate": 2.579004388622634e-05,
      "loss": 1.6434,
      "step": 32420
    },
    {
      "epoch": 5.359223300970874,
      "grad_norm": 7.331067085266113,
      "learning_rate": 2.5780862667327716e-05,
      "loss": 1.4821,
      "step": 32430
    },
    {
      "epoch": 5.360875852096674,
      "grad_norm": 9.73812484741211,
      "learning_rate": 2.5771681448429096e-05,
      "loss": 1.4768,
      "step": 32440
    },
    {
      "epoch": 5.362528403222475,
      "grad_norm": 9.941301345825195,
      "learning_rate": 2.576250022953047e-05,
      "loss": 1.5879,
      "step": 32450
    },
    {
      "epoch": 5.364180954348275,
      "grad_norm": 12.169427871704102,
      "learning_rate": 2.5753319010631854e-05,
      "loss": 1.4923,
      "step": 32460
    },
    {
      "epoch": 5.365833505474075,
      "grad_norm": 12.982132911682129,
      "learning_rate": 2.574413779173323e-05,
      "loss": 1.3733,
      "step": 32470
    },
    {
      "epoch": 5.367486056599876,
      "grad_norm": 7.65261173248291,
      "learning_rate": 2.5734956572834613e-05,
      "loss": 1.436,
      "step": 32480
    },
    {
      "epoch": 5.369138607725676,
      "grad_norm": 9.102943420410156,
      "learning_rate": 2.572577535393599e-05,
      "loss": 1.515,
      "step": 32490
    },
    {
      "epoch": 5.3707911588514765,
      "grad_norm": 5.440056800842285,
      "learning_rate": 2.571659413503737e-05,
      "loss": 1.4411,
      "step": 32500
    },
    {
      "epoch": 5.3724437099772775,
      "grad_norm": 6.815498352050781,
      "learning_rate": 2.5707412916138747e-05,
      "loss": 1.4195,
      "step": 32510
    },
    {
      "epoch": 5.374096261103078,
      "grad_norm": 9.051548957824707,
      "learning_rate": 2.5698231697240126e-05,
      "loss": 1.4627,
      "step": 32520
    },
    {
      "epoch": 5.375748812228879,
      "grad_norm": 10.097661972045898,
      "learning_rate": 2.5689050478341502e-05,
      "loss": 1.5081,
      "step": 32530
    },
    {
      "epoch": 5.377401363354679,
      "grad_norm": 8.773504257202148,
      "learning_rate": 2.5679869259442885e-05,
      "loss": 1.5666,
      "step": 32540
    },
    {
      "epoch": 5.37905391448048,
      "grad_norm": 9.091877937316895,
      "learning_rate": 2.5670688040544267e-05,
      "loss": 1.5096,
      "step": 32550
    },
    {
      "epoch": 5.38070646560628,
      "grad_norm": 6.017947196960449,
      "learning_rate": 2.5661506821645643e-05,
      "loss": 1.4604,
      "step": 32560
    },
    {
      "epoch": 5.38235901673208,
      "grad_norm": 8.248211860656738,
      "learning_rate": 2.5652325602747022e-05,
      "loss": 1.4782,
      "step": 32570
    },
    {
      "epoch": 5.384011567857881,
      "grad_norm": 8.899044036865234,
      "learning_rate": 2.5643144383848398e-05,
      "loss": 1.4193,
      "step": 32580
    },
    {
      "epoch": 5.385664118983681,
      "grad_norm": 10.997679710388184,
      "learning_rate": 2.563396316494978e-05,
      "loss": 1.469,
      "step": 32590
    },
    {
      "epoch": 5.387316670109482,
      "grad_norm": 14.518561363220215,
      "learning_rate": 2.5624781946051157e-05,
      "loss": 1.447,
      "step": 32600
    },
    {
      "epoch": 5.388969221235282,
      "grad_norm": 12.715332984924316,
      "learning_rate": 2.561560072715254e-05,
      "loss": 1.55,
      "step": 32610
    },
    {
      "epoch": 5.390621772361082,
      "grad_norm": 74.79273223876953,
      "learning_rate": 2.5606419508253915e-05,
      "loss": 1.3885,
      "step": 32620
    },
    {
      "epoch": 5.392274323486883,
      "grad_norm": 13.330283164978027,
      "learning_rate": 2.5597238289355298e-05,
      "loss": 1.5348,
      "step": 32630
    },
    {
      "epoch": 5.393926874612683,
      "grad_norm": 11.709308624267578,
      "learning_rate": 2.5588057070456674e-05,
      "loss": 1.5496,
      "step": 32640
    },
    {
      "epoch": 5.395579425738484,
      "grad_norm": 7.356724739074707,
      "learning_rate": 2.5578875851558053e-05,
      "loss": 1.3092,
      "step": 32650
    },
    {
      "epoch": 5.397231976864284,
      "grad_norm": 7.576127529144287,
      "learning_rate": 2.5569694632659436e-05,
      "loss": 1.3922,
      "step": 32660
    },
    {
      "epoch": 5.398884527990084,
      "grad_norm": 8.769586563110352,
      "learning_rate": 2.556051341376081e-05,
      "loss": 1.4997,
      "step": 32670
    },
    {
      "epoch": 5.400537079115885,
      "grad_norm": 8.266812324523926,
      "learning_rate": 2.5551332194862194e-05,
      "loss": 1.504,
      "step": 32680
    },
    {
      "epoch": 5.402189630241685,
      "grad_norm": 11.372197151184082,
      "learning_rate": 2.554215097596357e-05,
      "loss": 1.4409,
      "step": 32690
    },
    {
      "epoch": 5.403842181367486,
      "grad_norm": 19.209890365600586,
      "learning_rate": 2.553296975706495e-05,
      "loss": 1.5131,
      "step": 32700
    },
    {
      "epoch": 5.4054947324932865,
      "grad_norm": 11.207270622253418,
      "learning_rate": 2.5523788538166325e-05,
      "loss": 1.4532,
      "step": 32710
    },
    {
      "epoch": 5.407147283619087,
      "grad_norm": 7.834964752197266,
      "learning_rate": 2.5514607319267708e-05,
      "loss": 1.4809,
      "step": 32720
    },
    {
      "epoch": 5.408799834744888,
      "grad_norm": 10.717996597290039,
      "learning_rate": 2.5505426100369084e-05,
      "loss": 1.4442,
      "step": 32730
    },
    {
      "epoch": 5.410452385870688,
      "grad_norm": 7.571774005889893,
      "learning_rate": 2.5496244881470466e-05,
      "loss": 1.5859,
      "step": 32740
    },
    {
      "epoch": 5.412104936996489,
      "grad_norm": 9.474056243896484,
      "learning_rate": 2.5487063662571842e-05,
      "loss": 1.5151,
      "step": 32750
    },
    {
      "epoch": 5.413757488122289,
      "grad_norm": 8.241455078125,
      "learning_rate": 2.5477882443673225e-05,
      "loss": 1.4181,
      "step": 32760
    },
    {
      "epoch": 5.415410039248089,
      "grad_norm": 8.962974548339844,
      "learning_rate": 2.5468701224774604e-05,
      "loss": 1.4505,
      "step": 32770
    },
    {
      "epoch": 5.41706259037389,
      "grad_norm": 9.275091171264648,
      "learning_rate": 2.545952000587598e-05,
      "loss": 1.4723,
      "step": 32780
    },
    {
      "epoch": 5.41871514149969,
      "grad_norm": 9.588011741638184,
      "learning_rate": 2.5450338786977362e-05,
      "loss": 1.5176,
      "step": 32790
    },
    {
      "epoch": 5.420367692625491,
      "grad_norm": 6.4946441650390625,
      "learning_rate": 2.544115756807874e-05,
      "loss": 1.4376,
      "step": 32800
    },
    {
      "epoch": 5.422020243751291,
      "grad_norm": 13.272873878479004,
      "learning_rate": 2.543197634918012e-05,
      "loss": 1.5343,
      "step": 32810
    },
    {
      "epoch": 5.423672794877092,
      "grad_norm": 9.417807579040527,
      "learning_rate": 2.5422795130281497e-05,
      "loss": 1.4349,
      "step": 32820
    },
    {
      "epoch": 5.425325346002892,
      "grad_norm": 16.028039932250977,
      "learning_rate": 2.5413613911382876e-05,
      "loss": 1.464,
      "step": 32830
    },
    {
      "epoch": 5.426977897128692,
      "grad_norm": 6.217875003814697,
      "learning_rate": 2.5404432692484252e-05,
      "loss": 1.4224,
      "step": 32840
    },
    {
      "epoch": 5.428630448254493,
      "grad_norm": 4.6173529624938965,
      "learning_rate": 2.5395251473585635e-05,
      "loss": 1.389,
      "step": 32850
    },
    {
      "epoch": 5.430282999380293,
      "grad_norm": 16.59287452697754,
      "learning_rate": 2.538607025468701e-05,
      "loss": 1.406,
      "step": 32860
    },
    {
      "epoch": 5.431935550506093,
      "grad_norm": 28.710718154907227,
      "learning_rate": 2.5376889035788393e-05,
      "loss": 1.5599,
      "step": 32870
    },
    {
      "epoch": 5.433588101631894,
      "grad_norm": 6.974539756774902,
      "learning_rate": 2.5367707816889776e-05,
      "loss": 1.5168,
      "step": 32880
    },
    {
      "epoch": 5.435240652757694,
      "grad_norm": 18.166751861572266,
      "learning_rate": 2.535852659799115e-05,
      "loss": 1.4325,
      "step": 32890
    },
    {
      "epoch": 5.436893203883495,
      "grad_norm": 13.46557331085205,
      "learning_rate": 2.534934537909253e-05,
      "loss": 1.3934,
      "step": 32900
    },
    {
      "epoch": 5.4385457550092955,
      "grad_norm": 11.49219036102295,
      "learning_rate": 2.5340164160193907e-05,
      "loss": 1.4228,
      "step": 32910
    },
    {
      "epoch": 5.4401983061350965,
      "grad_norm": 12.140527725219727,
      "learning_rate": 2.533098294129529e-05,
      "loss": 1.5574,
      "step": 32920
    },
    {
      "epoch": 5.441850857260897,
      "grad_norm": 13.28624153137207,
      "learning_rate": 2.5321801722396665e-05,
      "loss": 1.4256,
      "step": 32930
    },
    {
      "epoch": 5.443503408386697,
      "grad_norm": 19.24834632873535,
      "learning_rate": 2.5312620503498048e-05,
      "loss": 1.4024,
      "step": 32940
    },
    {
      "epoch": 5.445155959512498,
      "grad_norm": 12.811565399169922,
      "learning_rate": 2.5303439284599424e-05,
      "loss": 1.5038,
      "step": 32950
    },
    {
      "epoch": 5.446808510638298,
      "grad_norm": 135.61062622070312,
      "learning_rate": 2.5294258065700803e-05,
      "loss": 1.5554,
      "step": 32960
    },
    {
      "epoch": 5.448461061764099,
      "grad_norm": 13.871460914611816,
      "learning_rate": 2.528507684680218e-05,
      "loss": 1.5125,
      "step": 32970
    },
    {
      "epoch": 5.450113612889899,
      "grad_norm": 7.033801078796387,
      "learning_rate": 2.527589562790356e-05,
      "loss": 1.6154,
      "step": 32980
    },
    {
      "epoch": 5.451766164015699,
      "grad_norm": 17.240575790405273,
      "learning_rate": 2.5266714409004944e-05,
      "loss": 1.4924,
      "step": 32990
    },
    {
      "epoch": 5.4534187151415,
      "grad_norm": 11.401257514953613,
      "learning_rate": 2.525753319010632e-05,
      "loss": 1.449,
      "step": 33000
    },
    {
      "epoch": 5.4550712662673,
      "grad_norm": 23.046972274780273,
      "learning_rate": 2.5248351971207703e-05,
      "loss": 1.3946,
      "step": 33010
    },
    {
      "epoch": 5.456723817393101,
      "grad_norm": 13.394057273864746,
      "learning_rate": 2.523917075230908e-05,
      "loss": 1.4518,
      "step": 33020
    },
    {
      "epoch": 5.458376368518901,
      "grad_norm": 7.417761325836182,
      "learning_rate": 2.5229989533410458e-05,
      "loss": 1.5553,
      "step": 33030
    },
    {
      "epoch": 5.460028919644701,
      "grad_norm": 15.697338104248047,
      "learning_rate": 2.5220808314511834e-05,
      "loss": 1.3501,
      "step": 33040
    },
    {
      "epoch": 5.461681470770502,
      "grad_norm": 13.63474178314209,
      "learning_rate": 2.5211627095613216e-05,
      "loss": 1.5208,
      "step": 33050
    },
    {
      "epoch": 5.463334021896302,
      "grad_norm": 15.581018447875977,
      "learning_rate": 2.5202445876714592e-05,
      "loss": 1.3335,
      "step": 33060
    },
    {
      "epoch": 5.464986573022103,
      "grad_norm": 15.496585845947266,
      "learning_rate": 2.5193264657815975e-05,
      "loss": 1.3965,
      "step": 33070
    },
    {
      "epoch": 5.466639124147903,
      "grad_norm": 11.815439224243164,
      "learning_rate": 2.518408343891735e-05,
      "loss": 1.4156,
      "step": 33080
    },
    {
      "epoch": 5.468291675273703,
      "grad_norm": 15.371979713439941,
      "learning_rate": 2.517490222001873e-05,
      "loss": 1.452,
      "step": 33090
    },
    {
      "epoch": 5.469944226399504,
      "grad_norm": 14.599471092224121,
      "learning_rate": 2.5165721001120106e-05,
      "loss": 1.4849,
      "step": 33100
    },
    {
      "epoch": 5.4715967775253045,
      "grad_norm": 16.83903694152832,
      "learning_rate": 2.5156539782221488e-05,
      "loss": 1.3825,
      "step": 33110
    },
    {
      "epoch": 5.4732493286511055,
      "grad_norm": 11.542963027954102,
      "learning_rate": 2.514735856332287e-05,
      "loss": 1.4628,
      "step": 33120
    },
    {
      "epoch": 5.474901879776906,
      "grad_norm": 44.345481872558594,
      "learning_rate": 2.5138177344424247e-05,
      "loss": 1.3834,
      "step": 33130
    },
    {
      "epoch": 5.476554430902706,
      "grad_norm": 13.720561981201172,
      "learning_rate": 2.512899612552563e-05,
      "loss": 1.3351,
      "step": 33140
    },
    {
      "epoch": 5.478206982028507,
      "grad_norm": 11.957843780517578,
      "learning_rate": 2.5119814906627005e-05,
      "loss": 1.4726,
      "step": 33150
    },
    {
      "epoch": 5.479859533154307,
      "grad_norm": 11.13582706451416,
      "learning_rate": 2.5110633687728384e-05,
      "loss": 1.3245,
      "step": 33160
    },
    {
      "epoch": 5.481512084280108,
      "grad_norm": 33.86077117919922,
      "learning_rate": 2.510145246882976e-05,
      "loss": 1.6398,
      "step": 33170
    },
    {
      "epoch": 5.483164635405908,
      "grad_norm": 19.655872344970703,
      "learning_rate": 2.5092271249931143e-05,
      "loss": 1.5043,
      "step": 33180
    },
    {
      "epoch": 5.484817186531708,
      "grad_norm": 10.309099197387695,
      "learning_rate": 2.508309003103252e-05,
      "loss": 1.4599,
      "step": 33190
    },
    {
      "epoch": 5.486469737657509,
      "grad_norm": 11.701153755187988,
      "learning_rate": 2.50739088121339e-05,
      "loss": 1.4324,
      "step": 33200
    },
    {
      "epoch": 5.488122288783309,
      "grad_norm": 12.09907054901123,
      "learning_rate": 2.5064727593235277e-05,
      "loss": 1.4819,
      "step": 33210
    },
    {
      "epoch": 5.48977483990911,
      "grad_norm": 11.532211303710938,
      "learning_rate": 2.5055546374336657e-05,
      "loss": 1.4336,
      "step": 33220
    },
    {
      "epoch": 5.49142739103491,
      "grad_norm": 7.926141738891602,
      "learning_rate": 2.504636515543804e-05,
      "loss": 1.4861,
      "step": 33230
    },
    {
      "epoch": 5.49307994216071,
      "grad_norm": 9.352069854736328,
      "learning_rate": 2.5037183936539415e-05,
      "loss": 1.403,
      "step": 33240
    },
    {
      "epoch": 5.494732493286511,
      "grad_norm": 8.540033340454102,
      "learning_rate": 2.5028002717640798e-05,
      "loss": 1.4863,
      "step": 33250
    },
    {
      "epoch": 5.496385044412311,
      "grad_norm": 119.00782775878906,
      "learning_rate": 2.5018821498742174e-05,
      "loss": 1.4733,
      "step": 33260
    },
    {
      "epoch": 5.498037595538112,
      "grad_norm": 9.270196914672852,
      "learning_rate": 2.5009640279843556e-05,
      "loss": 1.5347,
      "step": 33270
    },
    {
      "epoch": 5.499690146663912,
      "grad_norm": 7.050109386444092,
      "learning_rate": 2.5000459060944932e-05,
      "loss": 1.4698,
      "step": 33280
    },
    {
      "epoch": 5.501342697789713,
      "grad_norm": 15.336756706237793,
      "learning_rate": 2.499127784204631e-05,
      "loss": 1.3084,
      "step": 33290
    },
    {
      "epoch": 5.502995248915513,
      "grad_norm": 7.765586853027344,
      "learning_rate": 2.498209662314769e-05,
      "loss": 1.46,
      "step": 33300
    },
    {
      "epoch": 5.5046478000413135,
      "grad_norm": 22.45989227294922,
      "learning_rate": 2.497291540424907e-05,
      "loss": 1.5559,
      "step": 33310
    },
    {
      "epoch": 5.5063003511671145,
      "grad_norm": 16.461627960205078,
      "learning_rate": 2.496373418535045e-05,
      "loss": 1.5358,
      "step": 33320
    },
    {
      "epoch": 5.507952902292915,
      "grad_norm": 13.17129898071289,
      "learning_rate": 2.4954552966451828e-05,
      "loss": 1.4679,
      "step": 33330
    },
    {
      "epoch": 5.509605453418715,
      "grad_norm": 8.983606338500977,
      "learning_rate": 2.4945371747553208e-05,
      "loss": 1.4101,
      "step": 33340
    },
    {
      "epoch": 5.511258004544516,
      "grad_norm": 16.77591323852539,
      "learning_rate": 2.4936190528654583e-05,
      "loss": 1.4592,
      "step": 33350
    },
    {
      "epoch": 5.512910555670316,
      "grad_norm": 20.5835018157959,
      "learning_rate": 2.4927009309755963e-05,
      "loss": 1.4214,
      "step": 33360
    },
    {
      "epoch": 5.514563106796117,
      "grad_norm": 14.100784301757812,
      "learning_rate": 2.4917828090857342e-05,
      "loss": 1.4042,
      "step": 33370
    },
    {
      "epoch": 5.516215657921917,
      "grad_norm": 8.669175148010254,
      "learning_rate": 2.490864687195872e-05,
      "loss": 1.3463,
      "step": 33380
    },
    {
      "epoch": 5.517868209047718,
      "grad_norm": 19.604291915893555,
      "learning_rate": 2.4899465653060104e-05,
      "loss": 1.3455,
      "step": 33390
    },
    {
      "epoch": 5.519520760173518,
      "grad_norm": 13.911442756652832,
      "learning_rate": 2.4890284434161483e-05,
      "loss": 1.5867,
      "step": 33400
    },
    {
      "epoch": 5.521173311299318,
      "grad_norm": 9.792315483093262,
      "learning_rate": 2.488110321526286e-05,
      "loss": 1.4837,
      "step": 33410
    },
    {
      "epoch": 5.522825862425119,
      "grad_norm": 11.161901473999023,
      "learning_rate": 2.4871921996364238e-05,
      "loss": 1.4366,
      "step": 33420
    },
    {
      "epoch": 5.524478413550919,
      "grad_norm": 12.950854301452637,
      "learning_rate": 2.4862740777465617e-05,
      "loss": 1.3821,
      "step": 33430
    },
    {
      "epoch": 5.52613096467672,
      "grad_norm": 49.98655700683594,
      "learning_rate": 2.4853559558566997e-05,
      "loss": 1.5996,
      "step": 33440
    },
    {
      "epoch": 5.52778351580252,
      "grad_norm": 12.21288013458252,
      "learning_rate": 2.4844378339668376e-05,
      "loss": 1.5804,
      "step": 33450
    },
    {
      "epoch": 5.52943606692832,
      "grad_norm": 7.666236400604248,
      "learning_rate": 2.4835197120769755e-05,
      "loss": 1.665,
      "step": 33460
    },
    {
      "epoch": 5.531088618054121,
      "grad_norm": 11.325262069702148,
      "learning_rate": 2.4826015901871134e-05,
      "loss": 1.5233,
      "step": 33470
    },
    {
      "epoch": 5.532741169179921,
      "grad_norm": 9.03773021697998,
      "learning_rate": 2.481683468297251e-05,
      "loss": 1.5621,
      "step": 33480
    },
    {
      "epoch": 5.534393720305722,
      "grad_norm": 9.082817077636719,
      "learning_rate": 2.480765346407389e-05,
      "loss": 1.5058,
      "step": 33490
    },
    {
      "epoch": 5.536046271431522,
      "grad_norm": 7.868617534637451,
      "learning_rate": 2.479847224517527e-05,
      "loss": 1.4172,
      "step": 33500
    },
    {
      "epoch": 5.5376988225573225,
      "grad_norm": 16.261873245239258,
      "learning_rate": 2.478929102627665e-05,
      "loss": 1.4935,
      "step": 33510
    },
    {
      "epoch": 5.5393513736831235,
      "grad_norm": 8.341659545898438,
      "learning_rate": 2.478010980737803e-05,
      "loss": 1.6001,
      "step": 33520
    },
    {
      "epoch": 5.541003924808924,
      "grad_norm": 8.13469123840332,
      "learning_rate": 2.477092858847941e-05,
      "loss": 1.5278,
      "step": 33530
    },
    {
      "epoch": 5.5426564759347245,
      "grad_norm": 8.067828178405762,
      "learning_rate": 2.4761747369580786e-05,
      "loss": 1.3339,
      "step": 33540
    },
    {
      "epoch": 5.544309027060525,
      "grad_norm": 9.125925064086914,
      "learning_rate": 2.4752566150682165e-05,
      "loss": 1.4147,
      "step": 33550
    },
    {
      "epoch": 5.545961578186326,
      "grad_norm": 8.341336250305176,
      "learning_rate": 2.4743384931783544e-05,
      "loss": 1.4189,
      "step": 33560
    },
    {
      "epoch": 5.547614129312126,
      "grad_norm": 10.543508529663086,
      "learning_rate": 2.4734203712884923e-05,
      "loss": 1.4079,
      "step": 33570
    },
    {
      "epoch": 5.549266680437926,
      "grad_norm": 16.04317283630371,
      "learning_rate": 2.4725022493986303e-05,
      "loss": 1.6624,
      "step": 33580
    },
    {
      "epoch": 5.550919231563727,
      "grad_norm": 10.512799263000488,
      "learning_rate": 2.4715841275087682e-05,
      "loss": 1.4323,
      "step": 33590
    },
    {
      "epoch": 5.552571782689527,
      "grad_norm": 8.222831726074219,
      "learning_rate": 2.470666005618906e-05,
      "loss": 1.604,
      "step": 33600
    },
    {
      "epoch": 5.554224333815327,
      "grad_norm": 8.435620307922363,
      "learning_rate": 2.4697478837290437e-05,
      "loss": 1.5879,
      "step": 33610
    },
    {
      "epoch": 5.555876884941128,
      "grad_norm": 16.17806625366211,
      "learning_rate": 2.468829761839182e-05,
      "loss": 1.4183,
      "step": 33620
    },
    {
      "epoch": 5.557529436066928,
      "grad_norm": 10.076927185058594,
      "learning_rate": 2.46791163994932e-05,
      "loss": 1.332,
      "step": 33630
    },
    {
      "epoch": 5.559181987192729,
      "grad_norm": 8.407470703125,
      "learning_rate": 2.4669935180594578e-05,
      "loss": 1.5482,
      "step": 33640
    },
    {
      "epoch": 5.560834538318529,
      "grad_norm": 5.60792350769043,
      "learning_rate": 2.4660753961695957e-05,
      "loss": 1.4146,
      "step": 33650
    },
    {
      "epoch": 5.56248708944433,
      "grad_norm": 9.023984909057617,
      "learning_rate": 2.4651572742797337e-05,
      "loss": 1.4729,
      "step": 33660
    },
    {
      "epoch": 5.56413964057013,
      "grad_norm": 9.371933937072754,
      "learning_rate": 2.4642391523898713e-05,
      "loss": 1.4933,
      "step": 33670
    },
    {
      "epoch": 5.56579219169593,
      "grad_norm": 11.463881492614746,
      "learning_rate": 2.4633210305000092e-05,
      "loss": 1.3909,
      "step": 33680
    },
    {
      "epoch": 5.567444742821731,
      "grad_norm": 22.128860473632812,
      "learning_rate": 2.462402908610147e-05,
      "loss": 1.5586,
      "step": 33690
    },
    {
      "epoch": 5.569097293947531,
      "grad_norm": 7.879482746124268,
      "learning_rate": 2.461484786720285e-05,
      "loss": 1.4793,
      "step": 33700
    },
    {
      "epoch": 5.5707498450733315,
      "grad_norm": 16.057891845703125,
      "learning_rate": 2.460566664830423e-05,
      "loss": 1.4825,
      "step": 33710
    },
    {
      "epoch": 5.5724023961991325,
      "grad_norm": 21.956098556518555,
      "learning_rate": 2.459648542940561e-05,
      "loss": 1.4789,
      "step": 33720
    },
    {
      "epoch": 5.574054947324933,
      "grad_norm": 10.628647804260254,
      "learning_rate": 2.4587304210506988e-05,
      "loss": 1.4485,
      "step": 33730
    },
    {
      "epoch": 5.5757074984507335,
      "grad_norm": 22.851346969604492,
      "learning_rate": 2.4578122991608367e-05,
      "loss": 1.5087,
      "step": 33740
    },
    {
      "epoch": 5.577360049576534,
      "grad_norm": 7.7167744636535645,
      "learning_rate": 2.4568941772709747e-05,
      "loss": 1.4648,
      "step": 33750
    },
    {
      "epoch": 5.579012600702335,
      "grad_norm": 32.65045166015625,
      "learning_rate": 2.4559760553811126e-05,
      "loss": 1.4757,
      "step": 33760
    },
    {
      "epoch": 5.580665151828135,
      "grad_norm": 9.209506034851074,
      "learning_rate": 2.4550579334912505e-05,
      "loss": 1.4859,
      "step": 33770
    },
    {
      "epoch": 5.582317702953935,
      "grad_norm": 10.900243759155273,
      "learning_rate": 2.4541398116013884e-05,
      "loss": 1.5101,
      "step": 33780
    },
    {
      "epoch": 5.583970254079736,
      "grad_norm": 13.830432891845703,
      "learning_rate": 2.4532216897115264e-05,
      "loss": 1.5847,
      "step": 33790
    },
    {
      "epoch": 5.585622805205536,
      "grad_norm": 13.913232803344727,
      "learning_rate": 2.452303567821664e-05,
      "loss": 1.419,
      "step": 33800
    },
    {
      "epoch": 5.587275356331337,
      "grad_norm": 19.409276962280273,
      "learning_rate": 2.451385445931802e-05,
      "loss": 1.5777,
      "step": 33810
    },
    {
      "epoch": 5.588927907457137,
      "grad_norm": 14.467093467712402,
      "learning_rate": 2.4504673240419398e-05,
      "loss": 1.635,
      "step": 33820
    },
    {
      "epoch": 5.590580458582937,
      "grad_norm": 9.140510559082031,
      "learning_rate": 2.4495492021520777e-05,
      "loss": 1.5292,
      "step": 33830
    },
    {
      "epoch": 5.592233009708738,
      "grad_norm": 7.822348594665527,
      "learning_rate": 2.4486310802622156e-05,
      "loss": 1.3846,
      "step": 33840
    },
    {
      "epoch": 5.593885560834538,
      "grad_norm": 9.017069816589355,
      "learning_rate": 2.447712958372354e-05,
      "loss": 1.4916,
      "step": 33850
    },
    {
      "epoch": 5.595538111960339,
      "grad_norm": 11.385196685791016,
      "learning_rate": 2.4467948364824915e-05,
      "loss": 1.508,
      "step": 33860
    },
    {
      "epoch": 5.597190663086139,
      "grad_norm": 7.629830837249756,
      "learning_rate": 2.4458767145926294e-05,
      "loss": 1.4346,
      "step": 33870
    },
    {
      "epoch": 5.598843214211939,
      "grad_norm": 7.943394184112549,
      "learning_rate": 2.4449585927027673e-05,
      "loss": 1.4446,
      "step": 33880
    },
    {
      "epoch": 5.60049576533774,
      "grad_norm": 12.818558692932129,
      "learning_rate": 2.4440404708129053e-05,
      "loss": 1.4871,
      "step": 33890
    },
    {
      "epoch": 5.60214831646354,
      "grad_norm": 8.883275985717773,
      "learning_rate": 2.4431223489230432e-05,
      "loss": 1.4815,
      "step": 33900
    },
    {
      "epoch": 5.603800867589341,
      "grad_norm": 12.792842864990234,
      "learning_rate": 2.442204227033181e-05,
      "loss": 1.4886,
      "step": 33910
    },
    {
      "epoch": 5.6054534187151415,
      "grad_norm": 9.756999015808105,
      "learning_rate": 2.441286105143319e-05,
      "loss": 1.4582,
      "step": 33920
    },
    {
      "epoch": 5.6071059698409424,
      "grad_norm": 11.55717658996582,
      "learning_rate": 2.4403679832534566e-05,
      "loss": 1.6197,
      "step": 33930
    },
    {
      "epoch": 5.6087585209667425,
      "grad_norm": 6.848897933959961,
      "learning_rate": 2.4394498613635945e-05,
      "loss": 1.4918,
      "step": 33940
    },
    {
      "epoch": 5.610411072092543,
      "grad_norm": 7.045119762420654,
      "learning_rate": 2.4385317394737325e-05,
      "loss": 1.6231,
      "step": 33950
    },
    {
      "epoch": 5.612063623218344,
      "grad_norm": 12.93820571899414,
      "learning_rate": 2.4376136175838707e-05,
      "loss": 1.4263,
      "step": 33960
    },
    {
      "epoch": 5.613716174344144,
      "grad_norm": 11.486026763916016,
      "learning_rate": 2.4366954956940087e-05,
      "loss": 1.4914,
      "step": 33970
    },
    {
      "epoch": 5.615368725469944,
      "grad_norm": 9.217084884643555,
      "learning_rate": 2.4357773738041466e-05,
      "loss": 1.4765,
      "step": 33980
    },
    {
      "epoch": 5.617021276595745,
      "grad_norm": 20.32526206970215,
      "learning_rate": 2.4348592519142842e-05,
      "loss": 1.493,
      "step": 33990
    },
    {
      "epoch": 5.618673827721545,
      "grad_norm": 12.086780548095703,
      "learning_rate": 2.433941130024422e-05,
      "loss": 1.5825,
      "step": 34000
    },
    {
      "epoch": 5.620326378847346,
      "grad_norm": 11.141498565673828,
      "learning_rate": 2.43302300813456e-05,
      "loss": 1.4825,
      "step": 34010
    },
    {
      "epoch": 5.621978929973146,
      "grad_norm": 6.86226224899292,
      "learning_rate": 2.432104886244698e-05,
      "loss": 1.5988,
      "step": 34020
    },
    {
      "epoch": 5.623631481098947,
      "grad_norm": 7.528665542602539,
      "learning_rate": 2.431186764354836e-05,
      "loss": 1.5774,
      "step": 34030
    },
    {
      "epoch": 5.625284032224747,
      "grad_norm": 11.368343353271484,
      "learning_rate": 2.4302686424649738e-05,
      "loss": 1.3792,
      "step": 34040
    },
    {
      "epoch": 5.626936583350547,
      "grad_norm": 5.986963272094727,
      "learning_rate": 2.4293505205751117e-05,
      "loss": 1.4285,
      "step": 34050
    },
    {
      "epoch": 5.628589134476348,
      "grad_norm": 9.011881828308105,
      "learning_rate": 2.4284323986852493e-05,
      "loss": 1.476,
      "step": 34060
    },
    {
      "epoch": 5.630241685602148,
      "grad_norm": 8.9256010055542,
      "learning_rate": 2.4275142767953872e-05,
      "loss": 1.4706,
      "step": 34070
    },
    {
      "epoch": 5.631894236727948,
      "grad_norm": 34.386566162109375,
      "learning_rate": 2.4265961549055255e-05,
      "loss": 1.3625,
      "step": 34080
    },
    {
      "epoch": 5.633546787853749,
      "grad_norm": 15.83523178100586,
      "learning_rate": 2.4256780330156634e-05,
      "loss": 1.3465,
      "step": 34090
    },
    {
      "epoch": 5.635199338979549,
      "grad_norm": 9.349414825439453,
      "learning_rate": 2.4247599111258013e-05,
      "loss": 1.4396,
      "step": 34100
    },
    {
      "epoch": 5.63685189010535,
      "grad_norm": 13.072619438171387,
      "learning_rate": 2.4238417892359393e-05,
      "loss": 1.5748,
      "step": 34110
    },
    {
      "epoch": 5.6385044412311505,
      "grad_norm": 8.265012741088867,
      "learning_rate": 2.422923667346077e-05,
      "loss": 1.4326,
      "step": 34120
    },
    {
      "epoch": 5.6401569923569514,
      "grad_norm": 14.93253231048584,
      "learning_rate": 2.4220055454562148e-05,
      "loss": 1.4903,
      "step": 34130
    },
    {
      "epoch": 5.6418095434827515,
      "grad_norm": 8.096601486206055,
      "learning_rate": 2.4210874235663527e-05,
      "loss": 1.384,
      "step": 34140
    },
    {
      "epoch": 5.643462094608552,
      "grad_norm": 7.2971343994140625,
      "learning_rate": 2.4201693016764906e-05,
      "loss": 1.4753,
      "step": 34150
    },
    {
      "epoch": 5.645114645734353,
      "grad_norm": 7.540863990783691,
      "learning_rate": 2.4192511797866286e-05,
      "loss": 1.3667,
      "step": 34160
    },
    {
      "epoch": 5.646767196860153,
      "grad_norm": 5.347985744476318,
      "learning_rate": 2.4183330578967665e-05,
      "loss": 1.5807,
      "step": 34170
    },
    {
      "epoch": 5.648419747985954,
      "grad_norm": 10.622434616088867,
      "learning_rate": 2.4174149360069044e-05,
      "loss": 1.5232,
      "step": 34180
    },
    {
      "epoch": 5.650072299111754,
      "grad_norm": 17.396007537841797,
      "learning_rate": 2.4164968141170423e-05,
      "loss": 1.4468,
      "step": 34190
    },
    {
      "epoch": 5.651724850237554,
      "grad_norm": 17.4361629486084,
      "learning_rate": 2.4155786922271803e-05,
      "loss": 1.5216,
      "step": 34200
    },
    {
      "epoch": 5.653377401363355,
      "grad_norm": 12.362852096557617,
      "learning_rate": 2.4146605703373182e-05,
      "loss": 1.5796,
      "step": 34210
    },
    {
      "epoch": 5.655029952489155,
      "grad_norm": 8.334657669067383,
      "learning_rate": 2.413742448447456e-05,
      "loss": 1.5931,
      "step": 34220
    },
    {
      "epoch": 5.656682503614956,
      "grad_norm": 15.316583633422852,
      "learning_rate": 2.412824326557594e-05,
      "loss": 1.5869,
      "step": 34230
    },
    {
      "epoch": 5.658335054740756,
      "grad_norm": 17.858875274658203,
      "learning_rate": 2.411906204667732e-05,
      "loss": 1.5354,
      "step": 34240
    },
    {
      "epoch": 5.659987605866556,
      "grad_norm": 15.654756546020508,
      "learning_rate": 2.4109880827778695e-05,
      "loss": 1.5952,
      "step": 34250
    },
    {
      "epoch": 5.661640156992357,
      "grad_norm": 5.612504005432129,
      "learning_rate": 2.4100699608880075e-05,
      "loss": 1.3402,
      "step": 34260
    },
    {
      "epoch": 5.663292708118157,
      "grad_norm": 12.783700942993164,
      "learning_rate": 2.4091518389981454e-05,
      "loss": 1.4176,
      "step": 34270
    },
    {
      "epoch": 5.664945259243958,
      "grad_norm": 7.64247989654541,
      "learning_rate": 2.4082337171082833e-05,
      "loss": 1.4333,
      "step": 34280
    },
    {
      "epoch": 5.666597810369758,
      "grad_norm": 7.911691665649414,
      "learning_rate": 2.4073155952184212e-05,
      "loss": 1.3645,
      "step": 34290
    },
    {
      "epoch": 5.668250361495558,
      "grad_norm": 8.747984886169434,
      "learning_rate": 2.406397473328559e-05,
      "loss": 1.5304,
      "step": 34300
    },
    {
      "epoch": 5.669902912621359,
      "grad_norm": 12.237468719482422,
      "learning_rate": 2.405479351438697e-05,
      "loss": 1.3955,
      "step": 34310
    },
    {
      "epoch": 5.6715554637471595,
      "grad_norm": 9.102463722229004,
      "learning_rate": 2.404561229548835e-05,
      "loss": 1.4228,
      "step": 34320
    },
    {
      "epoch": 5.6732080148729604,
      "grad_norm": 11.882768630981445,
      "learning_rate": 2.403643107658973e-05,
      "loss": 1.3596,
      "step": 34330
    },
    {
      "epoch": 5.6748605659987605,
      "grad_norm": 10.731307029724121,
      "learning_rate": 2.402724985769111e-05,
      "loss": 1.6394,
      "step": 34340
    },
    {
      "epoch": 5.676513117124561,
      "grad_norm": 13.682147026062012,
      "learning_rate": 2.4018068638792488e-05,
      "loss": 1.4626,
      "step": 34350
    },
    {
      "epoch": 5.678165668250362,
      "grad_norm": 14.700006484985352,
      "learning_rate": 2.4008887419893867e-05,
      "loss": 1.5381,
      "step": 34360
    },
    {
      "epoch": 5.679818219376162,
      "grad_norm": 8.677534103393555,
      "learning_rate": 2.3999706200995246e-05,
      "loss": 1.3861,
      "step": 34370
    },
    {
      "epoch": 5.681470770501963,
      "grad_norm": 9.908595085144043,
      "learning_rate": 2.3990524982096622e-05,
      "loss": 1.3655,
      "step": 34380
    },
    {
      "epoch": 5.683123321627763,
      "grad_norm": 22.74381446838379,
      "learning_rate": 2.3981343763198e-05,
      "loss": 1.479,
      "step": 34390
    },
    {
      "epoch": 5.684775872753564,
      "grad_norm": 13.582764625549316,
      "learning_rate": 2.397216254429938e-05,
      "loss": 1.3896,
      "step": 34400
    },
    {
      "epoch": 5.686428423879364,
      "grad_norm": 21.719499588012695,
      "learning_rate": 2.396298132540076e-05,
      "loss": 1.6054,
      "step": 34410
    },
    {
      "epoch": 5.688080975005164,
      "grad_norm": 12.603348731994629,
      "learning_rate": 2.3953800106502143e-05,
      "loss": 1.4277,
      "step": 34420
    },
    {
      "epoch": 5.689733526130965,
      "grad_norm": 9.63592529296875,
      "learning_rate": 2.3944618887603522e-05,
      "loss": 1.485,
      "step": 34430
    },
    {
      "epoch": 5.691386077256765,
      "grad_norm": 13.966578483581543,
      "learning_rate": 2.3935437668704898e-05,
      "loss": 1.5003,
      "step": 34440
    },
    {
      "epoch": 5.693038628382565,
      "grad_norm": 13.241275787353516,
      "learning_rate": 2.3926256449806277e-05,
      "loss": 1.5059,
      "step": 34450
    },
    {
      "epoch": 5.694691179508366,
      "grad_norm": 18.274063110351562,
      "learning_rate": 2.3917075230907656e-05,
      "loss": 1.4191,
      "step": 34460
    },
    {
      "epoch": 5.696343730634166,
      "grad_norm": 6.728181838989258,
      "learning_rate": 2.3907894012009035e-05,
      "loss": 1.5287,
      "step": 34470
    },
    {
      "epoch": 5.697996281759967,
      "grad_norm": 9.49169921875,
      "learning_rate": 2.3898712793110415e-05,
      "loss": 1.5762,
      "step": 34480
    },
    {
      "epoch": 5.699648832885767,
      "grad_norm": 13.1586332321167,
      "learning_rate": 2.3889531574211794e-05,
      "loss": 1.4773,
      "step": 34490
    },
    {
      "epoch": 5.701301384011568,
      "grad_norm": 36.238487243652344,
      "learning_rate": 2.3880350355313173e-05,
      "loss": 1.4757,
      "step": 34500
    },
    {
      "epoch": 5.702953935137368,
      "grad_norm": 14.713821411132812,
      "learning_rate": 2.387116913641455e-05,
      "loss": 1.3241,
      "step": 34510
    },
    {
      "epoch": 5.7046064862631685,
      "grad_norm": 8.172168731689453,
      "learning_rate": 2.3861987917515928e-05,
      "loss": 1.4352,
      "step": 34520
    },
    {
      "epoch": 5.7062590373889694,
      "grad_norm": 12.12569808959961,
      "learning_rate": 2.3852806698617308e-05,
      "loss": 1.6369,
      "step": 34530
    },
    {
      "epoch": 5.7079115885147695,
      "grad_norm": 8.741676330566406,
      "learning_rate": 2.384362547971869e-05,
      "loss": 1.4149,
      "step": 34540
    },
    {
      "epoch": 5.70956413964057,
      "grad_norm": 15.845980644226074,
      "learning_rate": 2.383444426082007e-05,
      "loss": 1.4781,
      "step": 34550
    },
    {
      "epoch": 5.711216690766371,
      "grad_norm": 10.60511302947998,
      "learning_rate": 2.382526304192145e-05,
      "loss": 1.4277,
      "step": 34560
    },
    {
      "epoch": 5.712869241892171,
      "grad_norm": 12.912707328796387,
      "learning_rate": 2.3816081823022825e-05,
      "loss": 1.4863,
      "step": 34570
    },
    {
      "epoch": 5.714521793017972,
      "grad_norm": 7.532742977142334,
      "learning_rate": 2.3806900604124204e-05,
      "loss": 1.3594,
      "step": 34580
    },
    {
      "epoch": 5.716174344143772,
      "grad_norm": 109.51636505126953,
      "learning_rate": 2.3797719385225583e-05,
      "loss": 1.4491,
      "step": 34590
    },
    {
      "epoch": 5.717826895269573,
      "grad_norm": 25.65382957458496,
      "learning_rate": 2.3788538166326962e-05,
      "loss": 1.5347,
      "step": 34600
    },
    {
      "epoch": 5.719479446395373,
      "grad_norm": 6.745087146759033,
      "learning_rate": 2.377935694742834e-05,
      "loss": 1.5059,
      "step": 34610
    },
    {
      "epoch": 5.721131997521173,
      "grad_norm": 8.282033920288086,
      "learning_rate": 2.377017572852972e-05,
      "loss": 1.4472,
      "step": 34620
    },
    {
      "epoch": 5.722784548646974,
      "grad_norm": 12.051273345947266,
      "learning_rate": 2.37609945096311e-05,
      "loss": 1.4371,
      "step": 34630
    },
    {
      "epoch": 5.724437099772774,
      "grad_norm": 7.050853252410889,
      "learning_rate": 2.3751813290732476e-05,
      "loss": 1.5519,
      "step": 34640
    },
    {
      "epoch": 5.726089650898575,
      "grad_norm": 29.751380920410156,
      "learning_rate": 2.374263207183386e-05,
      "loss": 1.38,
      "step": 34650
    },
    {
      "epoch": 5.727742202024375,
      "grad_norm": 17.988632202148438,
      "learning_rate": 2.3733450852935238e-05,
      "loss": 1.4125,
      "step": 34660
    },
    {
      "epoch": 5.729394753150175,
      "grad_norm": 19.773149490356445,
      "learning_rate": 2.3724269634036617e-05,
      "loss": 1.3835,
      "step": 34670
    },
    {
      "epoch": 5.731047304275976,
      "grad_norm": 23.994436264038086,
      "learning_rate": 2.3715088415137996e-05,
      "loss": 1.4931,
      "step": 34680
    },
    {
      "epoch": 5.732699855401776,
      "grad_norm": 24.97893714904785,
      "learning_rate": 2.3705907196239375e-05,
      "loss": 1.3949,
      "step": 34690
    },
    {
      "epoch": 5.734352406527577,
      "grad_norm": 6.368709564208984,
      "learning_rate": 2.369672597734075e-05,
      "loss": 1.3729,
      "step": 34700
    },
    {
      "epoch": 5.736004957653377,
      "grad_norm": 14.183878898620605,
      "learning_rate": 2.368754475844213e-05,
      "loss": 1.5165,
      "step": 34710
    },
    {
      "epoch": 5.7376575087791775,
      "grad_norm": 11.956380844116211,
      "learning_rate": 2.367836353954351e-05,
      "loss": 1.4883,
      "step": 34720
    },
    {
      "epoch": 5.7393100599049784,
      "grad_norm": 16.989643096923828,
      "learning_rate": 2.366918232064489e-05,
      "loss": 1.5531,
      "step": 34730
    },
    {
      "epoch": 5.7409626110307785,
      "grad_norm": 14.136730194091797,
      "learning_rate": 2.366000110174627e-05,
      "loss": 1.4212,
      "step": 34740
    },
    {
      "epoch": 5.7426151621565795,
      "grad_norm": 10.436883926391602,
      "learning_rate": 2.3650819882847648e-05,
      "loss": 1.5093,
      "step": 34750
    },
    {
      "epoch": 5.74426771328238,
      "grad_norm": 23.723215103149414,
      "learning_rate": 2.3641638663949027e-05,
      "loss": 1.5963,
      "step": 34760
    },
    {
      "epoch": 5.745920264408181,
      "grad_norm": 29.90015983581543,
      "learning_rate": 2.3632457445050406e-05,
      "loss": 1.5902,
      "step": 34770
    },
    {
      "epoch": 5.747572815533981,
      "grad_norm": 20.19734764099121,
      "learning_rate": 2.3623276226151785e-05,
      "loss": 1.3964,
      "step": 34780
    },
    {
      "epoch": 5.749225366659781,
      "grad_norm": 7.617761611938477,
      "learning_rate": 2.3614095007253165e-05,
      "loss": 1.4068,
      "step": 34790
    },
    {
      "epoch": 5.750877917785582,
      "grad_norm": 12.926080703735352,
      "learning_rate": 2.3604913788354544e-05,
      "loss": 1.4668,
      "step": 34800
    },
    {
      "epoch": 5.752530468911382,
      "grad_norm": 10.092638969421387,
      "learning_rate": 2.3595732569455923e-05,
      "loss": 1.526,
      "step": 34810
    },
    {
      "epoch": 5.754183020037182,
      "grad_norm": 16.717660903930664,
      "learning_rate": 2.3586551350557302e-05,
      "loss": 1.5049,
      "step": 34820
    },
    {
      "epoch": 5.755835571162983,
      "grad_norm": 9.57276439666748,
      "learning_rate": 2.3577370131658678e-05,
      "loss": 1.4629,
      "step": 34830
    },
    {
      "epoch": 5.757488122288783,
      "grad_norm": 8.01769733428955,
      "learning_rate": 2.3568188912760057e-05,
      "loss": 1.4832,
      "step": 34840
    },
    {
      "epoch": 5.759140673414584,
      "grad_norm": 11.659310340881348,
      "learning_rate": 2.3559007693861437e-05,
      "loss": 1.4063,
      "step": 34850
    },
    {
      "epoch": 5.760793224540384,
      "grad_norm": 11.5115327835083,
      "learning_rate": 2.3549826474962816e-05,
      "loss": 1.4312,
      "step": 34860
    },
    {
      "epoch": 5.762445775666185,
      "grad_norm": 8.317569732666016,
      "learning_rate": 2.3540645256064195e-05,
      "loss": 1.4929,
      "step": 34870
    },
    {
      "epoch": 5.764098326791985,
      "grad_norm": 12.994110107421875,
      "learning_rate": 2.3531464037165578e-05,
      "loss": 1.4944,
      "step": 34880
    },
    {
      "epoch": 5.765750877917785,
      "grad_norm": 8.37218189239502,
      "learning_rate": 2.3522282818266954e-05,
      "loss": 1.4133,
      "step": 34890
    },
    {
      "epoch": 5.767403429043586,
      "grad_norm": 9.333100318908691,
      "learning_rate": 2.3513101599368333e-05,
      "loss": 1.4405,
      "step": 34900
    },
    {
      "epoch": 5.769055980169386,
      "grad_norm": 17.137237548828125,
      "learning_rate": 2.3503920380469712e-05,
      "loss": 1.513,
      "step": 34910
    },
    {
      "epoch": 5.7707085312951865,
      "grad_norm": 12.368675231933594,
      "learning_rate": 2.349473916157109e-05,
      "loss": 1.3646,
      "step": 34920
    },
    {
      "epoch": 5.772361082420987,
      "grad_norm": 14.37025260925293,
      "learning_rate": 2.348555794267247e-05,
      "loss": 1.5304,
      "step": 34930
    },
    {
      "epoch": 5.7740136335467875,
      "grad_norm": 40.79022216796875,
      "learning_rate": 2.347637672377385e-05,
      "loss": 1.3919,
      "step": 34940
    },
    {
      "epoch": 5.7756661846725885,
      "grad_norm": 8.884847640991211,
      "learning_rate": 2.346719550487523e-05,
      "loss": 1.4815,
      "step": 34950
    },
    {
      "epoch": 5.777318735798389,
      "grad_norm": 9.6146879196167,
      "learning_rate": 2.3458014285976605e-05,
      "loss": 1.4746,
      "step": 34960
    },
    {
      "epoch": 5.77897128692419,
      "grad_norm": 18.530406951904297,
      "learning_rate": 2.3448833067077984e-05,
      "loss": 1.644,
      "step": 34970
    },
    {
      "epoch": 5.78062383804999,
      "grad_norm": 26.01064109802246,
      "learning_rate": 2.3439651848179364e-05,
      "loss": 1.5561,
      "step": 34980
    },
    {
      "epoch": 5.78227638917579,
      "grad_norm": 6.7076239585876465,
      "learning_rate": 2.3430470629280746e-05,
      "loss": 1.441,
      "step": 34990
    },
    {
      "epoch": 5.783928940301591,
      "grad_norm": 58.96223831176758,
      "learning_rate": 2.3421289410382125e-05,
      "loss": 1.4621,
      "step": 35000
    },
    {
      "epoch": 5.785581491427391,
      "grad_norm": 12.242283821105957,
      "learning_rate": 2.3412108191483505e-05,
      "loss": 1.4991,
      "step": 35010
    },
    {
      "epoch": 5.787234042553192,
      "grad_norm": 60.0484619140625,
      "learning_rate": 2.340292697258488e-05,
      "loss": 1.4799,
      "step": 35020
    },
    {
      "epoch": 5.788886593678992,
      "grad_norm": 15.15187931060791,
      "learning_rate": 2.339374575368626e-05,
      "loss": 1.5587,
      "step": 35030
    },
    {
      "epoch": 5.790539144804792,
      "grad_norm": 14.363666534423828,
      "learning_rate": 2.338456453478764e-05,
      "loss": 1.5531,
      "step": 35040
    },
    {
      "epoch": 5.792191695930593,
      "grad_norm": 9.732187271118164,
      "learning_rate": 2.3375383315889018e-05,
      "loss": 1.5367,
      "step": 35050
    },
    {
      "epoch": 5.793844247056393,
      "grad_norm": 8.981781005859375,
      "learning_rate": 2.3366202096990397e-05,
      "loss": 1.4947,
      "step": 35060
    },
    {
      "epoch": 5.795496798182194,
      "grad_norm": 10.503220558166504,
      "learning_rate": 2.3357020878091777e-05,
      "loss": 1.3545,
      "step": 35070
    },
    {
      "epoch": 5.797149349307994,
      "grad_norm": 8.12704849243164,
      "learning_rate": 2.3347839659193156e-05,
      "loss": 1.5702,
      "step": 35080
    },
    {
      "epoch": 5.798801900433794,
      "grad_norm": 13.842459678649902,
      "learning_rate": 2.3338658440294532e-05,
      "loss": 1.4373,
      "step": 35090
    },
    {
      "epoch": 5.800454451559595,
      "grad_norm": 11.6895170211792,
      "learning_rate": 2.332947722139591e-05,
      "loss": 1.3984,
      "step": 35100
    },
    {
      "epoch": 5.802107002685395,
      "grad_norm": 55.09220886230469,
      "learning_rate": 2.3320296002497294e-05,
      "loss": 1.5888,
      "step": 35110
    },
    {
      "epoch": 5.803759553811196,
      "grad_norm": 10.366085052490234,
      "learning_rate": 2.3311114783598673e-05,
      "loss": 1.4724,
      "step": 35120
    },
    {
      "epoch": 5.805412104936996,
      "grad_norm": 12.669951438903809,
      "learning_rate": 2.3301933564700052e-05,
      "loss": 1.595,
      "step": 35130
    },
    {
      "epoch": 5.807064656062797,
      "grad_norm": 9.713029861450195,
      "learning_rate": 2.329275234580143e-05,
      "loss": 1.3403,
      "step": 35140
    },
    {
      "epoch": 5.8087172071885975,
      "grad_norm": 10.147992134094238,
      "learning_rate": 2.3283571126902807e-05,
      "loss": 1.4291,
      "step": 35150
    },
    {
      "epoch": 5.810369758314398,
      "grad_norm": 8.817466735839844,
      "learning_rate": 2.3274389908004187e-05,
      "loss": 1.4716,
      "step": 35160
    },
    {
      "epoch": 5.812022309440199,
      "grad_norm": 17.47516441345215,
      "learning_rate": 2.3265208689105566e-05,
      "loss": 1.4725,
      "step": 35170
    },
    {
      "epoch": 5.813674860565999,
      "grad_norm": 33.834625244140625,
      "learning_rate": 2.3256027470206945e-05,
      "loss": 1.5657,
      "step": 35180
    },
    {
      "epoch": 5.815327411691799,
      "grad_norm": 14.494001388549805,
      "learning_rate": 2.3246846251308324e-05,
      "loss": 1.486,
      "step": 35190
    },
    {
      "epoch": 5.8169799628176,
      "grad_norm": 10.042065620422363,
      "learning_rate": 2.3237665032409704e-05,
      "loss": 1.4782,
      "step": 35200
    },
    {
      "epoch": 5.8186325139434,
      "grad_norm": 8.86281681060791,
      "learning_rate": 2.3228483813511083e-05,
      "loss": 1.5148,
      "step": 35210
    },
    {
      "epoch": 5.820285065069201,
      "grad_norm": 17.04343605041504,
      "learning_rate": 2.3219302594612462e-05,
      "loss": 1.3849,
      "step": 35220
    },
    {
      "epoch": 5.821937616195001,
      "grad_norm": 11.681707382202148,
      "learning_rate": 2.321012137571384e-05,
      "loss": 1.4524,
      "step": 35230
    },
    {
      "epoch": 5.823590167320802,
      "grad_norm": 15.696507453918457,
      "learning_rate": 2.320094015681522e-05,
      "loss": 1.4256,
      "step": 35240
    },
    {
      "epoch": 5.825242718446602,
      "grad_norm": 21.752723693847656,
      "learning_rate": 2.31917589379166e-05,
      "loss": 1.376,
      "step": 35250
    },
    {
      "epoch": 5.826895269572402,
      "grad_norm": 16.060373306274414,
      "learning_rate": 2.318257771901798e-05,
      "loss": 1.514,
      "step": 35260
    },
    {
      "epoch": 5.828547820698203,
      "grad_norm": 51.685455322265625,
      "learning_rate": 2.3173396500119358e-05,
      "loss": 1.433,
      "step": 35270
    },
    {
      "epoch": 5.830200371824003,
      "grad_norm": 8.148418426513672,
      "learning_rate": 2.3164215281220734e-05,
      "loss": 1.4283,
      "step": 35280
    },
    {
      "epoch": 5.831852922949803,
      "grad_norm": 7.561429977416992,
      "learning_rate": 2.3155034062322113e-05,
      "loss": 1.4483,
      "step": 35290
    },
    {
      "epoch": 5.833505474075604,
      "grad_norm": 13.371988296508789,
      "learning_rate": 2.3145852843423493e-05,
      "loss": 1.4895,
      "step": 35300
    },
    {
      "epoch": 5.835158025201404,
      "grad_norm": 15.92422103881836,
      "learning_rate": 2.3136671624524872e-05,
      "loss": 1.3263,
      "step": 35310
    },
    {
      "epoch": 5.836810576327205,
      "grad_norm": 49.8977165222168,
      "learning_rate": 2.312749040562625e-05,
      "loss": 1.5131,
      "step": 35320
    },
    {
      "epoch": 5.838463127453005,
      "grad_norm": 11.140594482421875,
      "learning_rate": 2.311830918672763e-05,
      "loss": 1.5284,
      "step": 35330
    },
    {
      "epoch": 5.840115678578806,
      "grad_norm": 9.05160903930664,
      "learning_rate": 2.310912796782901e-05,
      "loss": 1.515,
      "step": 35340
    },
    {
      "epoch": 5.8417682297046065,
      "grad_norm": 11.951143264770508,
      "learning_rate": 2.309994674893039e-05,
      "loss": 1.5105,
      "step": 35350
    },
    {
      "epoch": 5.843420780830407,
      "grad_norm": 10.511120796203613,
      "learning_rate": 2.3090765530031768e-05,
      "loss": 1.4417,
      "step": 35360
    },
    {
      "epoch": 5.845073331956208,
      "grad_norm": 18.432544708251953,
      "learning_rate": 2.3081584311133147e-05,
      "loss": 1.5113,
      "step": 35370
    },
    {
      "epoch": 5.846725883082008,
      "grad_norm": 12.873449325561523,
      "learning_rate": 2.3072403092234527e-05,
      "loss": 1.4663,
      "step": 35380
    },
    {
      "epoch": 5.848378434207809,
      "grad_norm": 7.534992218017578,
      "learning_rate": 2.3063221873335906e-05,
      "loss": 1.4698,
      "step": 35390
    },
    {
      "epoch": 5.850030985333609,
      "grad_norm": 6.996005535125732,
      "learning_rate": 2.3054040654437285e-05,
      "loss": 1.3313,
      "step": 35400
    },
    {
      "epoch": 5.851683536459409,
      "grad_norm": 122.31879425048828,
      "learning_rate": 2.304485943553866e-05,
      "loss": 1.3871,
      "step": 35410
    },
    {
      "epoch": 5.85333608758521,
      "grad_norm": 10.604686737060547,
      "learning_rate": 2.303567821664004e-05,
      "loss": 1.5325,
      "step": 35420
    },
    {
      "epoch": 5.85498863871101,
      "grad_norm": 16.7020263671875,
      "learning_rate": 2.302649699774142e-05,
      "loss": 1.5011,
      "step": 35430
    },
    {
      "epoch": 5.856641189836811,
      "grad_norm": 17.591352462768555,
      "learning_rate": 2.30173157788428e-05,
      "loss": 1.4792,
      "step": 35440
    },
    {
      "epoch": 5.858293740962611,
      "grad_norm": 10.369803428649902,
      "learning_rate": 2.300813455994418e-05,
      "loss": 1.4176,
      "step": 35450
    },
    {
      "epoch": 5.859946292088411,
      "grad_norm": 41.722450256347656,
      "learning_rate": 2.299895334104556e-05,
      "loss": 1.3517,
      "step": 35460
    },
    {
      "epoch": 5.861598843214212,
      "grad_norm": 13.766722679138184,
      "learning_rate": 2.2989772122146936e-05,
      "loss": 1.5139,
      "step": 35470
    },
    {
      "epoch": 5.863251394340012,
      "grad_norm": 11.477117538452148,
      "learning_rate": 2.2980590903248316e-05,
      "loss": 1.3467,
      "step": 35480
    },
    {
      "epoch": 5.864903945465813,
      "grad_norm": 7.684751033782959,
      "learning_rate": 2.2971409684349695e-05,
      "loss": 1.3977,
      "step": 35490
    },
    {
      "epoch": 5.866556496591613,
      "grad_norm": 19.736038208007812,
      "learning_rate": 2.2962228465451074e-05,
      "loss": 1.4578,
      "step": 35500
    },
    {
      "epoch": 5.868209047717414,
      "grad_norm": 9.840457916259766,
      "learning_rate": 2.2953047246552453e-05,
      "loss": 1.5332,
      "step": 35510
    },
    {
      "epoch": 5.869861598843214,
      "grad_norm": 10.037358283996582,
      "learning_rate": 2.2943866027653833e-05,
      "loss": 1.4446,
      "step": 35520
    },
    {
      "epoch": 5.871514149969014,
      "grad_norm": 12.01092529296875,
      "learning_rate": 2.2934684808755212e-05,
      "loss": 1.4489,
      "step": 35530
    },
    {
      "epoch": 5.873166701094815,
      "grad_norm": 11.730144500732422,
      "learning_rate": 2.292550358985659e-05,
      "loss": 1.4665,
      "step": 35540
    },
    {
      "epoch": 5.8748192522206155,
      "grad_norm": 9.427178382873535,
      "learning_rate": 2.2916322370957967e-05,
      "loss": 1.4231,
      "step": 35550
    },
    {
      "epoch": 5.876471803346416,
      "grad_norm": 9.14688491821289,
      "learning_rate": 2.290714115205935e-05,
      "loss": 1.4917,
      "step": 35560
    },
    {
      "epoch": 5.878124354472217,
      "grad_norm": 8.813817977905273,
      "learning_rate": 2.289795993316073e-05,
      "loss": 1.386,
      "step": 35570
    },
    {
      "epoch": 5.879776905598017,
      "grad_norm": 13.4219970703125,
      "learning_rate": 2.2888778714262108e-05,
      "loss": 1.6158,
      "step": 35580
    },
    {
      "epoch": 5.881429456723818,
      "grad_norm": 18.55006217956543,
      "learning_rate": 2.2879597495363487e-05,
      "loss": 1.5049,
      "step": 35590
    },
    {
      "epoch": 5.883082007849618,
      "grad_norm": 10.324186325073242,
      "learning_rate": 2.2870416276464863e-05,
      "loss": 1.4228,
      "step": 35600
    },
    {
      "epoch": 5.884734558975419,
      "grad_norm": 14.822036743164062,
      "learning_rate": 2.2861235057566243e-05,
      "loss": 1.4369,
      "step": 35610
    },
    {
      "epoch": 5.886387110101219,
      "grad_norm": 9.402215003967285,
      "learning_rate": 2.2852053838667622e-05,
      "loss": 1.5191,
      "step": 35620
    },
    {
      "epoch": 5.888039661227019,
      "grad_norm": 15.71996021270752,
      "learning_rate": 2.2842872619769e-05,
      "loss": 1.4482,
      "step": 35630
    },
    {
      "epoch": 5.88969221235282,
      "grad_norm": 6.093387126922607,
      "learning_rate": 2.283369140087038e-05,
      "loss": 1.4423,
      "step": 35640
    },
    {
      "epoch": 5.89134476347862,
      "grad_norm": 9.502908706665039,
      "learning_rate": 2.282451018197176e-05,
      "loss": 1.4896,
      "step": 35650
    },
    {
      "epoch": 5.89299731460442,
      "grad_norm": 10.822142601013184,
      "learning_rate": 2.281532896307314e-05,
      "loss": 1.5373,
      "step": 35660
    },
    {
      "epoch": 5.894649865730221,
      "grad_norm": 12.028356552124023,
      "learning_rate": 2.2806147744174518e-05,
      "loss": 1.6253,
      "step": 35670
    },
    {
      "epoch": 5.896302416856021,
      "grad_norm": 9.975237846374512,
      "learning_rate": 2.2796966525275897e-05,
      "loss": 1.57,
      "step": 35680
    },
    {
      "epoch": 5.897954967981822,
      "grad_norm": 8.391766548156738,
      "learning_rate": 2.2787785306377277e-05,
      "loss": 1.5282,
      "step": 35690
    },
    {
      "epoch": 5.899607519107622,
      "grad_norm": 13.360061645507812,
      "learning_rate": 2.2778604087478656e-05,
      "loss": 1.5431,
      "step": 35700
    },
    {
      "epoch": 5.901260070233423,
      "grad_norm": 6.314427375793457,
      "learning_rate": 2.2769422868580035e-05,
      "loss": 1.5607,
      "step": 35710
    },
    {
      "epoch": 5.902912621359223,
      "grad_norm": 7.808109283447266,
      "learning_rate": 2.2760241649681414e-05,
      "loss": 1.4794,
      "step": 35720
    },
    {
      "epoch": 5.904565172485023,
      "grad_norm": 9.234506607055664,
      "learning_rate": 2.275106043078279e-05,
      "loss": 1.3842,
      "step": 35730
    },
    {
      "epoch": 5.906217723610824,
      "grad_norm": 35.93871307373047,
      "learning_rate": 2.274187921188417e-05,
      "loss": 1.3852,
      "step": 35740
    },
    {
      "epoch": 5.9078702747366245,
      "grad_norm": 10.612404823303223,
      "learning_rate": 2.273269799298555e-05,
      "loss": 1.5685,
      "step": 35750
    },
    {
      "epoch": 5.9095228258624255,
      "grad_norm": 43.04104232788086,
      "learning_rate": 2.2723516774086928e-05,
      "loss": 1.3909,
      "step": 35760
    },
    {
      "epoch": 5.911175376988226,
      "grad_norm": 16.839719772338867,
      "learning_rate": 2.2714335555188307e-05,
      "loss": 1.4255,
      "step": 35770
    },
    {
      "epoch": 5.912827928114026,
      "grad_norm": 16.71356773376465,
      "learning_rate": 2.2705154336289686e-05,
      "loss": 1.6229,
      "step": 35780
    },
    {
      "epoch": 5.914480479239827,
      "grad_norm": 28.990924835205078,
      "learning_rate": 2.2695973117391066e-05,
      "loss": 1.5036,
      "step": 35790
    },
    {
      "epoch": 5.916133030365627,
      "grad_norm": 8.961358070373535,
      "learning_rate": 2.2686791898492445e-05,
      "loss": 1.4317,
      "step": 35800
    },
    {
      "epoch": 5.917785581491428,
      "grad_norm": 9.525544166564941,
      "learning_rate": 2.2677610679593824e-05,
      "loss": 1.5382,
      "step": 35810
    },
    {
      "epoch": 5.919438132617228,
      "grad_norm": 16.726709365844727,
      "learning_rate": 2.2668429460695203e-05,
      "loss": 1.4789,
      "step": 35820
    },
    {
      "epoch": 5.921090683743028,
      "grad_norm": 26.31761360168457,
      "learning_rate": 2.2659248241796583e-05,
      "loss": 1.53,
      "step": 35830
    },
    {
      "epoch": 5.922743234868829,
      "grad_norm": 18.97273826599121,
      "learning_rate": 2.2650067022897962e-05,
      "loss": 1.4395,
      "step": 35840
    },
    {
      "epoch": 5.924395785994629,
      "grad_norm": 6.1228928565979,
      "learning_rate": 2.264088580399934e-05,
      "loss": 1.3988,
      "step": 35850
    },
    {
      "epoch": 5.92604833712043,
      "grad_norm": 16.22532081604004,
      "learning_rate": 2.263170458510072e-05,
      "loss": 1.5295,
      "step": 35860
    },
    {
      "epoch": 5.92770088824623,
      "grad_norm": 7.841466426849365,
      "learning_rate": 2.2622523366202096e-05,
      "loss": 1.2468,
      "step": 35870
    },
    {
      "epoch": 5.92935343937203,
      "grad_norm": 9.497376441955566,
      "learning_rate": 2.2613342147303475e-05,
      "loss": 1.5498,
      "step": 35880
    },
    {
      "epoch": 5.931005990497831,
      "grad_norm": 13.135749816894531,
      "learning_rate": 2.2604160928404855e-05,
      "loss": 1.3572,
      "step": 35890
    },
    {
      "epoch": 5.932658541623631,
      "grad_norm": 12.444751739501953,
      "learning_rate": 2.2594979709506234e-05,
      "loss": 1.4146,
      "step": 35900
    },
    {
      "epoch": 5.934311092749432,
      "grad_norm": 10.811918258666992,
      "learning_rate": 2.2585798490607617e-05,
      "loss": 1.5179,
      "step": 35910
    },
    {
      "epoch": 5.935963643875232,
      "grad_norm": 14.414387702941895,
      "learning_rate": 2.2576617271708992e-05,
      "loss": 1.4109,
      "step": 35920
    },
    {
      "epoch": 5.937616195001032,
      "grad_norm": 10.46023178100586,
      "learning_rate": 2.2567436052810372e-05,
      "loss": 1.4051,
      "step": 35930
    },
    {
      "epoch": 5.939268746126833,
      "grad_norm": 46.42742156982422,
      "learning_rate": 2.255825483391175e-05,
      "loss": 1.4066,
      "step": 35940
    },
    {
      "epoch": 5.9409212972526335,
      "grad_norm": 18.5954532623291,
      "learning_rate": 2.254907361501313e-05,
      "loss": 1.4041,
      "step": 35950
    },
    {
      "epoch": 5.9425738483784345,
      "grad_norm": 8.255085945129395,
      "learning_rate": 2.253989239611451e-05,
      "loss": 1.4775,
      "step": 35960
    },
    {
      "epoch": 5.944226399504235,
      "grad_norm": 16.39137840270996,
      "learning_rate": 2.253071117721589e-05,
      "loss": 1.495,
      "step": 35970
    },
    {
      "epoch": 5.945878950630036,
      "grad_norm": 22.78189468383789,
      "learning_rate": 2.2521529958317268e-05,
      "loss": 1.5671,
      "step": 35980
    },
    {
      "epoch": 5.947531501755836,
      "grad_norm": 10.982939720153809,
      "learning_rate": 2.2512348739418647e-05,
      "loss": 1.5311,
      "step": 35990
    },
    {
      "epoch": 5.949184052881636,
      "grad_norm": 8.388236045837402,
      "learning_rate": 2.2503167520520023e-05,
      "loss": 1.5759,
      "step": 36000
    },
    {
      "epoch": 5.950836604007437,
      "grad_norm": 19.174211502075195,
      "learning_rate": 2.2493986301621402e-05,
      "loss": 1.4469,
      "step": 36010
    },
    {
      "epoch": 5.952489155133237,
      "grad_norm": 9.777421951293945,
      "learning_rate": 2.2484805082722785e-05,
      "loss": 1.4646,
      "step": 36020
    },
    {
      "epoch": 5.954141706259037,
      "grad_norm": 33.97954559326172,
      "learning_rate": 2.2475623863824164e-05,
      "loss": 1.2839,
      "step": 36030
    },
    {
      "epoch": 5.955794257384838,
      "grad_norm": 10.13986587524414,
      "learning_rate": 2.2466442644925543e-05,
      "loss": 1.4924,
      "step": 36040
    },
    {
      "epoch": 5.957446808510638,
      "grad_norm": 11.5507173538208,
      "learning_rate": 2.245726142602692e-05,
      "loss": 1.53,
      "step": 36050
    },
    {
      "epoch": 5.959099359636439,
      "grad_norm": 7.484362602233887,
      "learning_rate": 2.24480802071283e-05,
      "loss": 1.3517,
      "step": 36060
    },
    {
      "epoch": 5.960751910762239,
      "grad_norm": 49.174068450927734,
      "learning_rate": 2.2438898988229678e-05,
      "loss": 1.4585,
      "step": 36070
    },
    {
      "epoch": 5.96240446188804,
      "grad_norm": 9.194089889526367,
      "learning_rate": 2.2429717769331057e-05,
      "loss": 1.5553,
      "step": 36080
    },
    {
      "epoch": 5.96405701301384,
      "grad_norm": 20.3260440826416,
      "learning_rate": 2.2420536550432436e-05,
      "loss": 1.5759,
      "step": 36090
    },
    {
      "epoch": 5.96570956413964,
      "grad_norm": 10.215295791625977,
      "learning_rate": 2.2411355331533816e-05,
      "loss": 1.4843,
      "step": 36100
    },
    {
      "epoch": 5.967362115265441,
      "grad_norm": 12.862039566040039,
      "learning_rate": 2.2402174112635195e-05,
      "loss": 1.5045,
      "step": 36110
    },
    {
      "epoch": 5.969014666391241,
      "grad_norm": 22.09560775756836,
      "learning_rate": 2.2392992893736574e-05,
      "loss": 1.4349,
      "step": 36120
    },
    {
      "epoch": 5.970667217517041,
      "grad_norm": 11.517844200134277,
      "learning_rate": 2.2383811674837953e-05,
      "loss": 1.3689,
      "step": 36130
    },
    {
      "epoch": 5.972319768642842,
      "grad_norm": 14.348694801330566,
      "learning_rate": 2.2374630455939333e-05,
      "loss": 1.4601,
      "step": 36140
    },
    {
      "epoch": 5.9739723197686425,
      "grad_norm": 9.823168754577637,
      "learning_rate": 2.2365449237040712e-05,
      "loss": 1.4595,
      "step": 36150
    },
    {
      "epoch": 5.9756248708944435,
      "grad_norm": 19.39566421508789,
      "learning_rate": 2.235626801814209e-05,
      "loss": 1.4817,
      "step": 36160
    },
    {
      "epoch": 5.977277422020244,
      "grad_norm": 12.078182220458984,
      "learning_rate": 2.234708679924347e-05,
      "loss": 1.5023,
      "step": 36170
    },
    {
      "epoch": 5.978929973146045,
      "grad_norm": 27.949277877807617,
      "learning_rate": 2.2337905580344846e-05,
      "loss": 1.4331,
      "step": 36180
    },
    {
      "epoch": 5.980582524271845,
      "grad_norm": 12.060958862304688,
      "learning_rate": 2.2328724361446225e-05,
      "loss": 1.4613,
      "step": 36190
    },
    {
      "epoch": 5.982235075397645,
      "grad_norm": 12.409534454345703,
      "learning_rate": 2.2319543142547605e-05,
      "loss": 1.3824,
      "step": 36200
    },
    {
      "epoch": 5.983887626523446,
      "grad_norm": 11.61839485168457,
      "learning_rate": 2.2310361923648984e-05,
      "loss": 1.4633,
      "step": 36210
    },
    {
      "epoch": 5.985540177649246,
      "grad_norm": 17.775489807128906,
      "learning_rate": 2.2301180704750363e-05,
      "loss": 1.5373,
      "step": 36220
    },
    {
      "epoch": 5.987192728775047,
      "grad_norm": 13.473231315612793,
      "learning_rate": 2.2291999485851742e-05,
      "loss": 1.282,
      "step": 36230
    },
    {
      "epoch": 5.988845279900847,
      "grad_norm": 11.443365097045898,
      "learning_rate": 2.228281826695312e-05,
      "loss": 1.4227,
      "step": 36240
    },
    {
      "epoch": 5.990497831026647,
      "grad_norm": 9.717024803161621,
      "learning_rate": 2.22736370480545e-05,
      "loss": 1.5218,
      "step": 36250
    },
    {
      "epoch": 5.992150382152448,
      "grad_norm": 17.857152938842773,
      "learning_rate": 2.226445582915588e-05,
      "loss": 1.5035,
      "step": 36260
    },
    {
      "epoch": 5.993802933278248,
      "grad_norm": 10.381898880004883,
      "learning_rate": 2.225527461025726e-05,
      "loss": 1.4807,
      "step": 36270
    },
    {
      "epoch": 5.995455484404049,
      "grad_norm": 16.007429122924805,
      "learning_rate": 2.224609339135864e-05,
      "loss": 1.6168,
      "step": 36280
    },
    {
      "epoch": 5.997108035529849,
      "grad_norm": 8.93297290802002,
      "learning_rate": 2.2236912172460018e-05,
      "loss": 1.4956,
      "step": 36290
    },
    {
      "epoch": 5.998760586655649,
      "grad_norm": 10.40009880065918,
      "learning_rate": 2.2227730953561397e-05,
      "loss": 1.4969,
      "step": 36300
    },
    {
      "epoch": 5.99991737244371,
      "eval_accuracy": 0.3295502908213931,
      "eval_loss": 2.154688835144043,
      "eval_runtime": 818.3335,
      "eval_samples_per_second": 34.455,
      "eval_steps_per_second": 8.614,
      "step": 36307
    },
    {
      "epoch": 6.00041313778145,
      "grad_norm": 9.945453643798828,
      "learning_rate": 2.2218549734662776e-05,
      "loss": 1.5087,
      "step": 36310
    },
    {
      "epoch": 6.00206568890725,
      "grad_norm": 9.44550609588623,
      "learning_rate": 2.2209368515764152e-05,
      "loss": 1.2805,
      "step": 36320
    },
    {
      "epoch": 6.003718240033051,
      "grad_norm": 52.20871353149414,
      "learning_rate": 2.220018729686553e-05,
      "loss": 1.32,
      "step": 36330
    },
    {
      "epoch": 6.005370791158851,
      "grad_norm": 9.028908729553223,
      "learning_rate": 2.219100607796691e-05,
      "loss": 1.3254,
      "step": 36340
    },
    {
      "epoch": 6.0070233422846515,
      "grad_norm": 10.057682991027832,
      "learning_rate": 2.218182485906829e-05,
      "loss": 1.3984,
      "step": 36350
    },
    {
      "epoch": 6.0086758934104525,
      "grad_norm": 13.148024559020996,
      "learning_rate": 2.2172643640169673e-05,
      "loss": 1.4118,
      "step": 36360
    },
    {
      "epoch": 6.010328444536253,
      "grad_norm": 7.585305690765381,
      "learning_rate": 2.216346242127105e-05,
      "loss": 1.4126,
      "step": 36370
    },
    {
      "epoch": 6.011980995662054,
      "grad_norm": 16.880674362182617,
      "learning_rate": 2.2154281202372428e-05,
      "loss": 1.4296,
      "step": 36380
    },
    {
      "epoch": 6.013633546787854,
      "grad_norm": 9.740089416503906,
      "learning_rate": 2.2145099983473807e-05,
      "loss": 1.4044,
      "step": 36390
    },
    {
      "epoch": 6.015286097913654,
      "grad_norm": 16.309249877929688,
      "learning_rate": 2.2135918764575186e-05,
      "loss": 1.3367,
      "step": 36400
    },
    {
      "epoch": 6.016938649039455,
      "grad_norm": 88.99636840820312,
      "learning_rate": 2.2126737545676565e-05,
      "loss": 1.4252,
      "step": 36410
    },
    {
      "epoch": 6.018591200165255,
      "grad_norm": 20.119401931762695,
      "learning_rate": 2.2117556326777945e-05,
      "loss": 1.5106,
      "step": 36420
    },
    {
      "epoch": 6.020243751291056,
      "grad_norm": 26.70507049560547,
      "learning_rate": 2.2108375107879324e-05,
      "loss": 1.4082,
      "step": 36430
    },
    {
      "epoch": 6.021896302416856,
      "grad_norm": 10.413867950439453,
      "learning_rate": 2.2099193888980703e-05,
      "loss": 1.4838,
      "step": 36440
    },
    {
      "epoch": 6.023548853542657,
      "grad_norm": 20.84398078918457,
      "learning_rate": 2.209001267008208e-05,
      "loss": 1.4627,
      "step": 36450
    },
    {
      "epoch": 6.025201404668457,
      "grad_norm": 7.607133388519287,
      "learning_rate": 2.2080831451183458e-05,
      "loss": 1.5133,
      "step": 36460
    },
    {
      "epoch": 6.026853955794257,
      "grad_norm": 9.936842918395996,
      "learning_rate": 2.2071650232284838e-05,
      "loss": 1.4638,
      "step": 36470
    },
    {
      "epoch": 6.028506506920058,
      "grad_norm": 9.79660415649414,
      "learning_rate": 2.206246901338622e-05,
      "loss": 1.3432,
      "step": 36480
    },
    {
      "epoch": 6.030159058045858,
      "grad_norm": 14.434847831726074,
      "learning_rate": 2.20532877944876e-05,
      "loss": 1.4072,
      "step": 36490
    },
    {
      "epoch": 6.031811609171659,
      "grad_norm": 11.613893508911133,
      "learning_rate": 2.2044106575588975e-05,
      "loss": 1.3748,
      "step": 36500
    },
    {
      "epoch": 6.033464160297459,
      "grad_norm": 14.374581336975098,
      "learning_rate": 2.2034925356690355e-05,
      "loss": 1.5375,
      "step": 36510
    },
    {
      "epoch": 6.035116711423259,
      "grad_norm": 10.08636474609375,
      "learning_rate": 2.2025744137791734e-05,
      "loss": 1.4055,
      "step": 36520
    },
    {
      "epoch": 6.03676926254906,
      "grad_norm": 11.865565299987793,
      "learning_rate": 2.2016562918893113e-05,
      "loss": 1.3775,
      "step": 36530
    },
    {
      "epoch": 6.03842181367486,
      "grad_norm": 88.45521545410156,
      "learning_rate": 2.2007381699994492e-05,
      "loss": 1.5659,
      "step": 36540
    },
    {
      "epoch": 6.040074364800661,
      "grad_norm": 9.256975173950195,
      "learning_rate": 2.199820048109587e-05,
      "loss": 1.3508,
      "step": 36550
    },
    {
      "epoch": 6.0417269159264615,
      "grad_norm": 14.002177238464355,
      "learning_rate": 2.198901926219725e-05,
      "loss": 1.4102,
      "step": 36560
    },
    {
      "epoch": 6.043379467052262,
      "grad_norm": 21.786453247070312,
      "learning_rate": 2.197983804329863e-05,
      "loss": 1.5863,
      "step": 36570
    },
    {
      "epoch": 6.045032018178063,
      "grad_norm": 9.75375747680664,
      "learning_rate": 2.1970656824400006e-05,
      "loss": 1.3894,
      "step": 36580
    },
    {
      "epoch": 6.046684569303863,
      "grad_norm": 7.86843729019165,
      "learning_rate": 2.196147560550139e-05,
      "loss": 1.3645,
      "step": 36590
    },
    {
      "epoch": 6.048337120429664,
      "grad_norm": 7.051876068115234,
      "learning_rate": 2.1952294386602768e-05,
      "loss": 1.3968,
      "step": 36600
    },
    {
      "epoch": 6.049989671555464,
      "grad_norm": 53.84910202026367,
      "learning_rate": 2.1943113167704147e-05,
      "loss": 1.4543,
      "step": 36610
    },
    {
      "epoch": 6.051642222681264,
      "grad_norm": 8.938406944274902,
      "learning_rate": 2.1933931948805526e-05,
      "loss": 1.3795,
      "step": 36620
    },
    {
      "epoch": 6.053294773807065,
      "grad_norm": 10.21081829071045,
      "learning_rate": 2.1924750729906905e-05,
      "loss": 1.5895,
      "step": 36630
    },
    {
      "epoch": 6.054947324932865,
      "grad_norm": 11.11033821105957,
      "learning_rate": 2.191556951100828e-05,
      "loss": 1.4376,
      "step": 36640
    },
    {
      "epoch": 6.056599876058666,
      "grad_norm": 12.69739818572998,
      "learning_rate": 2.190638829210966e-05,
      "loss": 1.4503,
      "step": 36650
    },
    {
      "epoch": 6.058252427184466,
      "grad_norm": 14.584735870361328,
      "learning_rate": 2.189720707321104e-05,
      "loss": 1.3427,
      "step": 36660
    },
    {
      "epoch": 6.059904978310266,
      "grad_norm": 10.828611373901367,
      "learning_rate": 2.188802585431242e-05,
      "loss": 1.3451,
      "step": 36670
    },
    {
      "epoch": 6.061557529436067,
      "grad_norm": 7.586915493011475,
      "learning_rate": 2.18788446354138e-05,
      "loss": 1.3483,
      "step": 36680
    },
    {
      "epoch": 6.063210080561867,
      "grad_norm": 11.18470573425293,
      "learning_rate": 2.1869663416515178e-05,
      "loss": 1.4135,
      "step": 36690
    },
    {
      "epoch": 6.064862631687668,
      "grad_norm": 12.794292449951172,
      "learning_rate": 2.1860482197616557e-05,
      "loss": 1.3996,
      "step": 36700
    },
    {
      "epoch": 6.066515182813468,
      "grad_norm": 10.54262924194336,
      "learning_rate": 2.1851300978717936e-05,
      "loss": 1.5032,
      "step": 36710
    },
    {
      "epoch": 6.068167733939268,
      "grad_norm": 16.40874671936035,
      "learning_rate": 2.1842119759819315e-05,
      "loss": 1.4283,
      "step": 36720
    },
    {
      "epoch": 6.069820285065069,
      "grad_norm": 10.831640243530273,
      "learning_rate": 2.1832938540920695e-05,
      "loss": 1.472,
      "step": 36730
    },
    {
      "epoch": 6.071472836190869,
      "grad_norm": 10.649149894714355,
      "learning_rate": 2.1823757322022074e-05,
      "loss": 1.294,
      "step": 36740
    },
    {
      "epoch": 6.07312538731667,
      "grad_norm": 30.559856414794922,
      "learning_rate": 2.1814576103123453e-05,
      "loss": 1.4514,
      "step": 36750
    },
    {
      "epoch": 6.0747779384424705,
      "grad_norm": 10.63766860961914,
      "learning_rate": 2.1805394884224832e-05,
      "loss": 1.4639,
      "step": 36760
    },
    {
      "epoch": 6.076430489568271,
      "grad_norm": 14.730107307434082,
      "learning_rate": 2.1796213665326208e-05,
      "loss": 1.4646,
      "step": 36770
    },
    {
      "epoch": 6.078083040694072,
      "grad_norm": 15.621095657348633,
      "learning_rate": 2.1787032446427587e-05,
      "loss": 1.3344,
      "step": 36780
    },
    {
      "epoch": 6.079735591819872,
      "grad_norm": 15.240421295166016,
      "learning_rate": 2.1777851227528967e-05,
      "loss": 1.4834,
      "step": 36790
    },
    {
      "epoch": 6.081388142945673,
      "grad_norm": 8.048723220825195,
      "learning_rate": 2.1768670008630346e-05,
      "loss": 1.3243,
      "step": 36800
    },
    {
      "epoch": 6.083040694071473,
      "grad_norm": 14.045221328735352,
      "learning_rate": 2.1759488789731725e-05,
      "loss": 1.4613,
      "step": 36810
    },
    {
      "epoch": 6.084693245197274,
      "grad_norm": 9.798041343688965,
      "learning_rate": 2.1750307570833104e-05,
      "loss": 1.4506,
      "step": 36820
    },
    {
      "epoch": 6.086345796323074,
      "grad_norm": 9.147562980651855,
      "learning_rate": 2.1741126351934484e-05,
      "loss": 1.4779,
      "step": 36830
    },
    {
      "epoch": 6.087998347448874,
      "grad_norm": 7.114458084106445,
      "learning_rate": 2.1731945133035863e-05,
      "loss": 1.3534,
      "step": 36840
    },
    {
      "epoch": 6.089650898574675,
      "grad_norm": 15.428686141967773,
      "learning_rate": 2.1722763914137242e-05,
      "loss": 1.4559,
      "step": 36850
    },
    {
      "epoch": 6.091303449700475,
      "grad_norm": 8.084449768066406,
      "learning_rate": 2.171358269523862e-05,
      "loss": 1.4538,
      "step": 36860
    },
    {
      "epoch": 6.092956000826276,
      "grad_norm": 13.73076343536377,
      "learning_rate": 2.170440147634e-05,
      "loss": 1.3203,
      "step": 36870
    },
    {
      "epoch": 6.094608551952076,
      "grad_norm": 12.01843547821045,
      "learning_rate": 2.169522025744138e-05,
      "loss": 1.3937,
      "step": 36880
    },
    {
      "epoch": 6.096261103077876,
      "grad_norm": 11.336709022521973,
      "learning_rate": 2.168603903854276e-05,
      "loss": 1.4382,
      "step": 36890
    },
    {
      "epoch": 6.097913654203677,
      "grad_norm": 20.88669776916504,
      "learning_rate": 2.1676857819644135e-05,
      "loss": 1.3326,
      "step": 36900
    },
    {
      "epoch": 6.099566205329477,
      "grad_norm": 12.064225196838379,
      "learning_rate": 2.1667676600745514e-05,
      "loss": 1.4931,
      "step": 36910
    },
    {
      "epoch": 6.101218756455278,
      "grad_norm": 10.633503913879395,
      "learning_rate": 2.1658495381846894e-05,
      "loss": 1.5304,
      "step": 36920
    },
    {
      "epoch": 6.102871307581078,
      "grad_norm": 11.907389640808105,
      "learning_rate": 2.1649314162948276e-05,
      "loss": 1.4893,
      "step": 36930
    },
    {
      "epoch": 6.104523858706878,
      "grad_norm": 11.007550239562988,
      "learning_rate": 2.1640132944049655e-05,
      "loss": 1.4218,
      "step": 36940
    },
    {
      "epoch": 6.106176409832679,
      "grad_norm": 7.1882405281066895,
      "learning_rate": 2.1630951725151035e-05,
      "loss": 1.347,
      "step": 36950
    },
    {
      "epoch": 6.1078289609584795,
      "grad_norm": 21.802770614624023,
      "learning_rate": 2.162177050625241e-05,
      "loss": 1.3269,
      "step": 36960
    },
    {
      "epoch": 6.1094815120842805,
      "grad_norm": 28.251907348632812,
      "learning_rate": 2.161258928735379e-05,
      "loss": 1.5196,
      "step": 36970
    },
    {
      "epoch": 6.111134063210081,
      "grad_norm": 13.803258895874023,
      "learning_rate": 2.160340806845517e-05,
      "loss": 1.45,
      "step": 36980
    },
    {
      "epoch": 6.112786614335881,
      "grad_norm": 26.397541046142578,
      "learning_rate": 2.1594226849556548e-05,
      "loss": 1.4174,
      "step": 36990
    },
    {
      "epoch": 6.114439165461682,
      "grad_norm": 16.90498924255371,
      "learning_rate": 2.1585045630657927e-05,
      "loss": 1.4125,
      "step": 37000
    },
    {
      "epoch": 6.116091716587482,
      "grad_norm": 17.150217056274414,
      "learning_rate": 2.1575864411759307e-05,
      "loss": 1.411,
      "step": 37010
    },
    {
      "epoch": 6.117744267713283,
      "grad_norm": 11.79568862915039,
      "learning_rate": 2.1566683192860686e-05,
      "loss": 1.4386,
      "step": 37020
    },
    {
      "epoch": 6.119396818839083,
      "grad_norm": 10.997562408447266,
      "learning_rate": 2.1557501973962062e-05,
      "loss": 1.4482,
      "step": 37030
    },
    {
      "epoch": 6.121049369964883,
      "grad_norm": 14.64095687866211,
      "learning_rate": 2.154832075506344e-05,
      "loss": 1.4078,
      "step": 37040
    },
    {
      "epoch": 6.122701921090684,
      "grad_norm": 7.124085426330566,
      "learning_rate": 2.1539139536164824e-05,
      "loss": 1.489,
      "step": 37050
    },
    {
      "epoch": 6.124354472216484,
      "grad_norm": 16.14177894592285,
      "learning_rate": 2.1529958317266203e-05,
      "loss": 1.449,
      "step": 37060
    },
    {
      "epoch": 6.126007023342285,
      "grad_norm": 14.747572898864746,
      "learning_rate": 2.1520777098367582e-05,
      "loss": 1.4958,
      "step": 37070
    },
    {
      "epoch": 6.127659574468085,
      "grad_norm": 15.584579467773438,
      "learning_rate": 2.151159587946896e-05,
      "loss": 1.4513,
      "step": 37080
    },
    {
      "epoch": 6.129312125593885,
      "grad_norm": 28.18446922302246,
      "learning_rate": 2.1502414660570337e-05,
      "loss": 1.3798,
      "step": 37090
    },
    {
      "epoch": 6.130964676719686,
      "grad_norm": 37.316654205322266,
      "learning_rate": 2.1493233441671717e-05,
      "loss": 1.5111,
      "step": 37100
    },
    {
      "epoch": 6.132617227845486,
      "grad_norm": 8.39609432220459,
      "learning_rate": 2.1484052222773096e-05,
      "loss": 1.5339,
      "step": 37110
    },
    {
      "epoch": 6.134269778971287,
      "grad_norm": 14.053455352783203,
      "learning_rate": 2.1474871003874475e-05,
      "loss": 1.4811,
      "step": 37120
    },
    {
      "epoch": 6.135922330097087,
      "grad_norm": 10.345773696899414,
      "learning_rate": 2.1465689784975854e-05,
      "loss": 1.4089,
      "step": 37130
    },
    {
      "epoch": 6.137574881222887,
      "grad_norm": 7.838363170623779,
      "learning_rate": 2.1456508566077234e-05,
      "loss": 1.4742,
      "step": 37140
    },
    {
      "epoch": 6.139227432348688,
      "grad_norm": 22.944093704223633,
      "learning_rate": 2.1447327347178613e-05,
      "loss": 1.4821,
      "step": 37150
    },
    {
      "epoch": 6.1408799834744885,
      "grad_norm": 8.07544994354248,
      "learning_rate": 2.1438146128279992e-05,
      "loss": 1.3206,
      "step": 37160
    },
    {
      "epoch": 6.1425325346002895,
      "grad_norm": 99.47127532958984,
      "learning_rate": 2.142896490938137e-05,
      "loss": 1.3522,
      "step": 37170
    },
    {
      "epoch": 6.14418508572609,
      "grad_norm": 10.434372901916504,
      "learning_rate": 2.141978369048275e-05,
      "loss": 1.4843,
      "step": 37180
    },
    {
      "epoch": 6.1458376368518906,
      "grad_norm": 9.524206161499023,
      "learning_rate": 2.141060247158413e-05,
      "loss": 1.4098,
      "step": 37190
    },
    {
      "epoch": 6.147490187977691,
      "grad_norm": 7.026861667633057,
      "learning_rate": 2.140142125268551e-05,
      "loss": 1.3957,
      "step": 37200
    },
    {
      "epoch": 6.149142739103491,
      "grad_norm": 12.716541290283203,
      "learning_rate": 2.1392240033786888e-05,
      "loss": 1.5463,
      "step": 37210
    },
    {
      "epoch": 6.150795290229292,
      "grad_norm": 38.672367095947266,
      "learning_rate": 2.1383058814888264e-05,
      "loss": 1.4292,
      "step": 37220
    },
    {
      "epoch": 6.152447841355092,
      "grad_norm": 14.989725112915039,
      "learning_rate": 2.1373877595989643e-05,
      "loss": 1.4267,
      "step": 37230
    },
    {
      "epoch": 6.154100392480893,
      "grad_norm": 9.272750854492188,
      "learning_rate": 2.1364696377091023e-05,
      "loss": 1.4773,
      "step": 37240
    },
    {
      "epoch": 6.155752943606693,
      "grad_norm": 13.461223602294922,
      "learning_rate": 2.1355515158192402e-05,
      "loss": 1.3765,
      "step": 37250
    },
    {
      "epoch": 6.157405494732493,
      "grad_norm": 11.42111587524414,
      "learning_rate": 2.134633393929378e-05,
      "loss": 1.399,
      "step": 37260
    },
    {
      "epoch": 6.159058045858294,
      "grad_norm": 9.016616821289062,
      "learning_rate": 2.133715272039516e-05,
      "loss": 1.596,
      "step": 37270
    },
    {
      "epoch": 6.160710596984094,
      "grad_norm": 8.332453727722168,
      "learning_rate": 2.132797150149654e-05,
      "loss": 1.388,
      "step": 37280
    },
    {
      "epoch": 6.162363148109895,
      "grad_norm": 17.8102970123291,
      "learning_rate": 2.131879028259792e-05,
      "loss": 1.3235,
      "step": 37290
    },
    {
      "epoch": 6.164015699235695,
      "grad_norm": 11.08936882019043,
      "learning_rate": 2.1309609063699298e-05,
      "loss": 1.3742,
      "step": 37300
    },
    {
      "epoch": 6.165668250361495,
      "grad_norm": 8.341262817382812,
      "learning_rate": 2.1300427844800677e-05,
      "loss": 1.4266,
      "step": 37310
    },
    {
      "epoch": 6.167320801487296,
      "grad_norm": 18.066402435302734,
      "learning_rate": 2.1291246625902057e-05,
      "loss": 1.6203,
      "step": 37320
    },
    {
      "epoch": 6.168973352613096,
      "grad_norm": 19.090791702270508,
      "learning_rate": 2.1282065407003436e-05,
      "loss": 1.4881,
      "step": 37330
    },
    {
      "epoch": 6.170625903738897,
      "grad_norm": 7.528576850891113,
      "learning_rate": 2.1272884188104815e-05,
      "loss": 1.3519,
      "step": 37340
    },
    {
      "epoch": 6.172278454864697,
      "grad_norm": 19.304250717163086,
      "learning_rate": 2.126370296920619e-05,
      "loss": 1.4546,
      "step": 37350
    },
    {
      "epoch": 6.1739310059904975,
      "grad_norm": 7.075794219970703,
      "learning_rate": 2.125452175030757e-05,
      "loss": 1.4554,
      "step": 37360
    },
    {
      "epoch": 6.1755835571162985,
      "grad_norm": 12.549452781677246,
      "learning_rate": 2.124534053140895e-05,
      "loss": 1.5573,
      "step": 37370
    },
    {
      "epoch": 6.177236108242099,
      "grad_norm": 9.695381164550781,
      "learning_rate": 2.123615931251033e-05,
      "loss": 1.3744,
      "step": 37380
    },
    {
      "epoch": 6.1788886593678995,
      "grad_norm": 21.600099563598633,
      "learning_rate": 2.122697809361171e-05,
      "loss": 1.5132,
      "step": 37390
    },
    {
      "epoch": 6.1805412104937,
      "grad_norm": 9.879249572753906,
      "learning_rate": 2.121779687471309e-05,
      "loss": 1.4469,
      "step": 37400
    },
    {
      "epoch": 6.1821937616195,
      "grad_norm": 12.330972671508789,
      "learning_rate": 2.1208615655814466e-05,
      "loss": 1.5801,
      "step": 37410
    },
    {
      "epoch": 6.183846312745301,
      "grad_norm": 12.194676399230957,
      "learning_rate": 2.1199434436915846e-05,
      "loss": 1.3992,
      "step": 37420
    },
    {
      "epoch": 6.185498863871101,
      "grad_norm": 14.351995468139648,
      "learning_rate": 2.1190253218017225e-05,
      "loss": 1.5508,
      "step": 37430
    },
    {
      "epoch": 6.187151414996902,
      "grad_norm": 9.946985244750977,
      "learning_rate": 2.1181071999118604e-05,
      "loss": 1.5336,
      "step": 37440
    },
    {
      "epoch": 6.188803966122702,
      "grad_norm": 10.623150825500488,
      "learning_rate": 2.1171890780219983e-05,
      "loss": 1.4028,
      "step": 37450
    },
    {
      "epoch": 6.190456517248502,
      "grad_norm": 12.726574897766113,
      "learning_rate": 2.1162709561321363e-05,
      "loss": 1.4419,
      "step": 37460
    },
    {
      "epoch": 6.192109068374303,
      "grad_norm": 11.567181587219238,
      "learning_rate": 2.1153528342422742e-05,
      "loss": 1.3462,
      "step": 37470
    },
    {
      "epoch": 6.193761619500103,
      "grad_norm": 15.38206672668457,
      "learning_rate": 2.1144347123524118e-05,
      "loss": 1.4159,
      "step": 37480
    },
    {
      "epoch": 6.195414170625904,
      "grad_norm": 9.006597518920898,
      "learning_rate": 2.1135165904625497e-05,
      "loss": 1.3282,
      "step": 37490
    },
    {
      "epoch": 6.197066721751704,
      "grad_norm": 16.817569732666016,
      "learning_rate": 2.112598468572688e-05,
      "loss": 1.3807,
      "step": 37500
    },
    {
      "epoch": 6.198719272877504,
      "grad_norm": 14.372156143188477,
      "learning_rate": 2.111680346682826e-05,
      "loss": 1.5622,
      "step": 37510
    },
    {
      "epoch": 6.200371824003305,
      "grad_norm": 10.266152381896973,
      "learning_rate": 2.1107622247929638e-05,
      "loss": 1.4466,
      "step": 37520
    },
    {
      "epoch": 6.202024375129105,
      "grad_norm": 15.889741897583008,
      "learning_rate": 2.1098441029031017e-05,
      "loss": 1.4577,
      "step": 37530
    },
    {
      "epoch": 6.203676926254906,
      "grad_norm": 11.235896110534668,
      "learning_rate": 2.1089259810132393e-05,
      "loss": 1.4971,
      "step": 37540
    },
    {
      "epoch": 6.205329477380706,
      "grad_norm": 10.462620735168457,
      "learning_rate": 2.1080078591233773e-05,
      "loss": 1.3453,
      "step": 37550
    },
    {
      "epoch": 6.206982028506507,
      "grad_norm": 9.256996154785156,
      "learning_rate": 2.1070897372335152e-05,
      "loss": 1.4294,
      "step": 37560
    },
    {
      "epoch": 6.2086345796323075,
      "grad_norm": 9.670038223266602,
      "learning_rate": 2.106171615343653e-05,
      "loss": 1.498,
      "step": 37570
    },
    {
      "epoch": 6.210287130758108,
      "grad_norm": 10.330695152282715,
      "learning_rate": 2.105253493453791e-05,
      "loss": 1.4457,
      "step": 37580
    },
    {
      "epoch": 6.2119396818839085,
      "grad_norm": 13.623739242553711,
      "learning_rate": 2.104335371563929e-05,
      "loss": 1.52,
      "step": 37590
    },
    {
      "epoch": 6.213592233009709,
      "grad_norm": 8.723990440368652,
      "learning_rate": 2.103417249674067e-05,
      "loss": 1.4207,
      "step": 37600
    },
    {
      "epoch": 6.215244784135509,
      "grad_norm": 10.08510684967041,
      "learning_rate": 2.1024991277842045e-05,
      "loss": 1.4864,
      "step": 37610
    },
    {
      "epoch": 6.21689733526131,
      "grad_norm": 13.55770206451416,
      "learning_rate": 2.1015810058943427e-05,
      "loss": 1.3481,
      "step": 37620
    },
    {
      "epoch": 6.21854988638711,
      "grad_norm": 26.282560348510742,
      "learning_rate": 2.1006628840044807e-05,
      "loss": 1.4423,
      "step": 37630
    },
    {
      "epoch": 6.220202437512911,
      "grad_norm": 13.998091697692871,
      "learning_rate": 2.0997447621146186e-05,
      "loss": 1.5324,
      "step": 37640
    },
    {
      "epoch": 6.221854988638711,
      "grad_norm": 10.550564765930176,
      "learning_rate": 2.0988266402247565e-05,
      "loss": 1.4653,
      "step": 37650
    },
    {
      "epoch": 6.223507539764512,
      "grad_norm": 9.701068878173828,
      "learning_rate": 2.0979085183348944e-05,
      "loss": 1.4896,
      "step": 37660
    },
    {
      "epoch": 6.225160090890312,
      "grad_norm": 19.163440704345703,
      "learning_rate": 2.096990396445032e-05,
      "loss": 1.3897,
      "step": 37670
    },
    {
      "epoch": 6.226812642016112,
      "grad_norm": 21.417760848999023,
      "learning_rate": 2.09607227455517e-05,
      "loss": 1.3892,
      "step": 37680
    },
    {
      "epoch": 6.228465193141913,
      "grad_norm": 13.961776733398438,
      "learning_rate": 2.095154152665308e-05,
      "loss": 1.3522,
      "step": 37690
    },
    {
      "epoch": 6.230117744267713,
      "grad_norm": 7.012969493865967,
      "learning_rate": 2.0942360307754458e-05,
      "loss": 1.3648,
      "step": 37700
    },
    {
      "epoch": 6.231770295393514,
      "grad_norm": 15.90501594543457,
      "learning_rate": 2.0933179088855837e-05,
      "loss": 1.4158,
      "step": 37710
    },
    {
      "epoch": 6.233422846519314,
      "grad_norm": 16.21110725402832,
      "learning_rate": 2.0923997869957216e-05,
      "loss": 1.5681,
      "step": 37720
    },
    {
      "epoch": 6.235075397645114,
      "grad_norm": 11.805452346801758,
      "learning_rate": 2.0914816651058596e-05,
      "loss": 1.387,
      "step": 37730
    },
    {
      "epoch": 6.236727948770915,
      "grad_norm": 13.68657398223877,
      "learning_rate": 2.0905635432159975e-05,
      "loss": 1.3717,
      "step": 37740
    },
    {
      "epoch": 6.238380499896715,
      "grad_norm": 9.461950302124023,
      "learning_rate": 2.0896454213261354e-05,
      "loss": 1.3141,
      "step": 37750
    },
    {
      "epoch": 6.240033051022516,
      "grad_norm": 9.578817367553711,
      "learning_rate": 2.0887272994362733e-05,
      "loss": 1.3581,
      "step": 37760
    },
    {
      "epoch": 6.2416856021483165,
      "grad_norm": 11.16032600402832,
      "learning_rate": 2.0878091775464113e-05,
      "loss": 1.4463,
      "step": 37770
    },
    {
      "epoch": 6.243338153274117,
      "grad_norm": 36.12056350708008,
      "learning_rate": 2.0868910556565492e-05,
      "loss": 1.3249,
      "step": 37780
    },
    {
      "epoch": 6.2449907043999175,
      "grad_norm": 17.16036605834961,
      "learning_rate": 2.085972933766687e-05,
      "loss": 1.4926,
      "step": 37790
    },
    {
      "epoch": 6.246643255525718,
      "grad_norm": 8.647866249084473,
      "learning_rate": 2.0850548118768247e-05,
      "loss": 1.3782,
      "step": 37800
    },
    {
      "epoch": 6.248295806651519,
      "grad_norm": 8.385272026062012,
      "learning_rate": 2.0841366899869626e-05,
      "loss": 1.4198,
      "step": 37810
    },
    {
      "epoch": 6.249948357777319,
      "grad_norm": 14.405436515808105,
      "learning_rate": 2.0832185680971005e-05,
      "loss": 1.3887,
      "step": 37820
    },
    {
      "epoch": 6.251600908903119,
      "grad_norm": 7.519594669342041,
      "learning_rate": 2.0823004462072385e-05,
      "loss": 1.3571,
      "step": 37830
    },
    {
      "epoch": 6.25325346002892,
      "grad_norm": 27.60797882080078,
      "learning_rate": 2.0813823243173764e-05,
      "loss": 1.5992,
      "step": 37840
    },
    {
      "epoch": 6.25490601115472,
      "grad_norm": 11.991599082946777,
      "learning_rate": 2.0804642024275147e-05,
      "loss": 1.4627,
      "step": 37850
    },
    {
      "epoch": 6.256558562280521,
      "grad_norm": 12.27054500579834,
      "learning_rate": 2.0795460805376522e-05,
      "loss": 1.3893,
      "step": 37860
    },
    {
      "epoch": 6.258211113406321,
      "grad_norm": 33.28194808959961,
      "learning_rate": 2.0786279586477902e-05,
      "loss": 1.4389,
      "step": 37870
    },
    {
      "epoch": 6.259863664532121,
      "grad_norm": 22.726533889770508,
      "learning_rate": 2.077709836757928e-05,
      "loss": 1.5418,
      "step": 37880
    },
    {
      "epoch": 6.261516215657922,
      "grad_norm": 13.192795753479004,
      "learning_rate": 2.076791714868066e-05,
      "loss": 1.3905,
      "step": 37890
    },
    {
      "epoch": 6.263168766783722,
      "grad_norm": 14.053077697753906,
      "learning_rate": 2.075873592978204e-05,
      "loss": 1.4147,
      "step": 37900
    },
    {
      "epoch": 6.264821317909523,
      "grad_norm": 16.81157875061035,
      "learning_rate": 2.074955471088342e-05,
      "loss": 1.3977,
      "step": 37910
    },
    {
      "epoch": 6.266473869035323,
      "grad_norm": 7.91226863861084,
      "learning_rate": 2.0740373491984798e-05,
      "loss": 1.3966,
      "step": 37920
    },
    {
      "epoch": 6.268126420161124,
      "grad_norm": 13.343158721923828,
      "learning_rate": 2.0731192273086174e-05,
      "loss": 1.4248,
      "step": 37930
    },
    {
      "epoch": 6.269778971286924,
      "grad_norm": 10.185911178588867,
      "learning_rate": 2.0722011054187553e-05,
      "loss": 1.4735,
      "step": 37940
    },
    {
      "epoch": 6.271431522412724,
      "grad_norm": 8.653183937072754,
      "learning_rate": 2.0712829835288932e-05,
      "loss": 1.4104,
      "step": 37950
    },
    {
      "epoch": 6.273084073538525,
      "grad_norm": 16.011350631713867,
      "learning_rate": 2.0703648616390315e-05,
      "loss": 1.4787,
      "step": 37960
    },
    {
      "epoch": 6.2747366246643255,
      "grad_norm": 11.425875663757324,
      "learning_rate": 2.0694467397491694e-05,
      "loss": 1.5623,
      "step": 37970
    },
    {
      "epoch": 6.276389175790126,
      "grad_norm": 10.677700996398926,
      "learning_rate": 2.0685286178593073e-05,
      "loss": 1.4261,
      "step": 37980
    },
    {
      "epoch": 6.2780417269159265,
      "grad_norm": 19.281002044677734,
      "learning_rate": 2.067610495969445e-05,
      "loss": 1.7146,
      "step": 37990
    },
    {
      "epoch": 6.279694278041727,
      "grad_norm": 16.140913009643555,
      "learning_rate": 2.066692374079583e-05,
      "loss": 1.4536,
      "step": 38000
    },
    {
      "epoch": 6.281346829167528,
      "grad_norm": 11.096719741821289,
      "learning_rate": 2.0657742521897208e-05,
      "loss": 1.3705,
      "step": 38010
    },
    {
      "epoch": 6.282999380293328,
      "grad_norm": 18.13045883178711,
      "learning_rate": 2.0648561302998587e-05,
      "loss": 1.4394,
      "step": 38020
    },
    {
      "epoch": 6.284651931419129,
      "grad_norm": 14.277056694030762,
      "learning_rate": 2.0639380084099966e-05,
      "loss": 1.46,
      "step": 38030
    },
    {
      "epoch": 6.286304482544929,
      "grad_norm": 12.03947925567627,
      "learning_rate": 2.0630198865201346e-05,
      "loss": 1.2961,
      "step": 38040
    },
    {
      "epoch": 6.287957033670729,
      "grad_norm": 11.471107482910156,
      "learning_rate": 2.0621017646302725e-05,
      "loss": 1.3797,
      "step": 38050
    },
    {
      "epoch": 6.28960958479653,
      "grad_norm": 14.753610610961914,
      "learning_rate": 2.06118364274041e-05,
      "loss": 1.5292,
      "step": 38060
    },
    {
      "epoch": 6.29126213592233,
      "grad_norm": 9.602330207824707,
      "learning_rate": 2.060265520850548e-05,
      "loss": 1.3428,
      "step": 38070
    },
    {
      "epoch": 6.292914687048131,
      "grad_norm": 12.456007957458496,
      "learning_rate": 2.0593473989606863e-05,
      "loss": 1.4892,
      "step": 38080
    },
    {
      "epoch": 6.294567238173931,
      "grad_norm": 8.093803405761719,
      "learning_rate": 2.0584292770708242e-05,
      "loss": 1.4064,
      "step": 38090
    },
    {
      "epoch": 6.296219789299731,
      "grad_norm": 10.36359691619873,
      "learning_rate": 2.057511155180962e-05,
      "loss": 1.4547,
      "step": 38100
    },
    {
      "epoch": 6.297872340425532,
      "grad_norm": 10.041078567504883,
      "learning_rate": 2.0565930332911e-05,
      "loss": 1.4408,
      "step": 38110
    },
    {
      "epoch": 6.299524891551332,
      "grad_norm": 12.253533363342285,
      "learning_rate": 2.0556749114012376e-05,
      "loss": 1.589,
      "step": 38120
    },
    {
      "epoch": 6.301177442677133,
      "grad_norm": 10.570530891418457,
      "learning_rate": 2.0547567895113755e-05,
      "loss": 1.3889,
      "step": 38130
    },
    {
      "epoch": 6.302829993802933,
      "grad_norm": 13.251456260681152,
      "learning_rate": 2.0538386676215135e-05,
      "loss": 1.5385,
      "step": 38140
    },
    {
      "epoch": 6.304482544928733,
      "grad_norm": 27.627914428710938,
      "learning_rate": 2.0529205457316514e-05,
      "loss": 1.5747,
      "step": 38150
    },
    {
      "epoch": 6.306135096054534,
      "grad_norm": 9.486604690551758,
      "learning_rate": 2.0520024238417893e-05,
      "loss": 1.4378,
      "step": 38160
    },
    {
      "epoch": 6.3077876471803345,
      "grad_norm": 6.460193157196045,
      "learning_rate": 2.0510843019519272e-05,
      "loss": 1.444,
      "step": 38170
    },
    {
      "epoch": 6.3094401983061354,
      "grad_norm": 48.902530670166016,
      "learning_rate": 2.050166180062065e-05,
      "loss": 1.3742,
      "step": 38180
    },
    {
      "epoch": 6.3110927494319355,
      "grad_norm": 8.491707801818848,
      "learning_rate": 2.049248058172203e-05,
      "loss": 1.4993,
      "step": 38190
    },
    {
      "epoch": 6.312745300557736,
      "grad_norm": 14.18362808227539,
      "learning_rate": 2.048329936282341e-05,
      "loss": 1.5047,
      "step": 38200
    },
    {
      "epoch": 6.314397851683537,
      "grad_norm": 15.740824699401855,
      "learning_rate": 2.047411814392479e-05,
      "loss": 1.3403,
      "step": 38210
    },
    {
      "epoch": 6.316050402809337,
      "grad_norm": 11.892169952392578,
      "learning_rate": 2.046493692502617e-05,
      "loss": 1.5607,
      "step": 38220
    },
    {
      "epoch": 6.317702953935138,
      "grad_norm": 8.451154708862305,
      "learning_rate": 2.0455755706127548e-05,
      "loss": 1.3493,
      "step": 38230
    },
    {
      "epoch": 6.319355505060938,
      "grad_norm": 14.7618408203125,
      "learning_rate": 2.0446574487228927e-05,
      "loss": 1.504,
      "step": 38240
    },
    {
      "epoch": 6.321008056186738,
      "grad_norm": 10.903997421264648,
      "learning_rate": 2.0437393268330303e-05,
      "loss": 1.3993,
      "step": 38250
    },
    {
      "epoch": 6.322660607312539,
      "grad_norm": 12.13855266571045,
      "learning_rate": 2.0428212049431682e-05,
      "loss": 1.478,
      "step": 38260
    },
    {
      "epoch": 6.324313158438339,
      "grad_norm": 13.594361305236816,
      "learning_rate": 2.041903083053306e-05,
      "loss": 1.5079,
      "step": 38270
    },
    {
      "epoch": 6.32596570956414,
      "grad_norm": 9.111920356750488,
      "learning_rate": 2.040984961163444e-05,
      "loss": 1.5341,
      "step": 38280
    },
    {
      "epoch": 6.32761826068994,
      "grad_norm": 8.151607513427734,
      "learning_rate": 2.040066839273582e-05,
      "loss": 1.4063,
      "step": 38290
    },
    {
      "epoch": 6.32927081181574,
      "grad_norm": 9.332478523254395,
      "learning_rate": 2.0391487173837203e-05,
      "loss": 1.5013,
      "step": 38300
    },
    {
      "epoch": 6.330923362941541,
      "grad_norm": 20.929075241088867,
      "learning_rate": 2.038230595493858e-05,
      "loss": 1.3706,
      "step": 38310
    },
    {
      "epoch": 6.332575914067341,
      "grad_norm": 16.67856788635254,
      "learning_rate": 2.0373124736039958e-05,
      "loss": 1.4256,
      "step": 38320
    },
    {
      "epoch": 6.334228465193142,
      "grad_norm": 9.978659629821777,
      "learning_rate": 2.0363943517141337e-05,
      "loss": 1.5506,
      "step": 38330
    },
    {
      "epoch": 6.335881016318942,
      "grad_norm": 7.140017032623291,
      "learning_rate": 2.0354762298242716e-05,
      "loss": 1.3496,
      "step": 38340
    },
    {
      "epoch": 6.337533567444742,
      "grad_norm": 7.583073616027832,
      "learning_rate": 2.0345581079344095e-05,
      "loss": 1.2338,
      "step": 38350
    },
    {
      "epoch": 6.339186118570543,
      "grad_norm": 27.26841163635254,
      "learning_rate": 2.0336399860445475e-05,
      "loss": 1.3861,
      "step": 38360
    },
    {
      "epoch": 6.3408386696963435,
      "grad_norm": 22.0368595123291,
      "learning_rate": 2.0327218641546854e-05,
      "loss": 1.4712,
      "step": 38370
    },
    {
      "epoch": 6.3424912208221444,
      "grad_norm": 13.927277565002441,
      "learning_rate": 2.031803742264823e-05,
      "loss": 1.5026,
      "step": 38380
    },
    {
      "epoch": 6.3441437719479445,
      "grad_norm": 13.221909523010254,
      "learning_rate": 2.030885620374961e-05,
      "loss": 1.3972,
      "step": 38390
    },
    {
      "epoch": 6.3457963230737455,
      "grad_norm": 13.610274314880371,
      "learning_rate": 2.0299674984850988e-05,
      "loss": 1.5108,
      "step": 38400
    },
    {
      "epoch": 6.347448874199546,
      "grad_norm": 16.176225662231445,
      "learning_rate": 2.0290493765952368e-05,
      "loss": 1.3653,
      "step": 38410
    },
    {
      "epoch": 6.349101425325346,
      "grad_norm": 21.335020065307617,
      "learning_rate": 2.028131254705375e-05,
      "loss": 1.5748,
      "step": 38420
    },
    {
      "epoch": 6.350753976451147,
      "grad_norm": 17.27007293701172,
      "learning_rate": 2.027213132815513e-05,
      "loss": 1.4135,
      "step": 38430
    },
    {
      "epoch": 6.352406527576947,
      "grad_norm": 19.558349609375,
      "learning_rate": 2.0262950109256505e-05,
      "loss": 1.398,
      "step": 38440
    },
    {
      "epoch": 6.354059078702748,
      "grad_norm": 7.04690408706665,
      "learning_rate": 2.0253768890357885e-05,
      "loss": 1.3721,
      "step": 38450
    },
    {
      "epoch": 6.355711629828548,
      "grad_norm": 9.508496284484863,
      "learning_rate": 2.0244587671459264e-05,
      "loss": 1.3303,
      "step": 38460
    },
    {
      "epoch": 6.357364180954348,
      "grad_norm": 21.893522262573242,
      "learning_rate": 2.0235406452560643e-05,
      "loss": 1.3006,
      "step": 38470
    },
    {
      "epoch": 6.359016732080149,
      "grad_norm": 10.605900764465332,
      "learning_rate": 2.0226225233662022e-05,
      "loss": 1.5355,
      "step": 38480
    },
    {
      "epoch": 6.360669283205949,
      "grad_norm": 43.09657669067383,
      "learning_rate": 2.02170440147634e-05,
      "loss": 1.4263,
      "step": 38490
    },
    {
      "epoch": 6.36232183433175,
      "grad_norm": 11.316243171691895,
      "learning_rate": 2.020786279586478e-05,
      "loss": 1.435,
      "step": 38500
    },
    {
      "epoch": 6.36397438545755,
      "grad_norm": 13.281632423400879,
      "learning_rate": 2.0198681576966157e-05,
      "loss": 1.496,
      "step": 38510
    },
    {
      "epoch": 6.36562693658335,
      "grad_norm": 11.309965133666992,
      "learning_rate": 2.0189500358067536e-05,
      "loss": 1.4598,
      "step": 38520
    },
    {
      "epoch": 6.367279487709151,
      "grad_norm": 13.029101371765137,
      "learning_rate": 2.018031913916892e-05,
      "loss": 1.5399,
      "step": 38530
    },
    {
      "epoch": 6.368932038834951,
      "grad_norm": 112.3958740234375,
      "learning_rate": 2.0171137920270298e-05,
      "loss": 1.3838,
      "step": 38540
    },
    {
      "epoch": 6.370584589960752,
      "grad_norm": 9.24577808380127,
      "learning_rate": 2.0161956701371677e-05,
      "loss": 1.3194,
      "step": 38550
    },
    {
      "epoch": 6.372237141086552,
      "grad_norm": 14.375222206115723,
      "learning_rate": 2.0152775482473056e-05,
      "loss": 1.1996,
      "step": 38560
    },
    {
      "epoch": 6.3738896922123525,
      "grad_norm": 19.481122970581055,
      "learning_rate": 2.0143594263574432e-05,
      "loss": 1.5943,
      "step": 38570
    },
    {
      "epoch": 6.3755422433381534,
      "grad_norm": 13.95479679107666,
      "learning_rate": 2.013441304467581e-05,
      "loss": 1.4706,
      "step": 38580
    },
    {
      "epoch": 6.3771947944639535,
      "grad_norm": 13.144954681396484,
      "learning_rate": 2.012523182577719e-05,
      "loss": 1.3673,
      "step": 38590
    },
    {
      "epoch": 6.3788473455897545,
      "grad_norm": 12.10051155090332,
      "learning_rate": 2.011605060687857e-05,
      "loss": 1.2891,
      "step": 38600
    },
    {
      "epoch": 6.380499896715555,
      "grad_norm": 27.080062866210938,
      "learning_rate": 2.010686938797995e-05,
      "loss": 1.4499,
      "step": 38610
    },
    {
      "epoch": 6.382152447841355,
      "grad_norm": 14.484885215759277,
      "learning_rate": 2.009768816908133e-05,
      "loss": 1.3468,
      "step": 38620
    },
    {
      "epoch": 6.383804998967156,
      "grad_norm": 10.011687278747559,
      "learning_rate": 2.0088506950182708e-05,
      "loss": 1.4775,
      "step": 38630
    },
    {
      "epoch": 6.385457550092956,
      "grad_norm": 16.136798858642578,
      "learning_rate": 2.0079325731284083e-05,
      "loss": 1.4223,
      "step": 38640
    },
    {
      "epoch": 6.387110101218757,
      "grad_norm": 14.336216926574707,
      "learning_rate": 2.0070144512385466e-05,
      "loss": 1.4514,
      "step": 38650
    },
    {
      "epoch": 6.388762652344557,
      "grad_norm": 19.590391159057617,
      "learning_rate": 2.0060963293486845e-05,
      "loss": 1.5656,
      "step": 38660
    },
    {
      "epoch": 6.390415203470357,
      "grad_norm": 12.93675422668457,
      "learning_rate": 2.0051782074588225e-05,
      "loss": 1.4404,
      "step": 38670
    },
    {
      "epoch": 6.392067754596158,
      "grad_norm": 12.619584083557129,
      "learning_rate": 2.0042600855689604e-05,
      "loss": 1.4569,
      "step": 38680
    },
    {
      "epoch": 6.393720305721958,
      "grad_norm": 9.457258224487305,
      "learning_rate": 2.0033419636790983e-05,
      "loss": 1.3339,
      "step": 38690
    },
    {
      "epoch": 6.395372856847759,
      "grad_norm": 13.199435234069824,
      "learning_rate": 2.002423841789236e-05,
      "loss": 1.3885,
      "step": 38700
    },
    {
      "epoch": 6.397025407973559,
      "grad_norm": 19.85384178161621,
      "learning_rate": 2.0015057198993738e-05,
      "loss": 1.5067,
      "step": 38710
    },
    {
      "epoch": 6.398677959099359,
      "grad_norm": 9.798479080200195,
      "learning_rate": 2.0005875980095117e-05,
      "loss": 1.4391,
      "step": 38720
    },
    {
      "epoch": 6.40033051022516,
      "grad_norm": 14.563024520874023,
      "learning_rate": 1.9996694761196497e-05,
      "loss": 1.4709,
      "step": 38730
    },
    {
      "epoch": 6.40198306135096,
      "grad_norm": 13.90085506439209,
      "learning_rate": 1.9987513542297876e-05,
      "loss": 1.5339,
      "step": 38740
    },
    {
      "epoch": 6.403635612476761,
      "grad_norm": 7.516844272613525,
      "learning_rate": 1.9978332323399255e-05,
      "loss": 1.3254,
      "step": 38750
    },
    {
      "epoch": 6.405288163602561,
      "grad_norm": 10.863202095031738,
      "learning_rate": 1.9969151104500634e-05,
      "loss": 1.3872,
      "step": 38760
    },
    {
      "epoch": 6.406940714728362,
      "grad_norm": 21.591825485229492,
      "learning_rate": 1.9959969885602014e-05,
      "loss": 1.5489,
      "step": 38770
    },
    {
      "epoch": 6.4085932658541624,
      "grad_norm": 17.203176498413086,
      "learning_rate": 1.9950788666703393e-05,
      "loss": 1.4613,
      "step": 38780
    },
    {
      "epoch": 6.4102458169799625,
      "grad_norm": 13.360187530517578,
      "learning_rate": 1.9941607447804772e-05,
      "loss": 1.4316,
      "step": 38790
    },
    {
      "epoch": 6.4118983681057635,
      "grad_norm": 30.37958526611328,
      "learning_rate": 1.993242622890615e-05,
      "loss": 1.4888,
      "step": 38800
    },
    {
      "epoch": 6.413550919231564,
      "grad_norm": 21.938684463500977,
      "learning_rate": 1.992324501000753e-05,
      "loss": 1.5436,
      "step": 38810
    },
    {
      "epoch": 6.415203470357364,
      "grad_norm": 9.949905395507812,
      "learning_rate": 1.991406379110891e-05,
      "loss": 1.45,
      "step": 38820
    },
    {
      "epoch": 6.416856021483165,
      "grad_norm": 15.345038414001465,
      "learning_rate": 1.9904882572210286e-05,
      "loss": 1.4534,
      "step": 38830
    },
    {
      "epoch": 6.418508572608965,
      "grad_norm": 11.959953308105469,
      "learning_rate": 1.9895701353311665e-05,
      "loss": 1.4728,
      "step": 38840
    },
    {
      "epoch": 6.420161123734766,
      "grad_norm": 10.856972694396973,
      "learning_rate": 1.9886520134413044e-05,
      "loss": 1.4479,
      "step": 38850
    },
    {
      "epoch": 6.421813674860566,
      "grad_norm": 10.778406143188477,
      "learning_rate": 1.9877338915514424e-05,
      "loss": 1.4299,
      "step": 38860
    },
    {
      "epoch": 6.423466225986367,
      "grad_norm": 10.005019187927246,
      "learning_rate": 1.9868157696615803e-05,
      "loss": 1.3999,
      "step": 38870
    },
    {
      "epoch": 6.425118777112167,
      "grad_norm": 8.668060302734375,
      "learning_rate": 1.9858976477717185e-05,
      "loss": 1.425,
      "step": 38880
    },
    {
      "epoch": 6.426771328237967,
      "grad_norm": 13.997286796569824,
      "learning_rate": 1.984979525881856e-05,
      "loss": 1.3868,
      "step": 38890
    },
    {
      "epoch": 6.428423879363768,
      "grad_norm": 15.211793899536133,
      "learning_rate": 1.984061403991994e-05,
      "loss": 1.5317,
      "step": 38900
    },
    {
      "epoch": 6.430076430489568,
      "grad_norm": 7.7943034172058105,
      "learning_rate": 1.983143282102132e-05,
      "loss": 1.299,
      "step": 38910
    },
    {
      "epoch": 6.431728981615369,
      "grad_norm": 7.966866493225098,
      "learning_rate": 1.98222516021227e-05,
      "loss": 1.3767,
      "step": 38920
    },
    {
      "epoch": 6.433381532741169,
      "grad_norm": 29.43184471130371,
      "learning_rate": 1.9813070383224078e-05,
      "loss": 1.3973,
      "step": 38930
    },
    {
      "epoch": 6.435034083866969,
      "grad_norm": 15.053828239440918,
      "learning_rate": 1.9803889164325457e-05,
      "loss": 1.4583,
      "step": 38940
    },
    {
      "epoch": 6.43668663499277,
      "grad_norm": 6.739525318145752,
      "learning_rate": 1.9794707945426837e-05,
      "loss": 1.2457,
      "step": 38950
    },
    {
      "epoch": 6.43833918611857,
      "grad_norm": 15.069477081298828,
      "learning_rate": 1.9785526726528213e-05,
      "loss": 1.3477,
      "step": 38960
    },
    {
      "epoch": 6.439991737244371,
      "grad_norm": 14.493611335754395,
      "learning_rate": 1.9776345507629592e-05,
      "loss": 1.4433,
      "step": 38970
    },
    {
      "epoch": 6.4416442883701714,
      "grad_norm": 9.453902244567871,
      "learning_rate": 1.976716428873097e-05,
      "loss": 1.5945,
      "step": 38980
    },
    {
      "epoch": 6.4432968394959715,
      "grad_norm": 42.37503433227539,
      "learning_rate": 1.9757983069832354e-05,
      "loss": 1.4771,
      "step": 38990
    },
    {
      "epoch": 6.4449493906217725,
      "grad_norm": 13.014341354370117,
      "learning_rate": 1.9748801850933733e-05,
      "loss": 1.4682,
      "step": 39000
    },
    {
      "epoch": 6.446601941747573,
      "grad_norm": 13.779463768005371,
      "learning_rate": 1.9739620632035112e-05,
      "loss": 1.3348,
      "step": 39010
    },
    {
      "epoch": 6.448254492873374,
      "grad_norm": 27.29125213623047,
      "learning_rate": 1.9730439413136488e-05,
      "loss": 1.3392,
      "step": 39020
    },
    {
      "epoch": 6.449907043999174,
      "grad_norm": 23.10088539123535,
      "learning_rate": 1.9721258194237867e-05,
      "loss": 1.4222,
      "step": 39030
    },
    {
      "epoch": 6.451559595124974,
      "grad_norm": 9.619447708129883,
      "learning_rate": 1.9712076975339247e-05,
      "loss": 1.4682,
      "step": 39040
    },
    {
      "epoch": 6.453212146250775,
      "grad_norm": 12.745689392089844,
      "learning_rate": 1.9702895756440626e-05,
      "loss": 1.4813,
      "step": 39050
    },
    {
      "epoch": 6.454864697376575,
      "grad_norm": 13.98746109008789,
      "learning_rate": 1.9693714537542005e-05,
      "loss": 1.4522,
      "step": 39060
    },
    {
      "epoch": 6.456517248502376,
      "grad_norm": 11.557574272155762,
      "learning_rate": 1.9684533318643384e-05,
      "loss": 1.4221,
      "step": 39070
    },
    {
      "epoch": 6.458169799628176,
      "grad_norm": 10.058650016784668,
      "learning_rate": 1.9675352099744764e-05,
      "loss": 1.5567,
      "step": 39080
    },
    {
      "epoch": 6.459822350753976,
      "grad_norm": 13.266910552978516,
      "learning_rate": 1.966617088084614e-05,
      "loss": 1.5052,
      "step": 39090
    },
    {
      "epoch": 6.461474901879777,
      "grad_norm": 13.270020484924316,
      "learning_rate": 1.9656989661947522e-05,
      "loss": 1.4501,
      "step": 39100
    },
    {
      "epoch": 6.463127453005577,
      "grad_norm": 9.27397346496582,
      "learning_rate": 1.96478084430489e-05,
      "loss": 1.5234,
      "step": 39110
    },
    {
      "epoch": 6.464780004131378,
      "grad_norm": 12.057477951049805,
      "learning_rate": 1.963862722415028e-05,
      "loss": 1.3752,
      "step": 39120
    },
    {
      "epoch": 6.466432555257178,
      "grad_norm": 8.3319730758667,
      "learning_rate": 1.962944600525166e-05,
      "loss": 1.4345,
      "step": 39130
    },
    {
      "epoch": 6.468085106382979,
      "grad_norm": 9.282336235046387,
      "learning_rate": 1.962026478635304e-05,
      "loss": 1.3447,
      "step": 39140
    },
    {
      "epoch": 6.469737657508779,
      "grad_norm": 30.67858123779297,
      "learning_rate": 1.9611083567454415e-05,
      "loss": 1.4171,
      "step": 39150
    },
    {
      "epoch": 6.471390208634579,
      "grad_norm": 13.620590209960938,
      "learning_rate": 1.9601902348555794e-05,
      "loss": 1.4623,
      "step": 39160
    },
    {
      "epoch": 6.47304275976038,
      "grad_norm": 10.19704818725586,
      "learning_rate": 1.9592721129657173e-05,
      "loss": 1.4476,
      "step": 39170
    },
    {
      "epoch": 6.47469531088618,
      "grad_norm": 8.967962265014648,
      "learning_rate": 1.9583539910758553e-05,
      "loss": 1.502,
      "step": 39180
    },
    {
      "epoch": 6.4763478620119805,
      "grad_norm": 68.98345184326172,
      "learning_rate": 1.9574358691859932e-05,
      "loss": 1.4766,
      "step": 39190
    },
    {
      "epoch": 6.4780004131377815,
      "grad_norm": 13.379420280456543,
      "learning_rate": 1.956517747296131e-05,
      "loss": 1.4901,
      "step": 39200
    },
    {
      "epoch": 6.479652964263582,
      "grad_norm": 10.130301475524902,
      "learning_rate": 1.955599625406269e-05,
      "loss": 1.4408,
      "step": 39210
    },
    {
      "epoch": 6.481305515389383,
      "grad_norm": 9.031115531921387,
      "learning_rate": 1.954681503516407e-05,
      "loss": 1.4477,
      "step": 39220
    },
    {
      "epoch": 6.482958066515183,
      "grad_norm": 10.593781471252441,
      "learning_rate": 1.953763381626545e-05,
      "loss": 1.3162,
      "step": 39230
    },
    {
      "epoch": 6.484610617640984,
      "grad_norm": 10.19831657409668,
      "learning_rate": 1.9528452597366828e-05,
      "loss": 1.393,
      "step": 39240
    },
    {
      "epoch": 6.486263168766784,
      "grad_norm": 28.23524284362793,
      "learning_rate": 1.9519271378468207e-05,
      "loss": 1.4287,
      "step": 39250
    },
    {
      "epoch": 6.487915719892584,
      "grad_norm": 10.968947410583496,
      "learning_rate": 1.9510090159569587e-05,
      "loss": 1.4497,
      "step": 39260
    },
    {
      "epoch": 6.489568271018385,
      "grad_norm": 11.46053409576416,
      "learning_rate": 1.9500908940670966e-05,
      "loss": 1.511,
      "step": 39270
    },
    {
      "epoch": 6.491220822144185,
      "grad_norm": 11.82854175567627,
      "learning_rate": 1.9491727721772342e-05,
      "loss": 1.3231,
      "step": 39280
    },
    {
      "epoch": 6.492873373269986,
      "grad_norm": 14.076804161071777,
      "learning_rate": 1.948254650287372e-05,
      "loss": 1.4442,
      "step": 39290
    },
    {
      "epoch": 6.494525924395786,
      "grad_norm": 8.95998477935791,
      "learning_rate": 1.94733652839751e-05,
      "loss": 1.4034,
      "step": 39300
    },
    {
      "epoch": 6.496178475521586,
      "grad_norm": 18.003774642944336,
      "learning_rate": 1.946418406507648e-05,
      "loss": 1.5485,
      "step": 39310
    },
    {
      "epoch": 6.497831026647387,
      "grad_norm": 44.281982421875,
      "learning_rate": 1.945500284617786e-05,
      "loss": 1.4739,
      "step": 39320
    },
    {
      "epoch": 6.499483577773187,
      "grad_norm": 8.51934814453125,
      "learning_rate": 1.944582162727924e-05,
      "loss": 1.5113,
      "step": 39330
    },
    {
      "epoch": 6.501136128898988,
      "grad_norm": 12.045053482055664,
      "learning_rate": 1.9436640408380617e-05,
      "loss": 1.4032,
      "step": 39340
    },
    {
      "epoch": 6.502788680024788,
      "grad_norm": 9.922347068786621,
      "learning_rate": 1.9427459189481996e-05,
      "loss": 1.3881,
      "step": 39350
    },
    {
      "epoch": 6.504441231150588,
      "grad_norm": 17.480976104736328,
      "learning_rate": 1.9418277970583376e-05,
      "loss": 1.4731,
      "step": 39360
    },
    {
      "epoch": 6.506093782276389,
      "grad_norm": 7.395370006561279,
      "learning_rate": 1.9409096751684755e-05,
      "loss": 1.3666,
      "step": 39370
    },
    {
      "epoch": 6.507746333402189,
      "grad_norm": 23.064151763916016,
      "learning_rate": 1.9399915532786134e-05,
      "loss": 1.4598,
      "step": 39380
    },
    {
      "epoch": 6.50939888452799,
      "grad_norm": 8.329906463623047,
      "learning_rate": 1.9390734313887513e-05,
      "loss": 1.367,
      "step": 39390
    },
    {
      "epoch": 6.5110514356537905,
      "grad_norm": 8.619010925292969,
      "learning_rate": 1.9381553094988893e-05,
      "loss": 1.4067,
      "step": 39400
    },
    {
      "epoch": 6.512703986779591,
      "grad_norm": 9.927018165588379,
      "learning_rate": 1.937237187609027e-05,
      "loss": 1.4338,
      "step": 39410
    },
    {
      "epoch": 6.514356537905392,
      "grad_norm": 13.935831069946289,
      "learning_rate": 1.9363190657191648e-05,
      "loss": 1.4524,
      "step": 39420
    },
    {
      "epoch": 6.516009089031192,
      "grad_norm": 8.87132453918457,
      "learning_rate": 1.9354009438293027e-05,
      "loss": 1.3038,
      "step": 39430
    },
    {
      "epoch": 6.517661640156993,
      "grad_norm": 9.636089324951172,
      "learning_rate": 1.9344828219394406e-05,
      "loss": 1.4682,
      "step": 39440
    },
    {
      "epoch": 6.519314191282793,
      "grad_norm": 21.424053192138672,
      "learning_rate": 1.933564700049579e-05,
      "loss": 1.3425,
      "step": 39450
    },
    {
      "epoch": 6.520966742408593,
      "grad_norm": 53.614646911621094,
      "learning_rate": 1.9326465781597168e-05,
      "loss": 1.4947,
      "step": 39460
    },
    {
      "epoch": 6.522619293534394,
      "grad_norm": 8.673842430114746,
      "learning_rate": 1.9317284562698544e-05,
      "loss": 1.4588,
      "step": 39470
    },
    {
      "epoch": 6.524271844660194,
      "grad_norm": 47.06062316894531,
      "learning_rate": 1.9308103343799923e-05,
      "loss": 1.2771,
      "step": 39480
    },
    {
      "epoch": 6.525924395785995,
      "grad_norm": 11.706918716430664,
      "learning_rate": 1.9298922124901303e-05,
      "loss": 1.4553,
      "step": 39490
    },
    {
      "epoch": 6.527576946911795,
      "grad_norm": 9.595712661743164,
      "learning_rate": 1.9289740906002682e-05,
      "loss": 1.4247,
      "step": 39500
    },
    {
      "epoch": 6.529229498037596,
      "grad_norm": 9.61540412902832,
      "learning_rate": 1.928055968710406e-05,
      "loss": 1.3642,
      "step": 39510
    },
    {
      "epoch": 6.530882049163396,
      "grad_norm": 34.72698974609375,
      "learning_rate": 1.927137846820544e-05,
      "loss": 1.4828,
      "step": 39520
    },
    {
      "epoch": 6.532534600289196,
      "grad_norm": 14.31153678894043,
      "learning_rate": 1.926219724930682e-05,
      "loss": 1.5188,
      "step": 39530
    },
    {
      "epoch": 6.534187151414997,
      "grad_norm": 12.134452819824219,
      "learning_rate": 1.9253016030408195e-05,
      "loss": 1.5148,
      "step": 39540
    },
    {
      "epoch": 6.535839702540797,
      "grad_norm": 8.781564712524414,
      "learning_rate": 1.9243834811509575e-05,
      "loss": 1.5178,
      "step": 39550
    },
    {
      "epoch": 6.537492253666597,
      "grad_norm": 16.55754852294922,
      "learning_rate": 1.9234653592610957e-05,
      "loss": 1.4356,
      "step": 39560
    },
    {
      "epoch": 6.539144804792398,
      "grad_norm": 14.849857330322266,
      "learning_rate": 1.9225472373712337e-05,
      "loss": 1.405,
      "step": 39570
    },
    {
      "epoch": 6.540797355918198,
      "grad_norm": 12.2333402633667,
      "learning_rate": 1.9216291154813716e-05,
      "loss": 1.4223,
      "step": 39580
    },
    {
      "epoch": 6.542449907043999,
      "grad_norm": 8.129743576049805,
      "learning_rate": 1.9207109935915095e-05,
      "loss": 1.4793,
      "step": 39590
    },
    {
      "epoch": 6.5441024581697995,
      "grad_norm": 10.34343433380127,
      "learning_rate": 1.919792871701647e-05,
      "loss": 1.4236,
      "step": 39600
    },
    {
      "epoch": 6.5457550092956005,
      "grad_norm": 18.99115753173828,
      "learning_rate": 1.918874749811785e-05,
      "loss": 1.4893,
      "step": 39610
    },
    {
      "epoch": 6.547407560421401,
      "grad_norm": 13.560120582580566,
      "learning_rate": 1.917956627921923e-05,
      "loss": 1.5145,
      "step": 39620
    },
    {
      "epoch": 6.549060111547201,
      "grad_norm": 20.85279655456543,
      "learning_rate": 1.917038506032061e-05,
      "loss": 1.3886,
      "step": 39630
    },
    {
      "epoch": 6.550712662673002,
      "grad_norm": 11.048171997070312,
      "learning_rate": 1.9161203841421988e-05,
      "loss": 1.4902,
      "step": 39640
    },
    {
      "epoch": 6.552365213798802,
      "grad_norm": 11.270243644714355,
      "learning_rate": 1.9152022622523367e-05,
      "loss": 1.4862,
      "step": 39650
    },
    {
      "epoch": 6.554017764924602,
      "grad_norm": 10.38618278503418,
      "learning_rate": 1.9142841403624746e-05,
      "loss": 1.478,
      "step": 39660
    },
    {
      "epoch": 6.555670316050403,
      "grad_norm": 11.706856727600098,
      "learning_rate": 1.9133660184726126e-05,
      "loss": 1.4313,
      "step": 39670
    },
    {
      "epoch": 6.557322867176203,
      "grad_norm": 19.566499710083008,
      "learning_rate": 1.9124478965827505e-05,
      "loss": 1.4815,
      "step": 39680
    },
    {
      "epoch": 6.558975418302004,
      "grad_norm": 10.643657684326172,
      "learning_rate": 1.9115297746928884e-05,
      "loss": 1.2795,
      "step": 39690
    },
    {
      "epoch": 6.560627969427804,
      "grad_norm": 12.18120288848877,
      "learning_rate": 1.9106116528030263e-05,
      "loss": 1.2629,
      "step": 39700
    },
    {
      "epoch": 6.562280520553605,
      "grad_norm": 9.885913848876953,
      "learning_rate": 1.9096935309131643e-05,
      "loss": 1.4513,
      "step": 39710
    },
    {
      "epoch": 6.563933071679405,
      "grad_norm": 7.771469593048096,
      "learning_rate": 1.9087754090233022e-05,
      "loss": 1.4502,
      "step": 39720
    },
    {
      "epoch": 6.565585622805205,
      "grad_norm": 8.415877342224121,
      "learning_rate": 1.9078572871334398e-05,
      "loss": 1.5465,
      "step": 39730
    },
    {
      "epoch": 6.567238173931006,
      "grad_norm": 30.650428771972656,
      "learning_rate": 1.9069391652435777e-05,
      "loss": 1.3623,
      "step": 39740
    },
    {
      "epoch": 6.568890725056806,
      "grad_norm": 19.83210563659668,
      "learning_rate": 1.9060210433537156e-05,
      "loss": 1.3031,
      "step": 39750
    },
    {
      "epoch": 6.570543276182607,
      "grad_norm": 11.888619422912598,
      "learning_rate": 1.9051029214638535e-05,
      "loss": 1.55,
      "step": 39760
    },
    {
      "epoch": 6.572195827308407,
      "grad_norm": 8.220413208007812,
      "learning_rate": 1.9041847995739915e-05,
      "loss": 1.395,
      "step": 39770
    },
    {
      "epoch": 6.573848378434207,
      "grad_norm": 10.484262466430664,
      "learning_rate": 1.9032666776841294e-05,
      "loss": 1.3441,
      "step": 39780
    },
    {
      "epoch": 6.575500929560008,
      "grad_norm": 8.3544921875,
      "learning_rate": 1.9023485557942673e-05,
      "loss": 1.4837,
      "step": 39790
    },
    {
      "epoch": 6.5771534806858085,
      "grad_norm": 10.449460983276367,
      "learning_rate": 1.9014304339044052e-05,
      "loss": 1.4106,
      "step": 39800
    },
    {
      "epoch": 6.5788060318116095,
      "grad_norm": 9.053803443908691,
      "learning_rate": 1.9005123120145432e-05,
      "loss": 1.309,
      "step": 39810
    },
    {
      "epoch": 6.58045858293741,
      "grad_norm": 26.58787727355957,
      "learning_rate": 1.899594190124681e-05,
      "loss": 1.4609,
      "step": 39820
    },
    {
      "epoch": 6.58211113406321,
      "grad_norm": 12.057016372680664,
      "learning_rate": 1.898676068234819e-05,
      "loss": 1.3254,
      "step": 39830
    },
    {
      "epoch": 6.583763685189011,
      "grad_norm": 7.667273998260498,
      "learning_rate": 1.897757946344957e-05,
      "loss": 1.4452,
      "step": 39840
    },
    {
      "epoch": 6.585416236314811,
      "grad_norm": 12.83427619934082,
      "learning_rate": 1.896839824455095e-05,
      "loss": 1.4146,
      "step": 39850
    },
    {
      "epoch": 6.587068787440612,
      "grad_norm": 13.283857345581055,
      "learning_rate": 1.8959217025652325e-05,
      "loss": 1.4297,
      "step": 39860
    },
    {
      "epoch": 6.588721338566412,
      "grad_norm": 13.097709655761719,
      "learning_rate": 1.8950035806753704e-05,
      "loss": 1.443,
      "step": 39870
    },
    {
      "epoch": 6.590373889692213,
      "grad_norm": 10.823288917541504,
      "learning_rate": 1.8940854587855083e-05,
      "loss": 1.4325,
      "step": 39880
    },
    {
      "epoch": 6.592026440818013,
      "grad_norm": 13.02402400970459,
      "learning_rate": 1.8931673368956462e-05,
      "loss": 1.4653,
      "step": 39890
    },
    {
      "epoch": 6.593678991943813,
      "grad_norm": 11.030945777893066,
      "learning_rate": 1.8922492150057845e-05,
      "loss": 1.2672,
      "step": 39900
    },
    {
      "epoch": 6.595331543069614,
      "grad_norm": 8.03734302520752,
      "learning_rate": 1.8913310931159224e-05,
      "loss": 1.3499,
      "step": 39910
    },
    {
      "epoch": 6.596984094195414,
      "grad_norm": 12.094477653503418,
      "learning_rate": 1.89041297122606e-05,
      "loss": 1.4214,
      "step": 39920
    },
    {
      "epoch": 6.598636645321214,
      "grad_norm": 10.550955772399902,
      "learning_rate": 1.889494849336198e-05,
      "loss": 1.4045,
      "step": 39930
    },
    {
      "epoch": 6.600289196447015,
      "grad_norm": 10.761007308959961,
      "learning_rate": 1.888576727446336e-05,
      "loss": 1.5118,
      "step": 39940
    },
    {
      "epoch": 6.601941747572815,
      "grad_norm": 19.835403442382812,
      "learning_rate": 1.8876586055564738e-05,
      "loss": 1.4687,
      "step": 39950
    },
    {
      "epoch": 6.603594298698616,
      "grad_norm": 11.144922256469727,
      "learning_rate": 1.8867404836666117e-05,
      "loss": 1.4642,
      "step": 39960
    },
    {
      "epoch": 6.605246849824416,
      "grad_norm": 12.04179573059082,
      "learning_rate": 1.8858223617767496e-05,
      "loss": 1.6522,
      "step": 39970
    },
    {
      "epoch": 6.606899400950217,
      "grad_norm": 8.051116943359375,
      "learning_rate": 1.8849042398868876e-05,
      "loss": 1.4801,
      "step": 39980
    },
    {
      "epoch": 6.608551952076017,
      "grad_norm": 13.712382316589355,
      "learning_rate": 1.883986117997025e-05,
      "loss": 1.4568,
      "step": 39990
    },
    {
      "epoch": 6.6102045032018175,
      "grad_norm": 7.790043830871582,
      "learning_rate": 1.883067996107163e-05,
      "loss": 1.3776,
      "step": 40000
    },
    {
      "epoch": 6.6118570543276185,
      "grad_norm": 16.25408172607422,
      "learning_rate": 1.882149874217301e-05,
      "loss": 1.3283,
      "step": 40010
    },
    {
      "epoch": 6.613509605453419,
      "grad_norm": 6.5616774559021,
      "learning_rate": 1.8812317523274393e-05,
      "loss": 1.3177,
      "step": 40020
    },
    {
      "epoch": 6.615162156579219,
      "grad_norm": 8.824763298034668,
      "learning_rate": 1.8803136304375772e-05,
      "loss": 1.3024,
      "step": 40030
    },
    {
      "epoch": 6.61681470770502,
      "grad_norm": 14.049736976623535,
      "learning_rate": 1.879395508547715e-05,
      "loss": 1.3356,
      "step": 40040
    },
    {
      "epoch": 6.61846725883082,
      "grad_norm": 16.632484436035156,
      "learning_rate": 1.8784773866578527e-05,
      "loss": 1.5723,
      "step": 40050
    },
    {
      "epoch": 6.620119809956621,
      "grad_norm": 21.031719207763672,
      "learning_rate": 1.8775592647679906e-05,
      "loss": 1.2901,
      "step": 40060
    },
    {
      "epoch": 6.621772361082421,
      "grad_norm": 15.199872016906738,
      "learning_rate": 1.8766411428781285e-05,
      "loss": 1.4082,
      "step": 40070
    },
    {
      "epoch": 6.623424912208222,
      "grad_norm": 10.436044692993164,
      "learning_rate": 1.8757230209882665e-05,
      "loss": 1.4145,
      "step": 40080
    },
    {
      "epoch": 6.625077463334022,
      "grad_norm": 8.306492805480957,
      "learning_rate": 1.8748048990984044e-05,
      "loss": 1.3091,
      "step": 40090
    },
    {
      "epoch": 6.626730014459822,
      "grad_norm": 10.09487247467041,
      "learning_rate": 1.8738867772085423e-05,
      "loss": 1.3551,
      "step": 40100
    },
    {
      "epoch": 6.628382565585623,
      "grad_norm": 13.294191360473633,
      "learning_rate": 1.8729686553186802e-05,
      "loss": 1.6028,
      "step": 40110
    },
    {
      "epoch": 6.630035116711423,
      "grad_norm": 11.726201057434082,
      "learning_rate": 1.8720505334288178e-05,
      "loss": 1.4088,
      "step": 40120
    },
    {
      "epoch": 6.631687667837224,
      "grad_norm": 9.069192886352539,
      "learning_rate": 1.871132411538956e-05,
      "loss": 1.4661,
      "step": 40130
    },
    {
      "epoch": 6.633340218963024,
      "grad_norm": 14.839059829711914,
      "learning_rate": 1.870214289649094e-05,
      "loss": 1.5112,
      "step": 40140
    },
    {
      "epoch": 6.634992770088824,
      "grad_norm": 15.229697227478027,
      "learning_rate": 1.869296167759232e-05,
      "loss": 1.289,
      "step": 40150
    },
    {
      "epoch": 6.636645321214625,
      "grad_norm": 9.121678352355957,
      "learning_rate": 1.86837804586937e-05,
      "loss": 1.5177,
      "step": 40160
    },
    {
      "epoch": 6.638297872340425,
      "grad_norm": 9.134458541870117,
      "learning_rate": 1.8674599239795078e-05,
      "loss": 1.3569,
      "step": 40170
    },
    {
      "epoch": 6.639950423466226,
      "grad_norm": 9.136309623718262,
      "learning_rate": 1.8665418020896454e-05,
      "loss": 1.4439,
      "step": 40180
    },
    {
      "epoch": 6.641602974592026,
      "grad_norm": 13.449373245239258,
      "learning_rate": 1.8656236801997833e-05,
      "loss": 1.4349,
      "step": 40190
    },
    {
      "epoch": 6.6432555257178265,
      "grad_norm": 14.80142879486084,
      "learning_rate": 1.8647055583099212e-05,
      "loss": 1.379,
      "step": 40200
    },
    {
      "epoch": 6.6449080768436275,
      "grad_norm": 12.268270492553711,
      "learning_rate": 1.863787436420059e-05,
      "loss": 1.4978,
      "step": 40210
    },
    {
      "epoch": 6.646560627969428,
      "grad_norm": 20.85672378540039,
      "learning_rate": 1.862869314530197e-05,
      "loss": 1.3803,
      "step": 40220
    },
    {
      "epoch": 6.648213179095229,
      "grad_norm": 15.535070419311523,
      "learning_rate": 1.861951192640335e-05,
      "loss": 1.383,
      "step": 40230
    },
    {
      "epoch": 6.649865730221029,
      "grad_norm": 9.89908504486084,
      "learning_rate": 1.861033070750473e-05,
      "loss": 1.4263,
      "step": 40240
    },
    {
      "epoch": 6.65151828134683,
      "grad_norm": 15.000544548034668,
      "learning_rate": 1.860114948860611e-05,
      "loss": 1.5781,
      "step": 40250
    },
    {
      "epoch": 6.65317083247263,
      "grad_norm": 11.549114227294922,
      "learning_rate": 1.8591968269707488e-05,
      "loss": 1.4254,
      "step": 40260
    },
    {
      "epoch": 6.65482338359843,
      "grad_norm": 16.330419540405273,
      "learning_rate": 1.8582787050808867e-05,
      "loss": 1.4272,
      "step": 40270
    },
    {
      "epoch": 6.656475934724231,
      "grad_norm": 12.878795623779297,
      "learning_rate": 1.8573605831910246e-05,
      "loss": 1.4654,
      "step": 40280
    },
    {
      "epoch": 6.658128485850031,
      "grad_norm": 35.6964111328125,
      "learning_rate": 1.8564424613011625e-05,
      "loss": 1.3814,
      "step": 40290
    },
    {
      "epoch": 6.659781036975831,
      "grad_norm": 14.159383773803711,
      "learning_rate": 1.8555243394113005e-05,
      "loss": 1.5742,
      "step": 40300
    },
    {
      "epoch": 6.661433588101632,
      "grad_norm": 8.858052253723145,
      "learning_rate": 1.854606217521438e-05,
      "loss": 1.4659,
      "step": 40310
    },
    {
      "epoch": 6.663086139227432,
      "grad_norm": 9.946170806884766,
      "learning_rate": 1.853688095631576e-05,
      "loss": 1.2612,
      "step": 40320
    },
    {
      "epoch": 6.664738690353233,
      "grad_norm": 9.768632888793945,
      "learning_rate": 1.852769973741714e-05,
      "loss": 1.4816,
      "step": 40330
    },
    {
      "epoch": 6.666391241479033,
      "grad_norm": 11.27400016784668,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 1.466,
      "step": 40340
    },
    {
      "epoch": 6.668043792604834,
      "grad_norm": 13.281896591186523,
      "learning_rate": 1.8509337299619898e-05,
      "loss": 1.356,
      "step": 40350
    },
    {
      "epoch": 6.669696343730634,
      "grad_norm": 14.230932235717773,
      "learning_rate": 1.850015608072128e-05,
      "loss": 1.3526,
      "step": 40360
    },
    {
      "epoch": 6.671348894856434,
      "grad_norm": 9.921380043029785,
      "learning_rate": 1.8490974861822656e-05,
      "loss": 1.4302,
      "step": 40370
    },
    {
      "epoch": 6.673001445982235,
      "grad_norm": 13.660683631896973,
      "learning_rate": 1.8481793642924035e-05,
      "loss": 1.3974,
      "step": 40380
    },
    {
      "epoch": 6.674653997108035,
      "grad_norm": 15.629735946655273,
      "learning_rate": 1.8472612424025415e-05,
      "loss": 1.4577,
      "step": 40390
    },
    {
      "epoch": 6.6763065482338355,
      "grad_norm": 10.573319435119629,
      "learning_rate": 1.8463431205126794e-05,
      "loss": 1.425,
      "step": 40400
    },
    {
      "epoch": 6.6779590993596365,
      "grad_norm": 9.454117774963379,
      "learning_rate": 1.8454249986228173e-05,
      "loss": 1.374,
      "step": 40410
    },
    {
      "epoch": 6.679611650485437,
      "grad_norm": 21.72432518005371,
      "learning_rate": 1.8445068767329552e-05,
      "loss": 1.499,
      "step": 40420
    },
    {
      "epoch": 6.681264201611238,
      "grad_norm": 10.876113891601562,
      "learning_rate": 1.843588754843093e-05,
      "loss": 1.5718,
      "step": 40430
    },
    {
      "epoch": 6.682916752737038,
      "grad_norm": 21.28350067138672,
      "learning_rate": 1.8426706329532307e-05,
      "loss": 1.399,
      "step": 40440
    },
    {
      "epoch": 6.684569303862839,
      "grad_norm": 10.992979049682617,
      "learning_rate": 1.8417525110633687e-05,
      "loss": 1.3749,
      "step": 40450
    },
    {
      "epoch": 6.686221854988639,
      "grad_norm": 11.09716510772705,
      "learning_rate": 1.8408343891735066e-05,
      "loss": 1.3887,
      "step": 40460
    },
    {
      "epoch": 6.687874406114439,
      "grad_norm": 14.070152282714844,
      "learning_rate": 1.839916267283645e-05,
      "loss": 1.541,
      "step": 40470
    },
    {
      "epoch": 6.68952695724024,
      "grad_norm": 16.888887405395508,
      "learning_rate": 1.8389981453937828e-05,
      "loss": 1.4876,
      "step": 40480
    },
    {
      "epoch": 6.69117950836604,
      "grad_norm": 13.392816543579102,
      "learning_rate": 1.8380800235039207e-05,
      "loss": 1.4925,
      "step": 40490
    },
    {
      "epoch": 6.692832059491841,
      "grad_norm": 12.64699649810791,
      "learning_rate": 1.8371619016140583e-05,
      "loss": 1.4564,
      "step": 40500
    },
    {
      "epoch": 6.694484610617641,
      "grad_norm": 48.2119026184082,
      "learning_rate": 1.8362437797241962e-05,
      "loss": 1.4863,
      "step": 40510
    },
    {
      "epoch": 6.696137161743441,
      "grad_norm": 10.135701179504395,
      "learning_rate": 1.835325657834334e-05,
      "loss": 1.4801,
      "step": 40520
    },
    {
      "epoch": 6.697789712869242,
      "grad_norm": 13.071991920471191,
      "learning_rate": 1.834407535944472e-05,
      "loss": 1.5195,
      "step": 40530
    },
    {
      "epoch": 6.699442263995042,
      "grad_norm": 10.40569019317627,
      "learning_rate": 1.83348941405461e-05,
      "loss": 1.4392,
      "step": 40540
    },
    {
      "epoch": 6.701094815120843,
      "grad_norm": 8.229533195495605,
      "learning_rate": 1.832571292164748e-05,
      "loss": 1.4115,
      "step": 40550
    },
    {
      "epoch": 6.702747366246643,
      "grad_norm": 10.888386726379395,
      "learning_rate": 1.831653170274886e-05,
      "loss": 1.3083,
      "step": 40560
    },
    {
      "epoch": 6.704399917372443,
      "grad_norm": 10.805375099182129,
      "learning_rate": 1.8307350483850234e-05,
      "loss": 1.4312,
      "step": 40570
    },
    {
      "epoch": 6.706052468498244,
      "grad_norm": 21.973468780517578,
      "learning_rate": 1.8298169264951613e-05,
      "loss": 1.4178,
      "step": 40580
    },
    {
      "epoch": 6.707705019624044,
      "grad_norm": 12.331151008605957,
      "learning_rate": 1.8288988046052996e-05,
      "loss": 1.361,
      "step": 40590
    },
    {
      "epoch": 6.709357570749845,
      "grad_norm": 17.735149383544922,
      "learning_rate": 1.8279806827154375e-05,
      "loss": 1.5563,
      "step": 40600
    },
    {
      "epoch": 6.7110101218756455,
      "grad_norm": 9.35239028930664,
      "learning_rate": 1.8270625608255755e-05,
      "loss": 1.2544,
      "step": 40610
    },
    {
      "epoch": 6.7126626730014465,
      "grad_norm": 10.340348243713379,
      "learning_rate": 1.8261444389357134e-05,
      "loss": 1.5049,
      "step": 40620
    },
    {
      "epoch": 6.714315224127247,
      "grad_norm": 7.2036871910095215,
      "learning_rate": 1.825226317045851e-05,
      "loss": 1.3728,
      "step": 40630
    },
    {
      "epoch": 6.715967775253047,
      "grad_norm": 21.527620315551758,
      "learning_rate": 1.824308195155989e-05,
      "loss": 1.4878,
      "step": 40640
    },
    {
      "epoch": 6.717620326378848,
      "grad_norm": 16.127460479736328,
      "learning_rate": 1.8233900732661268e-05,
      "loss": 1.4064,
      "step": 40650
    },
    {
      "epoch": 6.719272877504648,
      "grad_norm": 27.192962646484375,
      "learning_rate": 1.8224719513762647e-05,
      "loss": 1.3923,
      "step": 40660
    },
    {
      "epoch": 6.720925428630448,
      "grad_norm": 13.89322280883789,
      "learning_rate": 1.8215538294864027e-05,
      "loss": 1.4246,
      "step": 40670
    },
    {
      "epoch": 6.722577979756249,
      "grad_norm": 12.22103214263916,
      "learning_rate": 1.8206357075965406e-05,
      "loss": 1.3845,
      "step": 40680
    },
    {
      "epoch": 6.724230530882049,
      "grad_norm": 10.441896438598633,
      "learning_rate": 1.8197175857066785e-05,
      "loss": 1.6125,
      "step": 40690
    },
    {
      "epoch": 6.72588308200785,
      "grad_norm": 10.106922149658203,
      "learning_rate": 1.8187994638168164e-05,
      "loss": 1.3981,
      "step": 40700
    },
    {
      "epoch": 6.72753563313365,
      "grad_norm": 10.453673362731934,
      "learning_rate": 1.8178813419269544e-05,
      "loss": 1.4177,
      "step": 40710
    },
    {
      "epoch": 6.729188184259451,
      "grad_norm": 14.970731735229492,
      "learning_rate": 1.8169632200370923e-05,
      "loss": 1.4991,
      "step": 40720
    },
    {
      "epoch": 6.730840735385251,
      "grad_norm": 8.335892677307129,
      "learning_rate": 1.8160450981472302e-05,
      "loss": 1.354,
      "step": 40730
    },
    {
      "epoch": 6.732493286511051,
      "grad_norm": 15.514796257019043,
      "learning_rate": 1.815126976257368e-05,
      "loss": 1.4697,
      "step": 40740
    },
    {
      "epoch": 6.734145837636852,
      "grad_norm": 11.07535457611084,
      "learning_rate": 1.814208854367506e-05,
      "loss": 1.4608,
      "step": 40750
    },
    {
      "epoch": 6.735798388762652,
      "grad_norm": 14.842573165893555,
      "learning_rate": 1.8132907324776437e-05,
      "loss": 1.4753,
      "step": 40760
    },
    {
      "epoch": 6.737450939888452,
      "grad_norm": 9.721745491027832,
      "learning_rate": 1.8123726105877816e-05,
      "loss": 1.5479,
      "step": 40770
    },
    {
      "epoch": 6.739103491014253,
      "grad_norm": 8.41561222076416,
      "learning_rate": 1.8114544886979195e-05,
      "loss": 1.3348,
      "step": 40780
    },
    {
      "epoch": 6.740756042140053,
      "grad_norm": 9.427742004394531,
      "learning_rate": 1.8105363668080574e-05,
      "loss": 1.4775,
      "step": 40790
    },
    {
      "epoch": 6.742408593265854,
      "grad_norm": 11.878392219543457,
      "learning_rate": 1.8096182449181954e-05,
      "loss": 1.492,
      "step": 40800
    },
    {
      "epoch": 6.7440611443916545,
      "grad_norm": 14.520692825317383,
      "learning_rate": 1.8087001230283333e-05,
      "loss": 1.5121,
      "step": 40810
    },
    {
      "epoch": 6.7457136955174555,
      "grad_norm": 5.713458061218262,
      "learning_rate": 1.8077820011384712e-05,
      "loss": 1.4465,
      "step": 40820
    },
    {
      "epoch": 6.747366246643256,
      "grad_norm": 11.001376152038574,
      "learning_rate": 1.806863879248609e-05,
      "loss": 1.3831,
      "step": 40830
    },
    {
      "epoch": 6.749018797769056,
      "grad_norm": 11.784494400024414,
      "learning_rate": 1.805945757358747e-05,
      "loss": 1.4752,
      "step": 40840
    },
    {
      "epoch": 6.750671348894857,
      "grad_norm": 16.32159996032715,
      "learning_rate": 1.805027635468885e-05,
      "loss": 1.3969,
      "step": 40850
    },
    {
      "epoch": 6.752323900020657,
      "grad_norm": 8.51651382446289,
      "learning_rate": 1.804109513579023e-05,
      "loss": 1.4283,
      "step": 40860
    },
    {
      "epoch": 6.753976451146458,
      "grad_norm": 17.39286231994629,
      "learning_rate": 1.8031913916891608e-05,
      "loss": 1.482,
      "step": 40870
    },
    {
      "epoch": 6.755629002272258,
      "grad_norm": 19.771865844726562,
      "learning_rate": 1.8022732697992987e-05,
      "loss": 1.3334,
      "step": 40880
    },
    {
      "epoch": 6.757281553398058,
      "grad_norm": 8.667237281799316,
      "learning_rate": 1.8013551479094363e-05,
      "loss": 1.3109,
      "step": 40890
    },
    {
      "epoch": 6.758934104523859,
      "grad_norm": 16.53960609436035,
      "learning_rate": 1.8004370260195743e-05,
      "loss": 1.4221,
      "step": 40900
    },
    {
      "epoch": 6.760586655649659,
      "grad_norm": 18.91925048828125,
      "learning_rate": 1.7995189041297122e-05,
      "loss": 1.376,
      "step": 40910
    },
    {
      "epoch": 6.76223920677546,
      "grad_norm": 14.152482986450195,
      "learning_rate": 1.79860078223985e-05,
      "loss": 1.4648,
      "step": 40920
    },
    {
      "epoch": 6.76389175790126,
      "grad_norm": 10.383070945739746,
      "learning_rate": 1.7976826603499884e-05,
      "loss": 1.3391,
      "step": 40930
    },
    {
      "epoch": 6.76554430902706,
      "grad_norm": 15.106592178344727,
      "learning_rate": 1.7967645384601263e-05,
      "loss": 1.4528,
      "step": 40940
    },
    {
      "epoch": 6.767196860152861,
      "grad_norm": 13.787833213806152,
      "learning_rate": 1.795846416570264e-05,
      "loss": 1.3813,
      "step": 40950
    },
    {
      "epoch": 6.768849411278661,
      "grad_norm": 6.263724327087402,
      "learning_rate": 1.7949282946804018e-05,
      "loss": 1.5428,
      "step": 40960
    },
    {
      "epoch": 6.770501962404462,
      "grad_norm": 10.366127014160156,
      "learning_rate": 1.7940101727905397e-05,
      "loss": 1.4737,
      "step": 40970
    },
    {
      "epoch": 6.772154513530262,
      "grad_norm": 27.11163902282715,
      "learning_rate": 1.7930920509006777e-05,
      "loss": 1.4403,
      "step": 40980
    },
    {
      "epoch": 6.773807064656062,
      "grad_norm": 10.552127838134766,
      "learning_rate": 1.7921739290108156e-05,
      "loss": 1.4511,
      "step": 40990
    },
    {
      "epoch": 6.775459615781863,
      "grad_norm": 17.98050308227539,
      "learning_rate": 1.7912558071209535e-05,
      "loss": 1.5128,
      "step": 41000
    },
    {
      "epoch": 6.7771121669076635,
      "grad_norm": 9.192895889282227,
      "learning_rate": 1.7903376852310914e-05,
      "loss": 1.334,
      "step": 41010
    },
    {
      "epoch": 6.7787647180334645,
      "grad_norm": 11.638321876525879,
      "learning_rate": 1.789419563341229e-05,
      "loss": 1.3764,
      "step": 41020
    },
    {
      "epoch": 6.780417269159265,
      "grad_norm": 12.00686264038086,
      "learning_rate": 1.788501441451367e-05,
      "loss": 1.4813,
      "step": 41030
    },
    {
      "epoch": 6.782069820285065,
      "grad_norm": 11.396140098571777,
      "learning_rate": 1.7875833195615052e-05,
      "loss": 1.438,
      "step": 41040
    },
    {
      "epoch": 6.783722371410866,
      "grad_norm": 55.01451873779297,
      "learning_rate": 1.786665197671643e-05,
      "loss": 1.4997,
      "step": 41050
    },
    {
      "epoch": 6.785374922536666,
      "grad_norm": 9.244829177856445,
      "learning_rate": 1.785747075781781e-05,
      "loss": 1.3627,
      "step": 41060
    },
    {
      "epoch": 6.787027473662467,
      "grad_norm": 9.327887535095215,
      "learning_rate": 1.784828953891919e-05,
      "loss": 1.3883,
      "step": 41070
    },
    {
      "epoch": 6.788680024788267,
      "grad_norm": 9.212366104125977,
      "learning_rate": 1.7839108320020566e-05,
      "loss": 1.4265,
      "step": 41080
    },
    {
      "epoch": 6.790332575914068,
      "grad_norm": 41.623443603515625,
      "learning_rate": 1.7829927101121945e-05,
      "loss": 1.4342,
      "step": 41090
    },
    {
      "epoch": 6.791985127039868,
      "grad_norm": 10.537240982055664,
      "learning_rate": 1.7820745882223324e-05,
      "loss": 1.5631,
      "step": 41100
    },
    {
      "epoch": 6.793637678165668,
      "grad_norm": 14.747405052185059,
      "learning_rate": 1.7811564663324703e-05,
      "loss": 1.4803,
      "step": 41110
    },
    {
      "epoch": 6.795290229291469,
      "grad_norm": 10.192240715026855,
      "learning_rate": 1.7802383444426083e-05,
      "loss": 1.3918,
      "step": 41120
    },
    {
      "epoch": 6.796942780417269,
      "grad_norm": 9.440613746643066,
      "learning_rate": 1.7793202225527462e-05,
      "loss": 1.4808,
      "step": 41130
    },
    {
      "epoch": 6.798595331543069,
      "grad_norm": 14.262847900390625,
      "learning_rate": 1.778402100662884e-05,
      "loss": 1.3783,
      "step": 41140
    },
    {
      "epoch": 6.80024788266887,
      "grad_norm": 10.99431324005127,
      "learning_rate": 1.7774839787730217e-05,
      "loss": 1.4737,
      "step": 41150
    },
    {
      "epoch": 6.80190043379467,
      "grad_norm": 8.806312561035156,
      "learning_rate": 1.77656585688316e-05,
      "loss": 1.3779,
      "step": 41160
    },
    {
      "epoch": 6.803552984920471,
      "grad_norm": 39.46382141113281,
      "learning_rate": 1.775647734993298e-05,
      "loss": 1.4395,
      "step": 41170
    },
    {
      "epoch": 6.805205536046271,
      "grad_norm": 14.97314453125,
      "learning_rate": 1.7747296131034358e-05,
      "loss": 1.4276,
      "step": 41180
    },
    {
      "epoch": 6.806858087172072,
      "grad_norm": 9.130488395690918,
      "learning_rate": 1.7738114912135737e-05,
      "loss": 1.3061,
      "step": 41190
    },
    {
      "epoch": 6.808510638297872,
      "grad_norm": 9.451000213623047,
      "learning_rate": 1.7728933693237117e-05,
      "loss": 1.5621,
      "step": 41200
    },
    {
      "epoch": 6.8101631894236725,
      "grad_norm": 10.108766555786133,
      "learning_rate": 1.7719752474338492e-05,
      "loss": 1.387,
      "step": 41210
    },
    {
      "epoch": 6.8118157405494735,
      "grad_norm": 7.01558780670166,
      "learning_rate": 1.7710571255439872e-05,
      "loss": 1.3172,
      "step": 41220
    },
    {
      "epoch": 6.813468291675274,
      "grad_norm": 10.980785369873047,
      "learning_rate": 1.770139003654125e-05,
      "loss": 1.4921,
      "step": 41230
    },
    {
      "epoch": 6.815120842801074,
      "grad_norm": 12.635100364685059,
      "learning_rate": 1.769220881764263e-05,
      "loss": 1.5118,
      "step": 41240
    },
    {
      "epoch": 6.816773393926875,
      "grad_norm": 21.64072036743164,
      "learning_rate": 1.768302759874401e-05,
      "loss": 1.3484,
      "step": 41250
    },
    {
      "epoch": 6.818425945052675,
      "grad_norm": 10.687591552734375,
      "learning_rate": 1.767384637984539e-05,
      "loss": 1.3822,
      "step": 41260
    },
    {
      "epoch": 6.820078496178476,
      "grad_norm": 31.336519241333008,
      "learning_rate": 1.7664665160946768e-05,
      "loss": 1.4516,
      "step": 41270
    },
    {
      "epoch": 6.821731047304276,
      "grad_norm": 14.67437744140625,
      "learning_rate": 1.7655483942048147e-05,
      "loss": 1.4765,
      "step": 41280
    },
    {
      "epoch": 6.823383598430077,
      "grad_norm": 18.006410598754883,
      "learning_rate": 1.7646302723149526e-05,
      "loss": 1.3861,
      "step": 41290
    },
    {
      "epoch": 6.825036149555877,
      "grad_norm": 10.844664573669434,
      "learning_rate": 1.7637121504250906e-05,
      "loss": 1.4537,
      "step": 41300
    },
    {
      "epoch": 6.826688700681677,
      "grad_norm": 11.941879272460938,
      "learning_rate": 1.7627940285352285e-05,
      "loss": 1.4432,
      "step": 41310
    },
    {
      "epoch": 6.828341251807478,
      "grad_norm": 16.71599769592285,
      "learning_rate": 1.7618759066453664e-05,
      "loss": 1.4506,
      "step": 41320
    },
    {
      "epoch": 6.829993802933278,
      "grad_norm": 11.480480194091797,
      "learning_rate": 1.7609577847555043e-05,
      "loss": 1.5141,
      "step": 41330
    },
    {
      "epoch": 6.831646354059079,
      "grad_norm": 13.7313814163208,
      "learning_rate": 1.760039662865642e-05,
      "loss": 1.3246,
      "step": 41340
    },
    {
      "epoch": 6.833298905184879,
      "grad_norm": 6.67985725402832,
      "learning_rate": 1.75912154097578e-05,
      "loss": 1.5389,
      "step": 41350
    },
    {
      "epoch": 6.834951456310679,
      "grad_norm": 35.11752700805664,
      "learning_rate": 1.7582034190859178e-05,
      "loss": 1.3679,
      "step": 41360
    },
    {
      "epoch": 6.83660400743648,
      "grad_norm": 16.82154083251953,
      "learning_rate": 1.7572852971960557e-05,
      "loss": 1.5087,
      "step": 41370
    },
    {
      "epoch": 6.83825655856228,
      "grad_norm": 11.635119438171387,
      "learning_rate": 1.7563671753061936e-05,
      "loss": 1.5188,
      "step": 41380
    },
    {
      "epoch": 6.839909109688081,
      "grad_norm": 9.211424827575684,
      "learning_rate": 1.755449053416332e-05,
      "loss": 1.3765,
      "step": 41390
    },
    {
      "epoch": 6.841561660813881,
      "grad_norm": 7.994271755218506,
      "learning_rate": 1.7545309315264695e-05,
      "loss": 1.4092,
      "step": 41400
    },
    {
      "epoch": 6.8432142119396815,
      "grad_norm": 9.016519546508789,
      "learning_rate": 1.7536128096366074e-05,
      "loss": 1.5425,
      "step": 41410
    },
    {
      "epoch": 6.8448667630654825,
      "grad_norm": 15.688495635986328,
      "learning_rate": 1.7526946877467453e-05,
      "loss": 1.4526,
      "step": 41420
    },
    {
      "epoch": 6.846519314191283,
      "grad_norm": 11.424908638000488,
      "learning_rate": 1.7517765658568833e-05,
      "loss": 1.4075,
      "step": 41430
    },
    {
      "epoch": 6.8481718653170836,
      "grad_norm": 11.070535659790039,
      "learning_rate": 1.7508584439670212e-05,
      "loss": 1.3919,
      "step": 41440
    },
    {
      "epoch": 6.849824416442884,
      "grad_norm": 7.652846336364746,
      "learning_rate": 1.749940322077159e-05,
      "loss": 1.328,
      "step": 41450
    },
    {
      "epoch": 6.851476967568685,
      "grad_norm": 9.631972312927246,
      "learning_rate": 1.749022200187297e-05,
      "loss": 1.4173,
      "step": 41460
    },
    {
      "epoch": 6.853129518694485,
      "grad_norm": 41.709320068359375,
      "learning_rate": 1.7481040782974346e-05,
      "loss": 1.5274,
      "step": 41470
    },
    {
      "epoch": 6.854782069820285,
      "grad_norm": 11.905040740966797,
      "learning_rate": 1.7471859564075725e-05,
      "loss": 1.4682,
      "step": 41480
    },
    {
      "epoch": 6.856434620946086,
      "grad_norm": 20.44754981994629,
      "learning_rate": 1.7462678345177105e-05,
      "loss": 1.553,
      "step": 41490
    },
    {
      "epoch": 6.858087172071886,
      "grad_norm": 8.729151725769043,
      "learning_rate": 1.7453497126278487e-05,
      "loss": 1.4357,
      "step": 41500
    },
    {
      "epoch": 6.859739723197686,
      "grad_norm": 13.73019027709961,
      "learning_rate": 1.7444315907379867e-05,
      "loss": 1.4612,
      "step": 41510
    },
    {
      "epoch": 6.861392274323487,
      "grad_norm": 9.946063995361328,
      "learning_rate": 1.7435134688481246e-05,
      "loss": 1.4276,
      "step": 41520
    },
    {
      "epoch": 6.863044825449287,
      "grad_norm": 17.463390350341797,
      "learning_rate": 1.742595346958262e-05,
      "loss": 1.2973,
      "step": 41530
    },
    {
      "epoch": 6.864697376575088,
      "grad_norm": 25.70206642150879,
      "learning_rate": 1.7416772250684e-05,
      "loss": 1.5196,
      "step": 41540
    },
    {
      "epoch": 6.866349927700888,
      "grad_norm": 9.276856422424316,
      "learning_rate": 1.740759103178538e-05,
      "loss": 1.5073,
      "step": 41550
    },
    {
      "epoch": 6.868002478826689,
      "grad_norm": 9.336898803710938,
      "learning_rate": 1.739840981288676e-05,
      "loss": 1.4767,
      "step": 41560
    },
    {
      "epoch": 6.869655029952489,
      "grad_norm": 10.633883476257324,
      "learning_rate": 1.738922859398814e-05,
      "loss": 1.532,
      "step": 41570
    },
    {
      "epoch": 6.871307581078289,
      "grad_norm": 9.510002136230469,
      "learning_rate": 1.7380047375089518e-05,
      "loss": 1.4982,
      "step": 41580
    },
    {
      "epoch": 6.87296013220409,
      "grad_norm": 9.511191368103027,
      "learning_rate": 1.7370866156190897e-05,
      "loss": 1.4293,
      "step": 41590
    },
    {
      "epoch": 6.87461268332989,
      "grad_norm": 12.487457275390625,
      "learning_rate": 1.7361684937292273e-05,
      "loss": 1.5126,
      "step": 41600
    },
    {
      "epoch": 6.8762652344556905,
      "grad_norm": 11.388182640075684,
      "learning_rate": 1.7352503718393652e-05,
      "loss": 1.387,
      "step": 41610
    },
    {
      "epoch": 6.8779177855814915,
      "grad_norm": 12.775744438171387,
      "learning_rate": 1.7343322499495035e-05,
      "loss": 1.5719,
      "step": 41620
    },
    {
      "epoch": 6.879570336707292,
      "grad_norm": 9.344951629638672,
      "learning_rate": 1.7334141280596414e-05,
      "loss": 1.4063,
      "step": 41630
    },
    {
      "epoch": 6.8812228878330925,
      "grad_norm": 13.491334915161133,
      "learning_rate": 1.7324960061697793e-05,
      "loss": 1.2582,
      "step": 41640
    },
    {
      "epoch": 6.882875438958893,
      "grad_norm": 78.28813934326172,
      "learning_rate": 1.7315778842799173e-05,
      "loss": 1.5862,
      "step": 41650
    },
    {
      "epoch": 6.884527990084694,
      "grad_norm": 9.13769245147705,
      "learning_rate": 1.730659762390055e-05,
      "loss": 1.4646,
      "step": 41660
    },
    {
      "epoch": 6.886180541210494,
      "grad_norm": 7.546828269958496,
      "learning_rate": 1.7297416405001928e-05,
      "loss": 1.4563,
      "step": 41670
    },
    {
      "epoch": 6.887833092336294,
      "grad_norm": 8.716988563537598,
      "learning_rate": 1.7288235186103307e-05,
      "loss": 1.4902,
      "step": 41680
    },
    {
      "epoch": 6.889485643462095,
      "grad_norm": 11.223580360412598,
      "learning_rate": 1.7279053967204686e-05,
      "loss": 1.3886,
      "step": 41690
    },
    {
      "epoch": 6.891138194587895,
      "grad_norm": 9.644621849060059,
      "learning_rate": 1.7269872748306065e-05,
      "loss": 1.398,
      "step": 41700
    },
    {
      "epoch": 6.892790745713696,
      "grad_norm": 8.582539558410645,
      "learning_rate": 1.7260691529407445e-05,
      "loss": 1.3557,
      "step": 41710
    },
    {
      "epoch": 6.894443296839496,
      "grad_norm": 31.303890228271484,
      "learning_rate": 1.7251510310508824e-05,
      "loss": 1.4995,
      "step": 41720
    },
    {
      "epoch": 6.896095847965296,
      "grad_norm": 9.271944046020508,
      "learning_rate": 1.7242329091610203e-05,
      "loss": 1.2297,
      "step": 41730
    },
    {
      "epoch": 6.897748399091097,
      "grad_norm": 12.003678321838379,
      "learning_rate": 1.7233147872711582e-05,
      "loss": 1.3033,
      "step": 41740
    },
    {
      "epoch": 6.899400950216897,
      "grad_norm": 12.872206687927246,
      "learning_rate": 1.7223966653812962e-05,
      "loss": 1.5084,
      "step": 41750
    },
    {
      "epoch": 6.901053501342698,
      "grad_norm": 12.925690650939941,
      "learning_rate": 1.721478543491434e-05,
      "loss": 1.476,
      "step": 41760
    },
    {
      "epoch": 6.902706052468498,
      "grad_norm": 16.326433181762695,
      "learning_rate": 1.720560421601572e-05,
      "loss": 1.4717,
      "step": 41770
    },
    {
      "epoch": 6.904358603594298,
      "grad_norm": 10.100279808044434,
      "learning_rate": 1.71964229971171e-05,
      "loss": 1.5529,
      "step": 41780
    },
    {
      "epoch": 6.906011154720099,
      "grad_norm": 13.078716278076172,
      "learning_rate": 1.7187241778218475e-05,
      "loss": 1.4674,
      "step": 41790
    },
    {
      "epoch": 6.907663705845899,
      "grad_norm": 11.755579948425293,
      "learning_rate": 1.7178060559319855e-05,
      "loss": 1.3926,
      "step": 41800
    },
    {
      "epoch": 6.9093162569717,
      "grad_norm": 11.553741455078125,
      "learning_rate": 1.7168879340421234e-05,
      "loss": 1.352,
      "step": 41810
    },
    {
      "epoch": 6.9109688080975005,
      "grad_norm": 12.865326881408691,
      "learning_rate": 1.7159698121522613e-05,
      "loss": 1.3908,
      "step": 41820
    },
    {
      "epoch": 6.9126213592233015,
      "grad_norm": 9.914249420166016,
      "learning_rate": 1.7150516902623992e-05,
      "loss": 1.3497,
      "step": 41830
    },
    {
      "epoch": 6.9142739103491015,
      "grad_norm": 6.659285068511963,
      "learning_rate": 1.7141335683725375e-05,
      "loss": 1.3643,
      "step": 41840
    },
    {
      "epoch": 6.915926461474902,
      "grad_norm": 9.710965156555176,
      "learning_rate": 1.713215446482675e-05,
      "loss": 1.4596,
      "step": 41850
    },
    {
      "epoch": 6.917579012600703,
      "grad_norm": 9.427946090698242,
      "learning_rate": 1.712297324592813e-05,
      "loss": 1.4356,
      "step": 41860
    },
    {
      "epoch": 6.919231563726503,
      "grad_norm": 10.323373794555664,
      "learning_rate": 1.711379202702951e-05,
      "loss": 1.4465,
      "step": 41870
    },
    {
      "epoch": 6.920884114852303,
      "grad_norm": 19.84877586364746,
      "learning_rate": 1.710461080813089e-05,
      "loss": 1.4025,
      "step": 41880
    },
    {
      "epoch": 6.922536665978104,
      "grad_norm": 9.801685333251953,
      "learning_rate": 1.7095429589232268e-05,
      "loss": 1.3275,
      "step": 41890
    },
    {
      "epoch": 6.924189217103904,
      "grad_norm": 43.13996887207031,
      "learning_rate": 1.7086248370333647e-05,
      "loss": 1.4675,
      "step": 41900
    },
    {
      "epoch": 6.925841768229705,
      "grad_norm": 37.639007568359375,
      "learning_rate": 1.7077067151435026e-05,
      "loss": 1.2324,
      "step": 41910
    },
    {
      "epoch": 6.927494319355505,
      "grad_norm": 12.643664360046387,
      "learning_rate": 1.7067885932536402e-05,
      "loss": 1.3959,
      "step": 41920
    },
    {
      "epoch": 6.929146870481306,
      "grad_norm": 30.085914611816406,
      "learning_rate": 1.705870471363778e-05,
      "loss": 1.3784,
      "step": 41930
    },
    {
      "epoch": 6.930799421607106,
      "grad_norm": 17.809532165527344,
      "learning_rate": 1.704952349473916e-05,
      "loss": 1.4568,
      "step": 41940
    },
    {
      "epoch": 6.932451972732906,
      "grad_norm": 9.029561042785645,
      "learning_rate": 1.704034227584054e-05,
      "loss": 1.3388,
      "step": 41950
    },
    {
      "epoch": 6.934104523858707,
      "grad_norm": 12.235958099365234,
      "learning_rate": 1.7031161056941923e-05,
      "loss": 1.4447,
      "step": 41960
    },
    {
      "epoch": 6.935757074984507,
      "grad_norm": 12.634254455566406,
      "learning_rate": 1.7021979838043302e-05,
      "loss": 1.3982,
      "step": 41970
    },
    {
      "epoch": 6.937409626110307,
      "grad_norm": 15.774003982543945,
      "learning_rate": 1.7012798619144678e-05,
      "loss": 1.3463,
      "step": 41980
    },
    {
      "epoch": 6.939062177236108,
      "grad_norm": 8.786571502685547,
      "learning_rate": 1.7003617400246057e-05,
      "loss": 1.4888,
      "step": 41990
    },
    {
      "epoch": 6.940714728361908,
      "grad_norm": 20.433345794677734,
      "learning_rate": 1.6994436181347436e-05,
      "loss": 1.3355,
      "step": 42000
    },
    {
      "epoch": 6.942367279487709,
      "grad_norm": 7.788454055786133,
      "learning_rate": 1.6985254962448815e-05,
      "loss": 1.3896,
      "step": 42010
    },
    {
      "epoch": 6.9440198306135095,
      "grad_norm": 7.458805084228516,
      "learning_rate": 1.6976073743550195e-05,
      "loss": 1.393,
      "step": 42020
    },
    {
      "epoch": 6.9456723817393105,
      "grad_norm": 14.492402076721191,
      "learning_rate": 1.6966892524651574e-05,
      "loss": 1.5976,
      "step": 42030
    },
    {
      "epoch": 6.9473249328651105,
      "grad_norm": 8.02368450164795,
      "learning_rate": 1.6957711305752953e-05,
      "loss": 1.3769,
      "step": 42040
    },
    {
      "epoch": 6.948977483990911,
      "grad_norm": 8.434112548828125,
      "learning_rate": 1.694853008685433e-05,
      "loss": 1.4697,
      "step": 42050
    },
    {
      "epoch": 6.950630035116712,
      "grad_norm": 10.042356491088867,
      "learning_rate": 1.6939348867955708e-05,
      "loss": 1.4125,
      "step": 42060
    },
    {
      "epoch": 6.952282586242512,
      "grad_norm": 77.89820098876953,
      "learning_rate": 1.693016764905709e-05,
      "loss": 1.4607,
      "step": 42070
    },
    {
      "epoch": 6.953935137368313,
      "grad_norm": 48.598690032958984,
      "learning_rate": 1.692098643015847e-05,
      "loss": 1.3837,
      "step": 42080
    },
    {
      "epoch": 6.955587688494113,
      "grad_norm": 47.6584587097168,
      "learning_rate": 1.691180521125985e-05,
      "loss": 1.6172,
      "step": 42090
    },
    {
      "epoch": 6.957240239619913,
      "grad_norm": 7.663386344909668,
      "learning_rate": 1.690262399236123e-05,
      "loss": 1.383,
      "step": 42100
    },
    {
      "epoch": 6.958892790745714,
      "grad_norm": 7.526758670806885,
      "learning_rate": 1.6893442773462604e-05,
      "loss": 1.4312,
      "step": 42110
    },
    {
      "epoch": 6.960545341871514,
      "grad_norm": 16.803466796875,
      "learning_rate": 1.6884261554563984e-05,
      "loss": 1.4376,
      "step": 42120
    },
    {
      "epoch": 6.962197892997315,
      "grad_norm": 13.658576011657715,
      "learning_rate": 1.6875080335665363e-05,
      "loss": 1.5319,
      "step": 42130
    },
    {
      "epoch": 6.963850444123115,
      "grad_norm": 8.935698509216309,
      "learning_rate": 1.6865899116766742e-05,
      "loss": 1.439,
      "step": 42140
    },
    {
      "epoch": 6.965502995248915,
      "grad_norm": 11.672993659973145,
      "learning_rate": 1.685671789786812e-05,
      "loss": 1.4362,
      "step": 42150
    },
    {
      "epoch": 6.967155546374716,
      "grad_norm": 13.540828704833984,
      "learning_rate": 1.68475366789695e-05,
      "loss": 1.4189,
      "step": 42160
    },
    {
      "epoch": 6.968808097500516,
      "grad_norm": 11.229409217834473,
      "learning_rate": 1.683835546007088e-05,
      "loss": 1.4317,
      "step": 42170
    },
    {
      "epoch": 6.970460648626317,
      "grad_norm": 12.30543327331543,
      "learning_rate": 1.682917424117226e-05,
      "loss": 1.4359,
      "step": 42180
    },
    {
      "epoch": 6.972113199752117,
      "grad_norm": 11.764647483825684,
      "learning_rate": 1.681999302227364e-05,
      "loss": 1.5617,
      "step": 42190
    },
    {
      "epoch": 6.973765750877918,
      "grad_norm": 10.844428062438965,
      "learning_rate": 1.6810811803375018e-05,
      "loss": 1.4419,
      "step": 42200
    },
    {
      "epoch": 6.975418302003718,
      "grad_norm": 10.631561279296875,
      "learning_rate": 1.6801630584476397e-05,
      "loss": 1.3841,
      "step": 42210
    },
    {
      "epoch": 6.9770708531295185,
      "grad_norm": 13.280062675476074,
      "learning_rate": 1.6792449365577776e-05,
      "loss": 1.35,
      "step": 42220
    },
    {
      "epoch": 6.9787234042553195,
      "grad_norm": 15.788164138793945,
      "learning_rate": 1.6783268146679155e-05,
      "loss": 1.334,
      "step": 42230
    },
    {
      "epoch": 6.9803759553811195,
      "grad_norm": 12.476130485534668,
      "learning_rate": 1.677408692778053e-05,
      "loss": 1.5234,
      "step": 42240
    },
    {
      "epoch": 6.98202850650692,
      "grad_norm": 24.663625717163086,
      "learning_rate": 1.676490570888191e-05,
      "loss": 1.3422,
      "step": 42250
    },
    {
      "epoch": 6.983681057632721,
      "grad_norm": 16.133569717407227,
      "learning_rate": 1.675572448998329e-05,
      "loss": 1.4352,
      "step": 42260
    },
    {
      "epoch": 6.985333608758521,
      "grad_norm": 11.581048965454102,
      "learning_rate": 1.674654327108467e-05,
      "loss": 1.4443,
      "step": 42270
    },
    {
      "epoch": 6.986986159884322,
      "grad_norm": 10.542519569396973,
      "learning_rate": 1.6737362052186048e-05,
      "loss": 1.3883,
      "step": 42280
    },
    {
      "epoch": 6.988638711010122,
      "grad_norm": 7.865337371826172,
      "learning_rate": 1.6728180833287428e-05,
      "loss": 1.493,
      "step": 42290
    },
    {
      "epoch": 6.990291262135923,
      "grad_norm": 12.757125854492188,
      "learning_rate": 1.6718999614388807e-05,
      "loss": 1.4717,
      "step": 42300
    },
    {
      "epoch": 6.991943813261723,
      "grad_norm": 13.490144729614258,
      "learning_rate": 1.6709818395490186e-05,
      "loss": 1.4115,
      "step": 42310
    },
    {
      "epoch": 6.993596364387523,
      "grad_norm": 14.82015609741211,
      "learning_rate": 1.6700637176591565e-05,
      "loss": 1.4881,
      "step": 42320
    },
    {
      "epoch": 6.995248915513324,
      "grad_norm": 23.44951057434082,
      "learning_rate": 1.6691455957692945e-05,
      "loss": 1.4348,
      "step": 42330
    },
    {
      "epoch": 6.996901466639124,
      "grad_norm": 8.522165298461914,
      "learning_rate": 1.6682274738794324e-05,
      "loss": 1.4091,
      "step": 42340
    },
    {
      "epoch": 6.998554017764924,
      "grad_norm": 26.870582580566406,
      "learning_rate": 1.6673093519895703e-05,
      "loss": 1.4368,
      "step": 42350
    },
    {
      "epoch": 6.999876058665565,
      "eval_accuracy": 0.32497517378351537,
      "eval_loss": 2.2579329013824463,
      "eval_runtime": 821.7981,
      "eval_samples_per_second": 34.31,
      "eval_steps_per_second": 8.578,
      "step": 42358
    },
    {
      "epoch": 7.000206568890725,
      "grad_norm": 7.999766826629639,
      "learning_rate": 1.6663912300997082e-05,
      "loss": 1.289,
      "step": 42360
    },
    {
      "epoch": 7.001859120016525,
      "grad_norm": 8.853545188903809,
      "learning_rate": 1.6654731082098458e-05,
      "loss": 1.2426,
      "step": 42370
    },
    {
      "epoch": 7.003511671142326,
      "grad_norm": 9.978639602661133,
      "learning_rate": 1.6645549863199837e-05,
      "loss": 1.4188,
      "step": 42380
    },
    {
      "epoch": 7.005164222268126,
      "grad_norm": 12.857678413391113,
      "learning_rate": 1.6636368644301217e-05,
      "loss": 1.3589,
      "step": 42390
    },
    {
      "epoch": 7.006816773393927,
      "grad_norm": 12.013607025146484,
      "learning_rate": 1.6627187425402596e-05,
      "loss": 1.4138,
      "step": 42400
    },
    {
      "epoch": 7.008469324519727,
      "grad_norm": 6.538505554199219,
      "learning_rate": 1.6618006206503975e-05,
      "loss": 1.4131,
      "step": 42410
    },
    {
      "epoch": 7.0101218756455275,
      "grad_norm": 7.0642595291137695,
      "learning_rate": 1.6608824987605358e-05,
      "loss": 1.3476,
      "step": 42420
    },
    {
      "epoch": 7.0117744267713285,
      "grad_norm": 17.850061416625977,
      "learning_rate": 1.6599643768706734e-05,
      "loss": 1.4026,
      "step": 42430
    },
    {
      "epoch": 7.0134269778971285,
      "grad_norm": 12.280431747436523,
      "learning_rate": 1.6590462549808113e-05,
      "loss": 1.3851,
      "step": 42440
    },
    {
      "epoch": 7.0150795290229295,
      "grad_norm": 9.028722763061523,
      "learning_rate": 1.6581281330909492e-05,
      "loss": 1.3367,
      "step": 42450
    },
    {
      "epoch": 7.01673208014873,
      "grad_norm": 15.112545013427734,
      "learning_rate": 1.657210011201087e-05,
      "loss": 1.4602,
      "step": 42460
    },
    {
      "epoch": 7.01838463127453,
      "grad_norm": 13.523941993713379,
      "learning_rate": 1.656291889311225e-05,
      "loss": 1.4501,
      "step": 42470
    },
    {
      "epoch": 7.020037182400331,
      "grad_norm": 8.846076011657715,
      "learning_rate": 1.655373767421363e-05,
      "loss": 1.4001,
      "step": 42480
    },
    {
      "epoch": 7.021689733526131,
      "grad_norm": 7.614100456237793,
      "learning_rate": 1.654455645531501e-05,
      "loss": 1.3485,
      "step": 42490
    },
    {
      "epoch": 7.023342284651932,
      "grad_norm": 11.066787719726562,
      "learning_rate": 1.6535375236416385e-05,
      "loss": 1.4282,
      "step": 42500
    },
    {
      "epoch": 7.024994835777732,
      "grad_norm": 11.001472473144531,
      "learning_rate": 1.6526194017517764e-05,
      "loss": 1.3722,
      "step": 42510
    },
    {
      "epoch": 7.026647386903532,
      "grad_norm": 12.738889694213867,
      "learning_rate": 1.6517012798619143e-05,
      "loss": 1.4098,
      "step": 42520
    },
    {
      "epoch": 7.028299938029333,
      "grad_norm": 11.663647651672363,
      "learning_rate": 1.6507831579720526e-05,
      "loss": 1.4311,
      "step": 42530
    },
    {
      "epoch": 7.029952489155133,
      "grad_norm": 22.87346076965332,
      "learning_rate": 1.6498650360821905e-05,
      "loss": 1.316,
      "step": 42540
    },
    {
      "epoch": 7.031605040280934,
      "grad_norm": 8.73465347290039,
      "learning_rate": 1.6489469141923285e-05,
      "loss": 1.4358,
      "step": 42550
    },
    {
      "epoch": 7.033257591406734,
      "grad_norm": 10.015544891357422,
      "learning_rate": 1.648028792302466e-05,
      "loss": 1.3549,
      "step": 42560
    },
    {
      "epoch": 7.034910142532534,
      "grad_norm": 10.383803367614746,
      "learning_rate": 1.647110670412604e-05,
      "loss": 1.2933,
      "step": 42570
    },
    {
      "epoch": 7.036562693658335,
      "grad_norm": 6.737658977508545,
      "learning_rate": 1.646192548522742e-05,
      "loss": 1.3442,
      "step": 42580
    },
    {
      "epoch": 7.038215244784135,
      "grad_norm": 20.091114044189453,
      "learning_rate": 1.6452744266328798e-05,
      "loss": 1.4975,
      "step": 42590
    },
    {
      "epoch": 7.039867795909936,
      "grad_norm": 27.708118438720703,
      "learning_rate": 1.6443563047430177e-05,
      "loss": 1.5107,
      "step": 42600
    },
    {
      "epoch": 7.041520347035736,
      "grad_norm": 15.870881080627441,
      "learning_rate": 1.6434381828531557e-05,
      "loss": 1.4584,
      "step": 42610
    },
    {
      "epoch": 7.0431728981615365,
      "grad_norm": 20.80101203918457,
      "learning_rate": 1.6425200609632936e-05,
      "loss": 1.4254,
      "step": 42620
    },
    {
      "epoch": 7.0448254492873374,
      "grad_norm": 17.202617645263672,
      "learning_rate": 1.6416019390734315e-05,
      "loss": 1.3899,
      "step": 42630
    },
    {
      "epoch": 7.0464780004131375,
      "grad_norm": 10.804183006286621,
      "learning_rate": 1.6406838171835694e-05,
      "loss": 1.3267,
      "step": 42640
    },
    {
      "epoch": 7.0481305515389385,
      "grad_norm": 26.866113662719727,
      "learning_rate": 1.6397656952937074e-05,
      "loss": 1.4871,
      "step": 42650
    },
    {
      "epoch": 7.049783102664739,
      "grad_norm": 13.534538269042969,
      "learning_rate": 1.6388475734038453e-05,
      "loss": 1.5625,
      "step": 42660
    },
    {
      "epoch": 7.051435653790539,
      "grad_norm": 9.568449020385742,
      "learning_rate": 1.6379294515139832e-05,
      "loss": 1.3345,
      "step": 42670
    },
    {
      "epoch": 7.05308820491634,
      "grad_norm": 6.132664680480957,
      "learning_rate": 1.637011329624121e-05,
      "loss": 1.4646,
      "step": 42680
    },
    {
      "epoch": 7.05474075604214,
      "grad_norm": 13.083625793457031,
      "learning_rate": 1.6360932077342587e-05,
      "loss": 1.3717,
      "step": 42690
    },
    {
      "epoch": 7.056393307167941,
      "grad_norm": 22.688730239868164,
      "learning_rate": 1.6351750858443967e-05,
      "loss": 1.321,
      "step": 42700
    },
    {
      "epoch": 7.058045858293741,
      "grad_norm": 13.621474266052246,
      "learning_rate": 1.6342569639545346e-05,
      "loss": 1.4109,
      "step": 42710
    },
    {
      "epoch": 7.059698409419542,
      "grad_norm": 16.44967269897461,
      "learning_rate": 1.6333388420646725e-05,
      "loss": 1.3616,
      "step": 42720
    },
    {
      "epoch": 7.061350960545342,
      "grad_norm": 16.369356155395508,
      "learning_rate": 1.6324207201748104e-05,
      "loss": 1.4134,
      "step": 42730
    },
    {
      "epoch": 7.063003511671142,
      "grad_norm": 11.338850975036621,
      "learning_rate": 1.6315025982849484e-05,
      "loss": 1.3899,
      "step": 42740
    },
    {
      "epoch": 7.064656062796943,
      "grad_norm": 13.041457176208496,
      "learning_rate": 1.6305844763950863e-05,
      "loss": 1.3229,
      "step": 42750
    },
    {
      "epoch": 7.066308613922743,
      "grad_norm": 18.309200286865234,
      "learning_rate": 1.6296663545052242e-05,
      "loss": 1.4822,
      "step": 42760
    },
    {
      "epoch": 7.067961165048544,
      "grad_norm": 14.844568252563477,
      "learning_rate": 1.628748232615362e-05,
      "loss": 1.5301,
      "step": 42770
    },
    {
      "epoch": 7.069613716174344,
      "grad_norm": 8.260869979858398,
      "learning_rate": 1.6278301107255e-05,
      "loss": 1.3789,
      "step": 42780
    },
    {
      "epoch": 7.071266267300144,
      "grad_norm": 20.73078727722168,
      "learning_rate": 1.626911988835638e-05,
      "loss": 1.3875,
      "step": 42790
    },
    {
      "epoch": 7.072918818425945,
      "grad_norm": 7.812931537628174,
      "learning_rate": 1.625993866945776e-05,
      "loss": 1.3335,
      "step": 42800
    },
    {
      "epoch": 7.074571369551745,
      "grad_norm": 10.699761390686035,
      "learning_rate": 1.6250757450559138e-05,
      "loss": 1.2201,
      "step": 42810
    },
    {
      "epoch": 7.076223920677546,
      "grad_norm": 13.993300437927246,
      "learning_rate": 1.6241576231660514e-05,
      "loss": 1.3617,
      "step": 42820
    },
    {
      "epoch": 7.0778764718033464,
      "grad_norm": 7.880179405212402,
      "learning_rate": 1.6232395012761893e-05,
      "loss": 1.3303,
      "step": 42830
    },
    {
      "epoch": 7.0795290229291465,
      "grad_norm": 14.917696952819824,
      "learning_rate": 1.6223213793863273e-05,
      "loss": 1.383,
      "step": 42840
    },
    {
      "epoch": 7.0811815740549475,
      "grad_norm": 62.78154373168945,
      "learning_rate": 1.6214032574964652e-05,
      "loss": 1.5036,
      "step": 42850
    },
    {
      "epoch": 7.082834125180748,
      "grad_norm": 50.27464294433594,
      "learning_rate": 1.620485135606603e-05,
      "loss": 1.3522,
      "step": 42860
    },
    {
      "epoch": 7.084486676306549,
      "grad_norm": 19.16926383972168,
      "learning_rate": 1.6195670137167414e-05,
      "loss": 1.4757,
      "step": 42870
    },
    {
      "epoch": 7.086139227432349,
      "grad_norm": 13.433306694030762,
      "learning_rate": 1.618648891826879e-05,
      "loss": 1.3875,
      "step": 42880
    },
    {
      "epoch": 7.087791778558149,
      "grad_norm": 13.95352554321289,
      "learning_rate": 1.617730769937017e-05,
      "loss": 1.4077,
      "step": 42890
    },
    {
      "epoch": 7.08944432968395,
      "grad_norm": 11.004186630249023,
      "learning_rate": 1.6168126480471548e-05,
      "loss": 1.2435,
      "step": 42900
    },
    {
      "epoch": 7.09109688080975,
      "grad_norm": 9.82064151763916,
      "learning_rate": 1.6158945261572927e-05,
      "loss": 1.3743,
      "step": 42910
    },
    {
      "epoch": 7.092749431935551,
      "grad_norm": 11.183409690856934,
      "learning_rate": 1.6149764042674307e-05,
      "loss": 1.2179,
      "step": 42920
    },
    {
      "epoch": 7.094401983061351,
      "grad_norm": 12.038310050964355,
      "learning_rate": 1.6140582823775686e-05,
      "loss": 1.4433,
      "step": 42930
    },
    {
      "epoch": 7.096054534187151,
      "grad_norm": 14.014615058898926,
      "learning_rate": 1.6131401604877065e-05,
      "loss": 1.4148,
      "step": 42940
    },
    {
      "epoch": 7.097707085312952,
      "grad_norm": 11.169940948486328,
      "learning_rate": 1.6122220385978444e-05,
      "loss": 1.4004,
      "step": 42950
    },
    {
      "epoch": 7.099359636438752,
      "grad_norm": 12.410701751708984,
      "learning_rate": 1.611303916707982e-05,
      "loss": 1.4521,
      "step": 42960
    },
    {
      "epoch": 7.101012187564553,
      "grad_norm": 13.273151397705078,
      "learning_rate": 1.61038579481812e-05,
      "loss": 1.5151,
      "step": 42970
    },
    {
      "epoch": 7.102664738690353,
      "grad_norm": 8.128931999206543,
      "learning_rate": 1.609467672928258e-05,
      "loss": 1.4828,
      "step": 42980
    },
    {
      "epoch": 7.104317289816153,
      "grad_norm": 26.178449630737305,
      "learning_rate": 1.608549551038396e-05,
      "loss": 1.3804,
      "step": 42990
    },
    {
      "epoch": 7.105969840941954,
      "grad_norm": 9.145550727844238,
      "learning_rate": 1.607631429148534e-05,
      "loss": 1.3358,
      "step": 43000
    },
    {
      "epoch": 7.107622392067754,
      "grad_norm": 10.598913192749023,
      "learning_rate": 1.6067133072586716e-05,
      "loss": 1.4074,
      "step": 43010
    },
    {
      "epoch": 7.109274943193555,
      "grad_norm": 13.98646068572998,
      "learning_rate": 1.6057951853688096e-05,
      "loss": 1.4561,
      "step": 43020
    },
    {
      "epoch": 7.1109274943193554,
      "grad_norm": 15.383737564086914,
      "learning_rate": 1.6048770634789475e-05,
      "loss": 1.3876,
      "step": 43030
    },
    {
      "epoch": 7.1125800454451555,
      "grad_norm": 14.499407768249512,
      "learning_rate": 1.6039589415890854e-05,
      "loss": 1.3743,
      "step": 43040
    },
    {
      "epoch": 7.1142325965709565,
      "grad_norm": 21.623125076293945,
      "learning_rate": 1.6030408196992233e-05,
      "loss": 1.3869,
      "step": 43050
    },
    {
      "epoch": 7.115885147696757,
      "grad_norm": 19.47028350830078,
      "learning_rate": 1.6021226978093613e-05,
      "loss": 1.4454,
      "step": 43060
    },
    {
      "epoch": 7.117537698822558,
      "grad_norm": 12.078246116638184,
      "learning_rate": 1.6012045759194992e-05,
      "loss": 1.396,
      "step": 43070
    },
    {
      "epoch": 7.119190249948358,
      "grad_norm": 11.143622398376465,
      "learning_rate": 1.600286454029637e-05,
      "loss": 1.5037,
      "step": 43080
    },
    {
      "epoch": 7.120842801074158,
      "grad_norm": 35.760719299316406,
      "learning_rate": 1.5993683321397747e-05,
      "loss": 1.4383,
      "step": 43090
    },
    {
      "epoch": 7.122495352199959,
      "grad_norm": 23.67930030822754,
      "learning_rate": 1.598450210249913e-05,
      "loss": 1.415,
      "step": 43100
    },
    {
      "epoch": 7.124147903325759,
      "grad_norm": 8.731281280517578,
      "learning_rate": 1.597532088360051e-05,
      "loss": 1.4276,
      "step": 43110
    },
    {
      "epoch": 7.12580045445156,
      "grad_norm": 16.28318214416504,
      "learning_rate": 1.5966139664701888e-05,
      "loss": 1.4406,
      "step": 43120
    },
    {
      "epoch": 7.12745300557736,
      "grad_norm": 15.253238677978516,
      "learning_rate": 1.5956958445803267e-05,
      "loss": 1.4224,
      "step": 43130
    },
    {
      "epoch": 7.129105556703161,
      "grad_norm": 9.880094528198242,
      "learning_rate": 1.5947777226904643e-05,
      "loss": 1.289,
      "step": 43140
    },
    {
      "epoch": 7.130758107828961,
      "grad_norm": 45.78981399536133,
      "learning_rate": 1.5938596008006022e-05,
      "loss": 1.4237,
      "step": 43150
    },
    {
      "epoch": 7.132410658954761,
      "grad_norm": 15.530732154846191,
      "learning_rate": 1.5929414789107402e-05,
      "loss": 1.3385,
      "step": 43160
    },
    {
      "epoch": 7.134063210080562,
      "grad_norm": 7.943971157073975,
      "learning_rate": 1.592023357020878e-05,
      "loss": 1.3658,
      "step": 43170
    },
    {
      "epoch": 7.135715761206362,
      "grad_norm": 8.042201042175293,
      "learning_rate": 1.591105235131016e-05,
      "loss": 1.3703,
      "step": 43180
    },
    {
      "epoch": 7.137368312332163,
      "grad_norm": 14.290609359741211,
      "learning_rate": 1.590187113241154e-05,
      "loss": 1.433,
      "step": 43190
    },
    {
      "epoch": 7.139020863457963,
      "grad_norm": 12.84965991973877,
      "learning_rate": 1.589268991351292e-05,
      "loss": 1.3675,
      "step": 43200
    },
    {
      "epoch": 7.140673414583763,
      "grad_norm": 9.75621509552002,
      "learning_rate": 1.5883508694614298e-05,
      "loss": 1.3423,
      "step": 43210
    },
    {
      "epoch": 7.142325965709564,
      "grad_norm": 16.249305725097656,
      "learning_rate": 1.5874327475715677e-05,
      "loss": 1.5104,
      "step": 43220
    },
    {
      "epoch": 7.1439785168353644,
      "grad_norm": 9.561155319213867,
      "learning_rate": 1.5865146256817056e-05,
      "loss": 1.4316,
      "step": 43230
    },
    {
      "epoch": 7.145631067961165,
      "grad_norm": 9.332865715026855,
      "learning_rate": 1.5855965037918436e-05,
      "loss": 1.3956,
      "step": 43240
    },
    {
      "epoch": 7.1472836190869655,
      "grad_norm": 14.026449203491211,
      "learning_rate": 1.5846783819019815e-05,
      "loss": 1.3913,
      "step": 43250
    },
    {
      "epoch": 7.148936170212766,
      "grad_norm": 16.361709594726562,
      "learning_rate": 1.5837602600121194e-05,
      "loss": 1.4388,
      "step": 43260
    },
    {
      "epoch": 7.150588721338567,
      "grad_norm": 34.313411712646484,
      "learning_rate": 1.5828421381222573e-05,
      "loss": 1.4463,
      "step": 43270
    },
    {
      "epoch": 7.152241272464367,
      "grad_norm": 62.70059585571289,
      "learning_rate": 1.581924016232395e-05,
      "loss": 1.5965,
      "step": 43280
    },
    {
      "epoch": 7.153893823590168,
      "grad_norm": 12.427000999450684,
      "learning_rate": 1.581005894342533e-05,
      "loss": 1.4107,
      "step": 43290
    },
    {
      "epoch": 7.155546374715968,
      "grad_norm": 38.85627365112305,
      "learning_rate": 1.5800877724526708e-05,
      "loss": 1.4769,
      "step": 43300
    },
    {
      "epoch": 7.157198925841768,
      "grad_norm": 13.122133255004883,
      "learning_rate": 1.5791696505628087e-05,
      "loss": 1.3648,
      "step": 43310
    },
    {
      "epoch": 7.158851476967569,
      "grad_norm": 44.62786102294922,
      "learning_rate": 1.5782515286729466e-05,
      "loss": 1.5429,
      "step": 43320
    },
    {
      "epoch": 7.160504028093369,
      "grad_norm": 12.482089042663574,
      "learning_rate": 1.5773334067830846e-05,
      "loss": 1.3703,
      "step": 43330
    },
    {
      "epoch": 7.16215657921917,
      "grad_norm": 10.335782051086426,
      "learning_rate": 1.5764152848932225e-05,
      "loss": 1.3795,
      "step": 43340
    },
    {
      "epoch": 7.16380913034497,
      "grad_norm": 11.581258773803711,
      "learning_rate": 1.5754971630033604e-05,
      "loss": 1.3594,
      "step": 43350
    },
    {
      "epoch": 7.16546168147077,
      "grad_norm": 9.168930053710938,
      "learning_rate": 1.5745790411134983e-05,
      "loss": 1.3988,
      "step": 43360
    },
    {
      "epoch": 7.167114232596571,
      "grad_norm": 35.34038543701172,
      "learning_rate": 1.5736609192236363e-05,
      "loss": 1.367,
      "step": 43370
    },
    {
      "epoch": 7.168766783722371,
      "grad_norm": 13.737029075622559,
      "learning_rate": 1.5727427973337742e-05,
      "loss": 1.4333,
      "step": 43380
    },
    {
      "epoch": 7.170419334848172,
      "grad_norm": 11.130227088928223,
      "learning_rate": 1.571824675443912e-05,
      "loss": 1.3601,
      "step": 43390
    },
    {
      "epoch": 7.172071885973972,
      "grad_norm": 14.647340774536133,
      "learning_rate": 1.57090655355405e-05,
      "loss": 1.4556,
      "step": 43400
    },
    {
      "epoch": 7.173724437099772,
      "grad_norm": 11.120412826538086,
      "learning_rate": 1.5699884316641876e-05,
      "loss": 1.4372,
      "step": 43410
    },
    {
      "epoch": 7.175376988225573,
      "grad_norm": 51.608375549316406,
      "learning_rate": 1.5690703097743255e-05,
      "loss": 1.4654,
      "step": 43420
    },
    {
      "epoch": 7.177029539351373,
      "grad_norm": 10.668750762939453,
      "learning_rate": 1.5681521878844635e-05,
      "loss": 1.2519,
      "step": 43430
    },
    {
      "epoch": 7.178682090477174,
      "grad_norm": 12.485282897949219,
      "learning_rate": 1.5672340659946017e-05,
      "loss": 1.322,
      "step": 43440
    },
    {
      "epoch": 7.1803346416029745,
      "grad_norm": 12.183808326721191,
      "learning_rate": 1.5663159441047397e-05,
      "loss": 1.4264,
      "step": 43450
    },
    {
      "epoch": 7.181987192728775,
      "grad_norm": 13.21921443939209,
      "learning_rate": 1.5653978222148772e-05,
      "loss": 1.3703,
      "step": 43460
    },
    {
      "epoch": 7.183639743854576,
      "grad_norm": 10.9345064163208,
      "learning_rate": 1.564479700325015e-05,
      "loss": 1.2281,
      "step": 43470
    },
    {
      "epoch": 7.185292294980376,
      "grad_norm": 14.405402183532715,
      "learning_rate": 1.563561578435153e-05,
      "loss": 1.4111,
      "step": 43480
    },
    {
      "epoch": 7.186944846106177,
      "grad_norm": 17.535526275634766,
      "learning_rate": 1.562643456545291e-05,
      "loss": 1.3301,
      "step": 43490
    },
    {
      "epoch": 7.188597397231977,
      "grad_norm": 15.74835205078125,
      "learning_rate": 1.561725334655429e-05,
      "loss": 1.5023,
      "step": 43500
    },
    {
      "epoch": 7.190249948357778,
      "grad_norm": 29.075929641723633,
      "learning_rate": 1.560807212765567e-05,
      "loss": 1.443,
      "step": 43510
    },
    {
      "epoch": 7.191902499483578,
      "grad_norm": 8.865483283996582,
      "learning_rate": 1.5598890908757048e-05,
      "loss": 1.3097,
      "step": 43520
    },
    {
      "epoch": 7.193555050609378,
      "grad_norm": 14.490361213684082,
      "learning_rate": 1.5589709689858427e-05,
      "loss": 1.3695,
      "step": 43530
    },
    {
      "epoch": 7.195207601735179,
      "grad_norm": 21.584367752075195,
      "learning_rate": 1.5580528470959803e-05,
      "loss": 1.3773,
      "step": 43540
    },
    {
      "epoch": 7.196860152860979,
      "grad_norm": 10.539822578430176,
      "learning_rate": 1.5571347252061182e-05,
      "loss": 1.4,
      "step": 43550
    },
    {
      "epoch": 7.19851270398678,
      "grad_norm": 12.214553833007812,
      "learning_rate": 1.5562166033162565e-05,
      "loss": 1.3988,
      "step": 43560
    },
    {
      "epoch": 7.20016525511258,
      "grad_norm": 13.141803741455078,
      "learning_rate": 1.5552984814263944e-05,
      "loss": 1.3828,
      "step": 43570
    },
    {
      "epoch": 7.20181780623838,
      "grad_norm": 12.777125358581543,
      "learning_rate": 1.5543803595365323e-05,
      "loss": 1.3906,
      "step": 43580
    },
    {
      "epoch": 7.203470357364181,
      "grad_norm": 15.37719440460205,
      "learning_rate": 1.5534622376466703e-05,
      "loss": 1.4042,
      "step": 43590
    },
    {
      "epoch": 7.205122908489981,
      "grad_norm": 12.876169204711914,
      "learning_rate": 1.552544115756808e-05,
      "loss": 1.582,
      "step": 43600
    },
    {
      "epoch": 7.206775459615782,
      "grad_norm": 21.468862533569336,
      "learning_rate": 1.5516259938669458e-05,
      "loss": 1.4693,
      "step": 43610
    },
    {
      "epoch": 7.208428010741582,
      "grad_norm": 10.771493911743164,
      "learning_rate": 1.5507078719770837e-05,
      "loss": 1.4106,
      "step": 43620
    },
    {
      "epoch": 7.210080561867382,
      "grad_norm": 16.791486740112305,
      "learning_rate": 1.5497897500872216e-05,
      "loss": 1.4429,
      "step": 43630
    },
    {
      "epoch": 7.211733112993183,
      "grad_norm": 16.35462760925293,
      "learning_rate": 1.5488716281973595e-05,
      "loss": 1.3854,
      "step": 43640
    },
    {
      "epoch": 7.2133856641189835,
      "grad_norm": 12.617788314819336,
      "learning_rate": 1.5479535063074975e-05,
      "loss": 1.3077,
      "step": 43650
    },
    {
      "epoch": 7.2150382152447845,
      "grad_norm": 12.411581993103027,
      "learning_rate": 1.5470353844176354e-05,
      "loss": 1.3623,
      "step": 43660
    },
    {
      "epoch": 7.216690766370585,
      "grad_norm": 19.033109664916992,
      "learning_rate": 1.5461172625277733e-05,
      "loss": 1.4943,
      "step": 43670
    },
    {
      "epoch": 7.218343317496385,
      "grad_norm": 8.353110313415527,
      "learning_rate": 1.5451991406379112e-05,
      "loss": 1.3783,
      "step": 43680
    },
    {
      "epoch": 7.219995868622186,
      "grad_norm": 8.418465614318848,
      "learning_rate": 1.5442810187480492e-05,
      "loss": 1.3545,
      "step": 43690
    },
    {
      "epoch": 7.221648419747986,
      "grad_norm": 8.319928169250488,
      "learning_rate": 1.543362896858187e-05,
      "loss": 1.3794,
      "step": 43700
    },
    {
      "epoch": 7.223300970873787,
      "grad_norm": 13.188324928283691,
      "learning_rate": 1.542444774968325e-05,
      "loss": 1.2745,
      "step": 43710
    },
    {
      "epoch": 7.224953521999587,
      "grad_norm": 10.732872009277344,
      "learning_rate": 1.541526653078463e-05,
      "loss": 1.3477,
      "step": 43720
    },
    {
      "epoch": 7.226606073125387,
      "grad_norm": 47.982398986816406,
      "learning_rate": 1.5406085311886005e-05,
      "loss": 1.4172,
      "step": 43730
    },
    {
      "epoch": 7.228258624251188,
      "grad_norm": 44.85279846191406,
      "learning_rate": 1.5396904092987385e-05,
      "loss": 1.4017,
      "step": 43740
    },
    {
      "epoch": 7.229911175376988,
      "grad_norm": 9.955687522888184,
      "learning_rate": 1.5387722874088764e-05,
      "loss": 1.349,
      "step": 43750
    },
    {
      "epoch": 7.231563726502789,
      "grad_norm": 8.512934684753418,
      "learning_rate": 1.5378541655190143e-05,
      "loss": 1.4837,
      "step": 43760
    },
    {
      "epoch": 7.233216277628589,
      "grad_norm": 17.035743713378906,
      "learning_rate": 1.5369360436291522e-05,
      "loss": 1.4043,
      "step": 43770
    },
    {
      "epoch": 7.234868828754389,
      "grad_norm": 45.515899658203125,
      "learning_rate": 1.53601792173929e-05,
      "loss": 1.2978,
      "step": 43780
    },
    {
      "epoch": 7.23652137988019,
      "grad_norm": 13.9974946975708,
      "learning_rate": 1.535099799849428e-05,
      "loss": 1.4702,
      "step": 43790
    },
    {
      "epoch": 7.23817393100599,
      "grad_norm": 16.096546173095703,
      "learning_rate": 1.534181677959566e-05,
      "loss": 1.4132,
      "step": 43800
    },
    {
      "epoch": 7.239826482131791,
      "grad_norm": 11.778483390808105,
      "learning_rate": 1.533263556069704e-05,
      "loss": 1.3275,
      "step": 43810
    },
    {
      "epoch": 7.241479033257591,
      "grad_norm": 7.991126537322998,
      "learning_rate": 1.532345434179842e-05,
      "loss": 1.2556,
      "step": 43820
    },
    {
      "epoch": 7.243131584383391,
      "grad_norm": 12.626638412475586,
      "learning_rate": 1.5314273122899798e-05,
      "loss": 1.38,
      "step": 43830
    },
    {
      "epoch": 7.244784135509192,
      "grad_norm": 7.913274765014648,
      "learning_rate": 1.5305091904001177e-05,
      "loss": 1.253,
      "step": 43840
    },
    {
      "epoch": 7.2464366866349925,
      "grad_norm": 11.45130729675293,
      "learning_rate": 1.5295910685102556e-05,
      "loss": 1.4744,
      "step": 43850
    },
    {
      "epoch": 7.2480892377607935,
      "grad_norm": 7.596304416656494,
      "learning_rate": 1.5286729466203932e-05,
      "loss": 1.3495,
      "step": 43860
    },
    {
      "epoch": 7.249741788886594,
      "grad_norm": 16.072736740112305,
      "learning_rate": 1.527754824730531e-05,
      "loss": 1.3415,
      "step": 43870
    },
    {
      "epoch": 7.251394340012395,
      "grad_norm": 11.912080764770508,
      "learning_rate": 1.526836702840669e-05,
      "loss": 1.4074,
      "step": 43880
    },
    {
      "epoch": 7.253046891138195,
      "grad_norm": 14.602355003356934,
      "learning_rate": 1.525918580950807e-05,
      "loss": 1.4189,
      "step": 43890
    },
    {
      "epoch": 7.254699442263995,
      "grad_norm": 9.636602401733398,
      "learning_rate": 1.525000459060945e-05,
      "loss": 1.37,
      "step": 43900
    },
    {
      "epoch": 7.256351993389796,
      "grad_norm": 22.76931381225586,
      "learning_rate": 1.524082337171083e-05,
      "loss": 1.3947,
      "step": 43910
    },
    {
      "epoch": 7.258004544515596,
      "grad_norm": 15.332125663757324,
      "learning_rate": 1.523164215281221e-05,
      "loss": 1.433,
      "step": 43920
    },
    {
      "epoch": 7.259657095641396,
      "grad_norm": 13.930269241333008,
      "learning_rate": 1.5222460933913587e-05,
      "loss": 1.4699,
      "step": 43930
    },
    {
      "epoch": 7.261309646767197,
      "grad_norm": 13.538143157958984,
      "learning_rate": 1.5213279715014966e-05,
      "loss": 1.398,
      "step": 43940
    },
    {
      "epoch": 7.262962197892997,
      "grad_norm": 12.98928451538086,
      "learning_rate": 1.5204098496116345e-05,
      "loss": 1.2564,
      "step": 43950
    },
    {
      "epoch": 7.264614749018798,
      "grad_norm": 17.9935359954834,
      "learning_rate": 1.5194917277217725e-05,
      "loss": 1.4769,
      "step": 43960
    },
    {
      "epoch": 7.266267300144598,
      "grad_norm": 13.365751266479492,
      "learning_rate": 1.5185736058319102e-05,
      "loss": 1.4451,
      "step": 43970
    },
    {
      "epoch": 7.267919851270399,
      "grad_norm": 16.52039337158203,
      "learning_rate": 1.5176554839420481e-05,
      "loss": 1.4016,
      "step": 43980
    },
    {
      "epoch": 7.269572402396199,
      "grad_norm": 15.098515510559082,
      "learning_rate": 1.516737362052186e-05,
      "loss": 1.369,
      "step": 43990
    },
    {
      "epoch": 7.271224953521999,
      "grad_norm": 8.429466247558594,
      "learning_rate": 1.5158192401623238e-05,
      "loss": 1.4621,
      "step": 44000
    },
    {
      "epoch": 7.2728775046478,
      "grad_norm": 13.037130355834961,
      "learning_rate": 1.5149011182724621e-05,
      "loss": 1.4036,
      "step": 44010
    },
    {
      "epoch": 7.2745300557736,
      "grad_norm": 11.745615005493164,
      "learning_rate": 1.5139829963826e-05,
      "loss": 1.3655,
      "step": 44020
    },
    {
      "epoch": 7.276182606899401,
      "grad_norm": 25.232398986816406,
      "learning_rate": 1.5130648744927378e-05,
      "loss": 1.424,
      "step": 44030
    },
    {
      "epoch": 7.277835158025201,
      "grad_norm": 10.532912254333496,
      "learning_rate": 1.5121467526028757e-05,
      "loss": 1.4246,
      "step": 44040
    },
    {
      "epoch": 7.2794877091510015,
      "grad_norm": 9.035202026367188,
      "learning_rate": 1.5112286307130136e-05,
      "loss": 1.3259,
      "step": 44050
    },
    {
      "epoch": 7.2811402602768025,
      "grad_norm": 10.761393547058105,
      "learning_rate": 1.5103105088231514e-05,
      "loss": 1.4091,
      "step": 44060
    },
    {
      "epoch": 7.282792811402603,
      "grad_norm": 21.04129409790039,
      "learning_rate": 1.5093923869332893e-05,
      "loss": 1.3684,
      "step": 44070
    },
    {
      "epoch": 7.284445362528404,
      "grad_norm": 6.939519882202148,
      "learning_rate": 1.5084742650434272e-05,
      "loss": 1.5599,
      "step": 44080
    },
    {
      "epoch": 7.286097913654204,
      "grad_norm": 74.00979614257812,
      "learning_rate": 1.5075561431535651e-05,
      "loss": 1.4212,
      "step": 44090
    },
    {
      "epoch": 7.287750464780004,
      "grad_norm": 11.299165725708008,
      "learning_rate": 1.5066380212637029e-05,
      "loss": 1.3967,
      "step": 44100
    },
    {
      "epoch": 7.289403015905805,
      "grad_norm": 9.076863288879395,
      "learning_rate": 1.5057198993738408e-05,
      "loss": 1.3786,
      "step": 44110
    },
    {
      "epoch": 7.291055567031605,
      "grad_norm": 10.693045616149902,
      "learning_rate": 1.5048017774839787e-05,
      "loss": 1.4005,
      "step": 44120
    },
    {
      "epoch": 7.292708118157406,
      "grad_norm": 11.115889549255371,
      "learning_rate": 1.5038836555941168e-05,
      "loss": 1.512,
      "step": 44130
    },
    {
      "epoch": 7.294360669283206,
      "grad_norm": 10.336817741394043,
      "learning_rate": 1.5029655337042548e-05,
      "loss": 1.3774,
      "step": 44140
    },
    {
      "epoch": 7.296013220409006,
      "grad_norm": 11.22537612915039,
      "learning_rate": 1.5020474118143927e-05,
      "loss": 1.4058,
      "step": 44150
    },
    {
      "epoch": 7.297665771534807,
      "grad_norm": 43.39937210083008,
      "learning_rate": 1.5011292899245304e-05,
      "loss": 1.3649,
      "step": 44160
    },
    {
      "epoch": 7.299318322660607,
      "grad_norm": 22.82792091369629,
      "learning_rate": 1.5002111680346684e-05,
      "loss": 1.5085,
      "step": 44170
    },
    {
      "epoch": 7.300970873786408,
      "grad_norm": 30.476633071899414,
      "learning_rate": 1.4992930461448063e-05,
      "loss": 1.32,
      "step": 44180
    },
    {
      "epoch": 7.302623424912208,
      "grad_norm": 10.81251049041748,
      "learning_rate": 1.498374924254944e-05,
      "loss": 1.3552,
      "step": 44190
    },
    {
      "epoch": 7.304275976038008,
      "grad_norm": 17.479726791381836,
      "learning_rate": 1.497456802365082e-05,
      "loss": 1.437,
      "step": 44200
    },
    {
      "epoch": 7.305928527163809,
      "grad_norm": 14.184757232666016,
      "learning_rate": 1.4965386804752199e-05,
      "loss": 1.3092,
      "step": 44210
    },
    {
      "epoch": 7.307581078289609,
      "grad_norm": 16.718236923217773,
      "learning_rate": 1.4956205585853578e-05,
      "loss": 1.4491,
      "step": 44220
    },
    {
      "epoch": 7.30923362941541,
      "grad_norm": 13.425004005432129,
      "learning_rate": 1.4947024366954956e-05,
      "loss": 1.4319,
      "step": 44230
    },
    {
      "epoch": 7.31088618054121,
      "grad_norm": 12.716981887817383,
      "learning_rate": 1.4937843148056338e-05,
      "loss": 1.365,
      "step": 44240
    },
    {
      "epoch": 7.312538731667011,
      "grad_norm": 11.309150695800781,
      "learning_rate": 1.4928661929157716e-05,
      "loss": 1.3449,
      "step": 44250
    },
    {
      "epoch": 7.3141912827928115,
      "grad_norm": 9.994693756103516,
      "learning_rate": 1.4919480710259095e-05,
      "loss": 1.3902,
      "step": 44260
    },
    {
      "epoch": 7.315843833918612,
      "grad_norm": 12.569499015808105,
      "learning_rate": 1.4910299491360475e-05,
      "loss": 1.2932,
      "step": 44270
    },
    {
      "epoch": 7.317496385044413,
      "grad_norm": 14.061833381652832,
      "learning_rate": 1.4901118272461854e-05,
      "loss": 1.3171,
      "step": 44280
    },
    {
      "epoch": 7.319148936170213,
      "grad_norm": 17.114004135131836,
      "learning_rate": 1.4891937053563231e-05,
      "loss": 1.4204,
      "step": 44290
    },
    {
      "epoch": 7.320801487296013,
      "grad_norm": 20.603870391845703,
      "learning_rate": 1.488275583466461e-05,
      "loss": 1.4114,
      "step": 44300
    },
    {
      "epoch": 7.322454038421814,
      "grad_norm": 11.82662582397461,
      "learning_rate": 1.487357461576599e-05,
      "loss": 1.3978,
      "step": 44310
    },
    {
      "epoch": 7.324106589547614,
      "grad_norm": 8.087780952453613,
      "learning_rate": 1.4864393396867367e-05,
      "loss": 1.4033,
      "step": 44320
    },
    {
      "epoch": 7.325759140673415,
      "grad_norm": 14.538772583007812,
      "learning_rate": 1.4855212177968747e-05,
      "loss": 1.4692,
      "step": 44330
    },
    {
      "epoch": 7.327411691799215,
      "grad_norm": 11.816965103149414,
      "learning_rate": 1.4846030959070126e-05,
      "loss": 1.4003,
      "step": 44340
    },
    {
      "epoch": 7.329064242925016,
      "grad_norm": 11.297624588012695,
      "learning_rate": 1.4836849740171505e-05,
      "loss": 1.3688,
      "step": 44350
    },
    {
      "epoch": 7.330716794050816,
      "grad_norm": 29.10480308532715,
      "learning_rate": 1.4827668521272886e-05,
      "loss": 1.3417,
      "step": 44360
    },
    {
      "epoch": 7.332369345176616,
      "grad_norm": 16.348495483398438,
      "learning_rate": 1.4818487302374265e-05,
      "loss": 1.3799,
      "step": 44370
    },
    {
      "epoch": 7.334021896302417,
      "grad_norm": 15.20343017578125,
      "learning_rate": 1.4809306083475643e-05,
      "loss": 1.3703,
      "step": 44380
    },
    {
      "epoch": 7.335674447428217,
      "grad_norm": 10.832907676696777,
      "learning_rate": 1.4800124864577022e-05,
      "loss": 1.5541,
      "step": 44390
    },
    {
      "epoch": 7.337326998554018,
      "grad_norm": 13.253135681152344,
      "learning_rate": 1.4790943645678401e-05,
      "loss": 1.346,
      "step": 44400
    },
    {
      "epoch": 7.338979549679818,
      "grad_norm": 12.137068748474121,
      "learning_rate": 1.478176242677978e-05,
      "loss": 1.4282,
      "step": 44410
    },
    {
      "epoch": 7.340632100805618,
      "grad_norm": 17.6302547454834,
      "learning_rate": 1.4772581207881158e-05,
      "loss": 1.2844,
      "step": 44420
    },
    {
      "epoch": 7.342284651931419,
      "grad_norm": 8.67888355255127,
      "learning_rate": 1.4763399988982537e-05,
      "loss": 1.2272,
      "step": 44430
    },
    {
      "epoch": 7.343937203057219,
      "grad_norm": 12.13434886932373,
      "learning_rate": 1.4754218770083917e-05,
      "loss": 1.4605,
      "step": 44440
    },
    {
      "epoch": 7.34558975418302,
      "grad_norm": 13.558122634887695,
      "learning_rate": 1.4745037551185296e-05,
      "loss": 1.3914,
      "step": 44450
    },
    {
      "epoch": 7.3472423053088205,
      "grad_norm": 11.792350769042969,
      "learning_rate": 1.4735856332286673e-05,
      "loss": 1.3819,
      "step": 44460
    },
    {
      "epoch": 7.348894856434621,
      "grad_norm": 15.687477111816406,
      "learning_rate": 1.4726675113388056e-05,
      "loss": 1.4004,
      "step": 44470
    },
    {
      "epoch": 7.350547407560422,
      "grad_norm": 14.300453186035156,
      "learning_rate": 1.4717493894489434e-05,
      "loss": 1.3687,
      "step": 44480
    },
    {
      "epoch": 7.352199958686222,
      "grad_norm": 12.354840278625488,
      "learning_rate": 1.4708312675590813e-05,
      "loss": 1.3209,
      "step": 44490
    },
    {
      "epoch": 7.353852509812023,
      "grad_norm": 14.35772705078125,
      "learning_rate": 1.4699131456692192e-05,
      "loss": 1.428,
      "step": 44500
    },
    {
      "epoch": 7.355505060937823,
      "grad_norm": 16.01991844177246,
      "learning_rate": 1.468995023779357e-05,
      "loss": 1.3319,
      "step": 44510
    },
    {
      "epoch": 7.357157612063623,
      "grad_norm": 14.260498046875,
      "learning_rate": 1.4680769018894949e-05,
      "loss": 1.3704,
      "step": 44520
    },
    {
      "epoch": 7.358810163189424,
      "grad_norm": 17.772220611572266,
      "learning_rate": 1.4671587799996328e-05,
      "loss": 1.552,
      "step": 44530
    },
    {
      "epoch": 7.360462714315224,
      "grad_norm": 13.813607215881348,
      "learning_rate": 1.4662406581097707e-05,
      "loss": 1.4813,
      "step": 44540
    },
    {
      "epoch": 7.362115265441025,
      "grad_norm": 11.745838165283203,
      "learning_rate": 1.4653225362199085e-05,
      "loss": 1.396,
      "step": 44550
    },
    {
      "epoch": 7.363767816566825,
      "grad_norm": 24.77311897277832,
      "learning_rate": 1.4644044143300464e-05,
      "loss": 1.4673,
      "step": 44560
    },
    {
      "epoch": 7.365420367692625,
      "grad_norm": 12.793754577636719,
      "learning_rate": 1.4634862924401843e-05,
      "loss": 1.3443,
      "step": 44570
    },
    {
      "epoch": 7.367072918818426,
      "grad_norm": 10.67249870300293,
      "learning_rate": 1.4625681705503224e-05,
      "loss": 1.3569,
      "step": 44580
    },
    {
      "epoch": 7.368725469944226,
      "grad_norm": 15.902300834655762,
      "learning_rate": 1.4616500486604604e-05,
      "loss": 1.5068,
      "step": 44590
    },
    {
      "epoch": 7.370378021070027,
      "grad_norm": 9.883173942565918,
      "learning_rate": 1.4607319267705983e-05,
      "loss": 1.2706,
      "step": 44600
    },
    {
      "epoch": 7.372030572195827,
      "grad_norm": 17.580493927001953,
      "learning_rate": 1.459813804880736e-05,
      "loss": 1.3805,
      "step": 44610
    },
    {
      "epoch": 7.373683123321628,
      "grad_norm": 20.290407180786133,
      "learning_rate": 1.458895682990874e-05,
      "loss": 1.5256,
      "step": 44620
    },
    {
      "epoch": 7.375335674447428,
      "grad_norm": 11.272130012512207,
      "learning_rate": 1.4579775611010119e-05,
      "loss": 1.4099,
      "step": 44630
    },
    {
      "epoch": 7.376988225573228,
      "grad_norm": 14.19814395904541,
      "learning_rate": 1.4570594392111497e-05,
      "loss": 1.4281,
      "step": 44640
    },
    {
      "epoch": 7.378640776699029,
      "grad_norm": 13.382776260375977,
      "learning_rate": 1.4561413173212876e-05,
      "loss": 1.3977,
      "step": 44650
    },
    {
      "epoch": 7.3802933278248295,
      "grad_norm": 13.100677490234375,
      "learning_rate": 1.4552231954314255e-05,
      "loss": 1.3892,
      "step": 44660
    },
    {
      "epoch": 7.38194587895063,
      "grad_norm": 23.83889389038086,
      "learning_rate": 1.4543050735415634e-05,
      "loss": 1.3258,
      "step": 44670
    },
    {
      "epoch": 7.383598430076431,
      "grad_norm": 15.009169578552246,
      "learning_rate": 1.4533869516517012e-05,
      "loss": 1.3531,
      "step": 44680
    },
    {
      "epoch": 7.385250981202231,
      "grad_norm": 12.051997184753418,
      "learning_rate": 1.4524688297618391e-05,
      "loss": 1.3819,
      "step": 44690
    },
    {
      "epoch": 7.386903532328032,
      "grad_norm": 15.543947219848633,
      "learning_rate": 1.4515507078719772e-05,
      "loss": 1.3308,
      "step": 44700
    },
    {
      "epoch": 7.388556083453832,
      "grad_norm": 10.858241081237793,
      "learning_rate": 1.4506325859821151e-05,
      "loss": 1.32,
      "step": 44710
    },
    {
      "epoch": 7.390208634579633,
      "grad_norm": 11.575108528137207,
      "learning_rate": 1.449714464092253e-05,
      "loss": 1.3483,
      "step": 44720
    },
    {
      "epoch": 7.391861185705433,
      "grad_norm": 14.963944435119629,
      "learning_rate": 1.448796342202391e-05,
      "loss": 1.5092,
      "step": 44730
    },
    {
      "epoch": 7.393513736831233,
      "grad_norm": 24.254188537597656,
      "learning_rate": 1.4478782203125287e-05,
      "loss": 1.5216,
      "step": 44740
    },
    {
      "epoch": 7.395166287957034,
      "grad_norm": 11.076996803283691,
      "learning_rate": 1.4469600984226667e-05,
      "loss": 1.3176,
      "step": 44750
    },
    {
      "epoch": 7.396818839082834,
      "grad_norm": 17.357913970947266,
      "learning_rate": 1.4460419765328046e-05,
      "loss": 1.4716,
      "step": 44760
    },
    {
      "epoch": 7.398471390208635,
      "grad_norm": 15.097737312316895,
      "learning_rate": 1.4451238546429425e-05,
      "loss": 1.328,
      "step": 44770
    },
    {
      "epoch": 7.400123941334435,
      "grad_norm": 30.251018524169922,
      "learning_rate": 1.4442057327530803e-05,
      "loss": 1.3193,
      "step": 44780
    },
    {
      "epoch": 7.401776492460235,
      "grad_norm": 11.265031814575195,
      "learning_rate": 1.4432876108632182e-05,
      "loss": 1.3426,
      "step": 44790
    },
    {
      "epoch": 7.403429043586036,
      "grad_norm": 13.859474182128906,
      "learning_rate": 1.4423694889733561e-05,
      "loss": 1.4333,
      "step": 44800
    },
    {
      "epoch": 7.405081594711836,
      "grad_norm": 9.234789848327637,
      "learning_rate": 1.4414513670834942e-05,
      "loss": 1.3847,
      "step": 44810
    },
    {
      "epoch": 7.406734145837637,
      "grad_norm": 7.183691024780273,
      "learning_rate": 1.4405332451936321e-05,
      "loss": 1.3156,
      "step": 44820
    },
    {
      "epoch": 7.408386696963437,
      "grad_norm": 16.125627517700195,
      "learning_rate": 1.4396151233037699e-05,
      "loss": 1.4502,
      "step": 44830
    },
    {
      "epoch": 7.410039248089237,
      "grad_norm": 12.971248626708984,
      "learning_rate": 1.4386970014139078e-05,
      "loss": 1.4037,
      "step": 44840
    },
    {
      "epoch": 7.411691799215038,
      "grad_norm": 13.183744430541992,
      "learning_rate": 1.4377788795240457e-05,
      "loss": 1.3773,
      "step": 44850
    },
    {
      "epoch": 7.4133443503408385,
      "grad_norm": 12.956613540649414,
      "learning_rate": 1.4368607576341837e-05,
      "loss": 1.4324,
      "step": 44860
    },
    {
      "epoch": 7.4149969014666395,
      "grad_norm": 13.061296463012695,
      "learning_rate": 1.4359426357443214e-05,
      "loss": 1.3835,
      "step": 44870
    },
    {
      "epoch": 7.41664945259244,
      "grad_norm": 16.211750030517578,
      "learning_rate": 1.4350245138544593e-05,
      "loss": 1.2608,
      "step": 44880
    },
    {
      "epoch": 7.41830200371824,
      "grad_norm": 27.55082893371582,
      "learning_rate": 1.4341063919645973e-05,
      "loss": 1.3529,
      "step": 44890
    },
    {
      "epoch": 7.419954554844041,
      "grad_norm": 10.643492698669434,
      "learning_rate": 1.4331882700747352e-05,
      "loss": 1.3314,
      "step": 44900
    },
    {
      "epoch": 7.421607105969841,
      "grad_norm": 17.30533790588379,
      "learning_rate": 1.432270148184873e-05,
      "loss": 1.426,
      "step": 44910
    },
    {
      "epoch": 7.423259657095642,
      "grad_norm": 14.52156925201416,
      "learning_rate": 1.4313520262950109e-05,
      "loss": 1.3467,
      "step": 44920
    },
    {
      "epoch": 7.424912208221442,
      "grad_norm": 14.510608673095703,
      "learning_rate": 1.430433904405149e-05,
      "loss": 1.3018,
      "step": 44930
    },
    {
      "epoch": 7.426564759347242,
      "grad_norm": 15.366539001464844,
      "learning_rate": 1.4295157825152869e-05,
      "loss": 1.4342,
      "step": 44940
    },
    {
      "epoch": 7.428217310473043,
      "grad_norm": 66.76383972167969,
      "learning_rate": 1.4285976606254248e-05,
      "loss": 1.2281,
      "step": 44950
    },
    {
      "epoch": 7.429869861598843,
      "grad_norm": 12.815417289733887,
      "learning_rate": 1.4276795387355626e-05,
      "loss": 1.3599,
      "step": 44960
    },
    {
      "epoch": 7.431522412724644,
      "grad_norm": 11.356534004211426,
      "learning_rate": 1.4267614168457005e-05,
      "loss": 1.3074,
      "step": 44970
    },
    {
      "epoch": 7.433174963850444,
      "grad_norm": 12.682281494140625,
      "learning_rate": 1.4258432949558384e-05,
      "loss": 1.4034,
      "step": 44980
    },
    {
      "epoch": 7.434827514976244,
      "grad_norm": 217.0580596923828,
      "learning_rate": 1.4249251730659763e-05,
      "loss": 1.495,
      "step": 44990
    },
    {
      "epoch": 7.436480066102045,
      "grad_norm": 13.032721519470215,
      "learning_rate": 1.4240070511761141e-05,
      "loss": 1.373,
      "step": 45000
    },
    {
      "epoch": 7.438132617227845,
      "grad_norm": 13.059027671813965,
      "learning_rate": 1.423088929286252e-05,
      "loss": 1.4275,
      "step": 45010
    },
    {
      "epoch": 7.439785168353646,
      "grad_norm": 21.048809051513672,
      "learning_rate": 1.42217080739639e-05,
      "loss": 1.4316,
      "step": 45020
    },
    {
      "epoch": 7.441437719479446,
      "grad_norm": 15.526588439941406,
      "learning_rate": 1.4212526855065279e-05,
      "loss": 1.4637,
      "step": 45030
    },
    {
      "epoch": 7.443090270605246,
      "grad_norm": 12.340840339660645,
      "learning_rate": 1.420334563616666e-05,
      "loss": 1.4365,
      "step": 45040
    },
    {
      "epoch": 7.444742821731047,
      "grad_norm": 13.411223411560059,
      "learning_rate": 1.4194164417268039e-05,
      "loss": 1.4682,
      "step": 45050
    },
    {
      "epoch": 7.4463953728568475,
      "grad_norm": 9.226353645324707,
      "learning_rate": 1.4184983198369416e-05,
      "loss": 1.2974,
      "step": 45060
    },
    {
      "epoch": 7.4480479239826485,
      "grad_norm": 11.580143928527832,
      "learning_rate": 1.4175801979470796e-05,
      "loss": 1.517,
      "step": 45070
    },
    {
      "epoch": 7.449700475108449,
      "grad_norm": 20.38673973083496,
      "learning_rate": 1.4166620760572175e-05,
      "loss": 1.3698,
      "step": 45080
    },
    {
      "epoch": 7.4513530262342496,
      "grad_norm": 8.279924392700195,
      "learning_rate": 1.4157439541673554e-05,
      "loss": 1.5279,
      "step": 45090
    },
    {
      "epoch": 7.45300557736005,
      "grad_norm": 12.932108879089355,
      "learning_rate": 1.4148258322774932e-05,
      "loss": 1.5431,
      "step": 45100
    },
    {
      "epoch": 7.45465812848585,
      "grad_norm": 16.382911682128906,
      "learning_rate": 1.4139077103876311e-05,
      "loss": 1.3692,
      "step": 45110
    },
    {
      "epoch": 7.456310679611651,
      "grad_norm": 9.782100677490234,
      "learning_rate": 1.412989588497769e-05,
      "loss": 1.3877,
      "step": 45120
    },
    {
      "epoch": 7.457963230737451,
      "grad_norm": 13.853610038757324,
      "learning_rate": 1.4120714666079068e-05,
      "loss": 1.3601,
      "step": 45130
    },
    {
      "epoch": 7.459615781863252,
      "grad_norm": 15.717327117919922,
      "learning_rate": 1.4111533447180447e-05,
      "loss": 1.5425,
      "step": 45140
    },
    {
      "epoch": 7.461268332989052,
      "grad_norm": 11.447760581970215,
      "learning_rate": 1.4102352228281826e-05,
      "loss": 1.3657,
      "step": 45150
    },
    {
      "epoch": 7.462920884114852,
      "grad_norm": 15.681619644165039,
      "learning_rate": 1.4093171009383207e-05,
      "loss": 1.3663,
      "step": 45160
    },
    {
      "epoch": 7.464573435240653,
      "grad_norm": 11.067183494567871,
      "learning_rate": 1.4083989790484586e-05,
      "loss": 1.4642,
      "step": 45170
    },
    {
      "epoch": 7.466225986366453,
      "grad_norm": 21.230554580688477,
      "learning_rate": 1.4074808571585966e-05,
      "loss": 1.2956,
      "step": 45180
    },
    {
      "epoch": 7.467878537492254,
      "grad_norm": 12.558381080627441,
      "learning_rate": 1.4065627352687343e-05,
      "loss": 1.48,
      "step": 45190
    },
    {
      "epoch": 7.469531088618054,
      "grad_norm": 8.757758140563965,
      "learning_rate": 1.4056446133788723e-05,
      "loss": 1.1943,
      "step": 45200
    },
    {
      "epoch": 7.471183639743854,
      "grad_norm": 16.4716796875,
      "learning_rate": 1.4047264914890102e-05,
      "loss": 1.3091,
      "step": 45210
    },
    {
      "epoch": 7.472836190869655,
      "grad_norm": 22.595905303955078,
      "learning_rate": 1.4038083695991481e-05,
      "loss": 1.5014,
      "step": 45220
    },
    {
      "epoch": 7.474488741995455,
      "grad_norm": 12.184328079223633,
      "learning_rate": 1.4028902477092859e-05,
      "loss": 1.2568,
      "step": 45230
    },
    {
      "epoch": 7.476141293121256,
      "grad_norm": 14.307938575744629,
      "learning_rate": 1.4019721258194238e-05,
      "loss": 1.366,
      "step": 45240
    },
    {
      "epoch": 7.477793844247056,
      "grad_norm": 15.05473518371582,
      "learning_rate": 1.4010540039295617e-05,
      "loss": 1.3869,
      "step": 45250
    },
    {
      "epoch": 7.4794463953728565,
      "grad_norm": 12.070158958435059,
      "learning_rate": 1.4001358820396995e-05,
      "loss": 1.296,
      "step": 45260
    },
    {
      "epoch": 7.4810989464986575,
      "grad_norm": 23.16536521911621,
      "learning_rate": 1.3992177601498377e-05,
      "loss": 1.3242,
      "step": 45270
    },
    {
      "epoch": 7.482751497624458,
      "grad_norm": 11.528528213500977,
      "learning_rate": 1.3982996382599755e-05,
      "loss": 1.4554,
      "step": 45280
    },
    {
      "epoch": 7.4844040487502586,
      "grad_norm": 10.557455062866211,
      "learning_rate": 1.3973815163701134e-05,
      "loss": 1.3526,
      "step": 45290
    },
    {
      "epoch": 7.486056599876059,
      "grad_norm": 6.784778594970703,
      "learning_rate": 1.3964633944802513e-05,
      "loss": 1.3421,
      "step": 45300
    },
    {
      "epoch": 7.487709151001859,
      "grad_norm": 10.750931739807129,
      "learning_rate": 1.3955452725903893e-05,
      "loss": 1.305,
      "step": 45310
    },
    {
      "epoch": 7.48936170212766,
      "grad_norm": 16.1955623626709,
      "learning_rate": 1.394627150700527e-05,
      "loss": 1.4377,
      "step": 45320
    },
    {
      "epoch": 7.49101425325346,
      "grad_norm": 14.3265962600708,
      "learning_rate": 1.393709028810665e-05,
      "loss": 1.3428,
      "step": 45330
    },
    {
      "epoch": 7.492666804379261,
      "grad_norm": 31.921222686767578,
      "learning_rate": 1.3927909069208029e-05,
      "loss": 1.4143,
      "step": 45340
    },
    {
      "epoch": 7.494319355505061,
      "grad_norm": 20.369794845581055,
      "learning_rate": 1.3918727850309408e-05,
      "loss": 1.3892,
      "step": 45350
    },
    {
      "epoch": 7.495971906630861,
      "grad_norm": 14.980151176452637,
      "learning_rate": 1.3909546631410785e-05,
      "loss": 1.3714,
      "step": 45360
    },
    {
      "epoch": 7.497624457756662,
      "grad_norm": 21.37506675720215,
      "learning_rate": 1.3900365412512165e-05,
      "loss": 1.3442,
      "step": 45370
    },
    {
      "epoch": 7.499277008882462,
      "grad_norm": 9.15767765045166,
      "learning_rate": 1.3891184193613546e-05,
      "loss": 1.2651,
      "step": 45380
    },
    {
      "epoch": 7.500929560008263,
      "grad_norm": 19.919265747070312,
      "learning_rate": 1.3882002974714925e-05,
      "loss": 1.4399,
      "step": 45390
    },
    {
      "epoch": 7.502582111134063,
      "grad_norm": 18.08030891418457,
      "learning_rate": 1.3872821755816304e-05,
      "loss": 1.2924,
      "step": 45400
    },
    {
      "epoch": 7.504234662259863,
      "grad_norm": 14.912238121032715,
      "learning_rate": 1.3863640536917683e-05,
      "loss": 1.356,
      "step": 45410
    },
    {
      "epoch": 7.505887213385664,
      "grad_norm": 18.654098510742188,
      "learning_rate": 1.3854459318019061e-05,
      "loss": 1.3739,
      "step": 45420
    },
    {
      "epoch": 7.507539764511464,
      "grad_norm": 37.90325164794922,
      "learning_rate": 1.384527809912044e-05,
      "loss": 1.4239,
      "step": 45430
    },
    {
      "epoch": 7.509192315637265,
      "grad_norm": 10.297859191894531,
      "learning_rate": 1.383609688022182e-05,
      "loss": 1.3303,
      "step": 45440
    },
    {
      "epoch": 7.510844866763065,
      "grad_norm": 16.992605209350586,
      "learning_rate": 1.3826915661323197e-05,
      "loss": 1.4006,
      "step": 45450
    },
    {
      "epoch": 7.512497417888866,
      "grad_norm": 11.61043930053711,
      "learning_rate": 1.3817734442424576e-05,
      "loss": 1.3269,
      "step": 45460
    },
    {
      "epoch": 7.5141499690146665,
      "grad_norm": 11.616755485534668,
      "learning_rate": 1.3808553223525955e-05,
      "loss": 1.2896,
      "step": 45470
    },
    {
      "epoch": 7.515802520140467,
      "grad_norm": 13.678464889526367,
      "learning_rate": 1.3799372004627335e-05,
      "loss": 1.4308,
      "step": 45480
    },
    {
      "epoch": 7.5174550712662676,
      "grad_norm": 15.36746597290039,
      "learning_rate": 1.3790190785728712e-05,
      "loss": 1.2909,
      "step": 45490
    },
    {
      "epoch": 7.519107622392068,
      "grad_norm": 17.820960998535156,
      "learning_rate": 1.3781009566830095e-05,
      "loss": 1.273,
      "step": 45500
    },
    {
      "epoch": 7.520760173517868,
      "grad_norm": 34.905059814453125,
      "learning_rate": 1.3771828347931472e-05,
      "loss": 1.3707,
      "step": 45510
    },
    {
      "epoch": 7.522412724643669,
      "grad_norm": 9.051143646240234,
      "learning_rate": 1.3762647129032852e-05,
      "loss": 1.3682,
      "step": 45520
    },
    {
      "epoch": 7.524065275769469,
      "grad_norm": 12.66955852508545,
      "learning_rate": 1.3753465910134231e-05,
      "loss": 1.361,
      "step": 45530
    },
    {
      "epoch": 7.52571782689527,
      "grad_norm": 12.091620445251465,
      "learning_rate": 1.374428469123561e-05,
      "loss": 1.3856,
      "step": 45540
    },
    {
      "epoch": 7.52737037802107,
      "grad_norm": 9.986820220947266,
      "learning_rate": 1.3735103472336988e-05,
      "loss": 1.362,
      "step": 45550
    },
    {
      "epoch": 7.529022929146871,
      "grad_norm": 40.41029357910156,
      "learning_rate": 1.3725922253438367e-05,
      "loss": 1.4566,
      "step": 45560
    },
    {
      "epoch": 7.530675480272671,
      "grad_norm": 51.40917205810547,
      "learning_rate": 1.3716741034539746e-05,
      "loss": 1.4305,
      "step": 45570
    },
    {
      "epoch": 7.532328031398471,
      "grad_norm": 12.97027587890625,
      "learning_rate": 1.3707559815641124e-05,
      "loss": 1.444,
      "step": 45580
    },
    {
      "epoch": 7.533980582524272,
      "grad_norm": 10.593807220458984,
      "learning_rate": 1.3698378596742503e-05,
      "loss": 1.4671,
      "step": 45590
    },
    {
      "epoch": 7.535633133650072,
      "grad_norm": 14.09102725982666,
      "learning_rate": 1.3689197377843882e-05,
      "loss": 1.2288,
      "step": 45600
    },
    {
      "epoch": 7.537285684775873,
      "grad_norm": 13.20085620880127,
      "learning_rate": 1.3680016158945263e-05,
      "loss": 1.5422,
      "step": 45610
    },
    {
      "epoch": 7.538938235901673,
      "grad_norm": 13.723567008972168,
      "learning_rate": 1.3670834940046642e-05,
      "loss": 1.3606,
      "step": 45620
    },
    {
      "epoch": 7.540590787027473,
      "grad_norm": 12.14212703704834,
      "learning_rate": 1.3661653721148022e-05,
      "loss": 1.3494,
      "step": 45630
    },
    {
      "epoch": 7.542243338153274,
      "grad_norm": 18.38138771057129,
      "learning_rate": 1.36524725022494e-05,
      "loss": 1.3763,
      "step": 45640
    },
    {
      "epoch": 7.543895889279074,
      "grad_norm": 9.355786323547363,
      "learning_rate": 1.3643291283350778e-05,
      "loss": 1.4087,
      "step": 45650
    },
    {
      "epoch": 7.545548440404875,
      "grad_norm": 17.49809455871582,
      "learning_rate": 1.3634110064452158e-05,
      "loss": 1.3953,
      "step": 45660
    },
    {
      "epoch": 7.5472009915306755,
      "grad_norm": 29.25096893310547,
      "learning_rate": 1.3624928845553537e-05,
      "loss": 1.3762,
      "step": 45670
    },
    {
      "epoch": 7.548853542656476,
      "grad_norm": 21.453706741333008,
      "learning_rate": 1.3615747626654915e-05,
      "loss": 1.4828,
      "step": 45680
    },
    {
      "epoch": 7.5505060937822766,
      "grad_norm": 9.639589309692383,
      "learning_rate": 1.3606566407756294e-05,
      "loss": 1.4189,
      "step": 45690
    },
    {
      "epoch": 7.552158644908077,
      "grad_norm": 18.783720016479492,
      "learning_rate": 1.3597385188857673e-05,
      "loss": 1.3584,
      "step": 45700
    },
    {
      "epoch": 7.553811196033878,
      "grad_norm": 10.537060737609863,
      "learning_rate": 1.358820396995905e-05,
      "loss": 1.3764,
      "step": 45710
    },
    {
      "epoch": 7.555463747159678,
      "grad_norm": 17.940895080566406,
      "learning_rate": 1.357902275106043e-05,
      "loss": 1.5005,
      "step": 45720
    },
    {
      "epoch": 7.557116298285479,
      "grad_norm": 13.430806159973145,
      "learning_rate": 1.3569841532161812e-05,
      "loss": 1.4229,
      "step": 45730
    },
    {
      "epoch": 7.558768849411279,
      "grad_norm": 37.275325775146484,
      "learning_rate": 1.356066031326319e-05,
      "loss": 1.332,
      "step": 45740
    },
    {
      "epoch": 7.560421400537079,
      "grad_norm": 14.86922550201416,
      "learning_rate": 1.355147909436457e-05,
      "loss": 1.3686,
      "step": 45750
    },
    {
      "epoch": 7.56207395166288,
      "grad_norm": 87.05618286132812,
      "learning_rate": 1.3542297875465949e-05,
      "loss": 1.4918,
      "step": 45760
    },
    {
      "epoch": 7.56372650278868,
      "grad_norm": 11.52795696258545,
      "learning_rate": 1.3533116656567326e-05,
      "loss": 1.3878,
      "step": 45770
    },
    {
      "epoch": 7.56537905391448,
      "grad_norm": 13.127462387084961,
      "learning_rate": 1.3523935437668705e-05,
      "loss": 1.3601,
      "step": 45780
    },
    {
      "epoch": 7.567031605040281,
      "grad_norm": 14.65783977508545,
      "learning_rate": 1.3514754218770085e-05,
      "loss": 1.3726,
      "step": 45790
    },
    {
      "epoch": 7.568684156166081,
      "grad_norm": 12.145511627197266,
      "learning_rate": 1.3505572999871464e-05,
      "loss": 1.4446,
      "step": 45800
    },
    {
      "epoch": 7.570336707291882,
      "grad_norm": 20.19838523864746,
      "learning_rate": 1.3496391780972841e-05,
      "loss": 1.472,
      "step": 45810
    },
    {
      "epoch": 7.571989258417682,
      "grad_norm": 8.90953254699707,
      "learning_rate": 1.348721056207422e-05,
      "loss": 1.3158,
      "step": 45820
    },
    {
      "epoch": 7.573641809543483,
      "grad_norm": 14.088125228881836,
      "learning_rate": 1.34780293431756e-05,
      "loss": 1.2182,
      "step": 45830
    },
    {
      "epoch": 7.575294360669283,
      "grad_norm": 14.895604133605957,
      "learning_rate": 1.346884812427698e-05,
      "loss": 1.501,
      "step": 45840
    },
    {
      "epoch": 7.576946911795083,
      "grad_norm": 11.624994277954102,
      "learning_rate": 1.345966690537836e-05,
      "loss": 1.4957,
      "step": 45850
    },
    {
      "epoch": 7.578599462920884,
      "grad_norm": 19.928787231445312,
      "learning_rate": 1.345048568647974e-05,
      "loss": 1.2905,
      "step": 45860
    },
    {
      "epoch": 7.5802520140466845,
      "grad_norm": 12.225404739379883,
      "learning_rate": 1.3441304467581117e-05,
      "loss": 1.5096,
      "step": 45870
    },
    {
      "epoch": 7.581904565172485,
      "grad_norm": 12.218575477600098,
      "learning_rate": 1.3432123248682496e-05,
      "loss": 1.289,
      "step": 45880
    },
    {
      "epoch": 7.5835571162982855,
      "grad_norm": 17.526824951171875,
      "learning_rate": 1.3422942029783875e-05,
      "loss": 1.4192,
      "step": 45890
    },
    {
      "epoch": 7.585209667424086,
      "grad_norm": 13.405013084411621,
      "learning_rate": 1.3413760810885253e-05,
      "loss": 1.4625,
      "step": 45900
    },
    {
      "epoch": 7.586862218549887,
      "grad_norm": 10.898892402648926,
      "learning_rate": 1.3404579591986632e-05,
      "loss": 1.3915,
      "step": 45910
    },
    {
      "epoch": 7.588514769675687,
      "grad_norm": 17.272350311279297,
      "learning_rate": 1.3395398373088011e-05,
      "loss": 1.409,
      "step": 45920
    },
    {
      "epoch": 7.590167320801488,
      "grad_norm": 13.411367416381836,
      "learning_rate": 1.338621715418939e-05,
      "loss": 1.4557,
      "step": 45930
    },
    {
      "epoch": 7.591819871927288,
      "grad_norm": 33.857017517089844,
      "learning_rate": 1.3377035935290768e-05,
      "loss": 1.2233,
      "step": 45940
    },
    {
      "epoch": 7.593472423053088,
      "grad_norm": 18.17672348022461,
      "learning_rate": 1.3367854716392147e-05,
      "loss": 1.3755,
      "step": 45950
    },
    {
      "epoch": 7.595124974178889,
      "grad_norm": 11.827006340026855,
      "learning_rate": 1.3358673497493528e-05,
      "loss": 1.3968,
      "step": 45960
    },
    {
      "epoch": 7.596777525304689,
      "grad_norm": 10.973575592041016,
      "learning_rate": 1.3349492278594908e-05,
      "loss": 1.4225,
      "step": 45970
    },
    {
      "epoch": 7.59843007643049,
      "grad_norm": 13.734282493591309,
      "learning_rate": 1.3340311059696287e-05,
      "loss": 1.3545,
      "step": 45980
    },
    {
      "epoch": 7.60008262755629,
      "grad_norm": 20.4477596282959,
      "learning_rate": 1.3331129840797666e-05,
      "loss": 1.4661,
      "step": 45990
    },
    {
      "epoch": 7.60173517868209,
      "grad_norm": 11.852956771850586,
      "learning_rate": 1.3321948621899044e-05,
      "loss": 1.28,
      "step": 46000
    },
    {
      "epoch": 7.603387729807891,
      "grad_norm": 11.33363151550293,
      "learning_rate": 1.3312767403000423e-05,
      "loss": 1.4586,
      "step": 46010
    },
    {
      "epoch": 7.605040280933691,
      "grad_norm": 10.724862098693848,
      "learning_rate": 1.3303586184101802e-05,
      "loss": 1.5499,
      "step": 46020
    },
    {
      "epoch": 7.606692832059492,
      "grad_norm": 12.029391288757324,
      "learning_rate": 1.329440496520318e-05,
      "loss": 1.3334,
      "step": 46030
    },
    {
      "epoch": 7.608345383185292,
      "grad_norm": 27.66431999206543,
      "learning_rate": 1.3285223746304559e-05,
      "loss": 1.4114,
      "step": 46040
    },
    {
      "epoch": 7.609997934311092,
      "grad_norm": 11.550406455993652,
      "learning_rate": 1.3276042527405938e-05,
      "loss": 1.3458,
      "step": 46050
    },
    {
      "epoch": 7.611650485436893,
      "grad_norm": 19.206697463989258,
      "learning_rate": 1.3266861308507317e-05,
      "loss": 1.4311,
      "step": 46060
    },
    {
      "epoch": 7.6133030365626935,
      "grad_norm": 13.49929428100586,
      "learning_rate": 1.3257680089608698e-05,
      "loss": 1.4536,
      "step": 46070
    },
    {
      "epoch": 7.6149555876884945,
      "grad_norm": 12.506552696228027,
      "learning_rate": 1.3248498870710078e-05,
      "loss": 1.4649,
      "step": 46080
    },
    {
      "epoch": 7.6166081388142945,
      "grad_norm": 16.31134605407715,
      "learning_rate": 1.3239317651811455e-05,
      "loss": 1.4549,
      "step": 46090
    },
    {
      "epoch": 7.618260689940095,
      "grad_norm": 9.935333251953125,
      "learning_rate": 1.3230136432912834e-05,
      "loss": 1.387,
      "step": 46100
    },
    {
      "epoch": 7.619913241065896,
      "grad_norm": 14.679750442504883,
      "learning_rate": 1.3220955214014214e-05,
      "loss": 1.4526,
      "step": 46110
    },
    {
      "epoch": 7.621565792191696,
      "grad_norm": 11.663803100585938,
      "learning_rate": 1.3211773995115593e-05,
      "loss": 1.4375,
      "step": 46120
    },
    {
      "epoch": 7.623218343317497,
      "grad_norm": 333.21502685546875,
      "learning_rate": 1.320259277621697e-05,
      "loss": 1.5769,
      "step": 46130
    },
    {
      "epoch": 7.624870894443297,
      "grad_norm": 17.140153884887695,
      "learning_rate": 1.319341155731835e-05,
      "loss": 1.3195,
      "step": 46140
    },
    {
      "epoch": 7.626523445569097,
      "grad_norm": 7.242269039154053,
      "learning_rate": 1.3184230338419729e-05,
      "loss": 1.4281,
      "step": 46150
    },
    {
      "epoch": 7.628175996694898,
      "grad_norm": 13.725931167602539,
      "learning_rate": 1.3175049119521107e-05,
      "loss": 1.3323,
      "step": 46160
    },
    {
      "epoch": 7.629828547820698,
      "grad_norm": 14.671615600585938,
      "learning_rate": 1.3165867900622486e-05,
      "loss": 1.4224,
      "step": 46170
    },
    {
      "epoch": 7.631481098946499,
      "grad_norm": 12.745248794555664,
      "learning_rate": 1.3156686681723868e-05,
      "loss": 1.3534,
      "step": 46180
    },
    {
      "epoch": 7.633133650072299,
      "grad_norm": 24.55314064025879,
      "learning_rate": 1.3147505462825246e-05,
      "loss": 1.2832,
      "step": 46190
    },
    {
      "epoch": 7.6347862011981,
      "grad_norm": 17.937116622924805,
      "learning_rate": 1.3138324243926625e-05,
      "loss": 1.2754,
      "step": 46200
    },
    {
      "epoch": 7.6364387523239,
      "grad_norm": 26.932025909423828,
      "learning_rate": 1.3129143025028005e-05,
      "loss": 1.3059,
      "step": 46210
    },
    {
      "epoch": 7.6380913034497,
      "grad_norm": 11.607211112976074,
      "learning_rate": 1.3119961806129382e-05,
      "loss": 1.4191,
      "step": 46220
    },
    {
      "epoch": 7.639743854575501,
      "grad_norm": 30.498897552490234,
      "learning_rate": 1.3110780587230761e-05,
      "loss": 1.6439,
      "step": 46230
    },
    {
      "epoch": 7.641396405701301,
      "grad_norm": 9.308640480041504,
      "learning_rate": 1.310159936833214e-05,
      "loss": 1.3958,
      "step": 46240
    },
    {
      "epoch": 7.643048956827101,
      "grad_norm": 12.674652099609375,
      "learning_rate": 1.309241814943352e-05,
      "loss": 1.3947,
      "step": 46250
    },
    {
      "epoch": 7.644701507952902,
      "grad_norm": 18.442964553833008,
      "learning_rate": 1.3083236930534897e-05,
      "loss": 1.3829,
      "step": 46260
    },
    {
      "epoch": 7.6463540590787025,
      "grad_norm": 9.009620666503906,
      "learning_rate": 1.3074055711636277e-05,
      "loss": 1.382,
      "step": 46270
    },
    {
      "epoch": 7.6480066102045035,
      "grad_norm": 11.820587158203125,
      "learning_rate": 1.3064874492737656e-05,
      "loss": 1.2625,
      "step": 46280
    },
    {
      "epoch": 7.6496591613303035,
      "grad_norm": 13.428104400634766,
      "learning_rate": 1.3055693273839033e-05,
      "loss": 1.2081,
      "step": 46290
    },
    {
      "epoch": 7.6513117124561045,
      "grad_norm": 226.61851501464844,
      "learning_rate": 1.3046512054940416e-05,
      "loss": 1.3743,
      "step": 46300
    },
    {
      "epoch": 7.652964263581905,
      "grad_norm": 19.351016998291016,
      "learning_rate": 1.3037330836041795e-05,
      "loss": 1.3831,
      "step": 46310
    },
    {
      "epoch": 7.654616814707705,
      "grad_norm": 9.168930053710938,
      "learning_rate": 1.3028149617143173e-05,
      "loss": 1.3555,
      "step": 46320
    },
    {
      "epoch": 7.656269365833506,
      "grad_norm": 17.819650650024414,
      "learning_rate": 1.3018968398244552e-05,
      "loss": 1.4432,
      "step": 46330
    },
    {
      "epoch": 7.657921916959306,
      "grad_norm": 57.033180236816406,
      "learning_rate": 1.3009787179345931e-05,
      "loss": 1.3021,
      "step": 46340
    },
    {
      "epoch": 7.659574468085106,
      "grad_norm": 25.294410705566406,
      "learning_rate": 1.3000605960447309e-05,
      "loss": 1.4219,
      "step": 46350
    },
    {
      "epoch": 7.661227019210907,
      "grad_norm": 22.175817489624023,
      "learning_rate": 1.2991424741548688e-05,
      "loss": 1.4499,
      "step": 46360
    },
    {
      "epoch": 7.662879570336707,
      "grad_norm": 12.002792358398438,
      "learning_rate": 1.2982243522650067e-05,
      "loss": 1.3654,
      "step": 46370
    },
    {
      "epoch": 7.664532121462508,
      "grad_norm": 9.988709449768066,
      "learning_rate": 1.2973062303751447e-05,
      "loss": 1.4325,
      "step": 46380
    },
    {
      "epoch": 7.666184672588308,
      "grad_norm": 14.610112190246582,
      "learning_rate": 1.2963881084852824e-05,
      "loss": 1.2868,
      "step": 46390
    },
    {
      "epoch": 7.667837223714109,
      "grad_norm": 14.141910552978516,
      "learning_rate": 1.2954699865954203e-05,
      "loss": 1.3978,
      "step": 46400
    },
    {
      "epoch": 7.669489774839909,
      "grad_norm": 24.699037551879883,
      "learning_rate": 1.2945518647055584e-05,
      "loss": 1.3868,
      "step": 46410
    },
    {
      "epoch": 7.671142325965709,
      "grad_norm": 13.397317886352539,
      "learning_rate": 1.2936337428156964e-05,
      "loss": 1.329,
      "step": 46420
    },
    {
      "epoch": 7.67279487709151,
      "grad_norm": 13.475739479064941,
      "learning_rate": 1.2927156209258343e-05,
      "loss": 1.4917,
      "step": 46430
    },
    {
      "epoch": 7.67444742821731,
      "grad_norm": 14.52394962310791,
      "learning_rate": 1.2917974990359722e-05,
      "loss": 1.3478,
      "step": 46440
    },
    {
      "epoch": 7.676099979343111,
      "grad_norm": 20.19123077392578,
      "learning_rate": 1.29087937714611e-05,
      "loss": 1.5876,
      "step": 46450
    },
    {
      "epoch": 7.677752530468911,
      "grad_norm": 10.427851676940918,
      "learning_rate": 1.2899612552562479e-05,
      "loss": 1.3178,
      "step": 46460
    },
    {
      "epoch": 7.6794050815947115,
      "grad_norm": 15.901928901672363,
      "learning_rate": 1.2890431333663858e-05,
      "loss": 1.4723,
      "step": 46470
    },
    {
      "epoch": 7.6810576327205125,
      "grad_norm": 16.793914794921875,
      "learning_rate": 1.2881250114765236e-05,
      "loss": 1.3749,
      "step": 46480
    },
    {
      "epoch": 7.6827101838463125,
      "grad_norm": 9.323737144470215,
      "learning_rate": 1.2872068895866615e-05,
      "loss": 1.3149,
      "step": 46490
    },
    {
      "epoch": 7.6843627349721135,
      "grad_norm": 13.843609809875488,
      "learning_rate": 1.2862887676967994e-05,
      "loss": 1.4424,
      "step": 46500
    },
    {
      "epoch": 7.686015286097914,
      "grad_norm": 16.021211624145508,
      "learning_rate": 1.2853706458069373e-05,
      "loss": 1.3349,
      "step": 46510
    },
    {
      "epoch": 7.687667837223714,
      "grad_norm": 12.02914810180664,
      "learning_rate": 1.2844525239170751e-05,
      "loss": 1.3076,
      "step": 46520
    },
    {
      "epoch": 7.689320388349515,
      "grad_norm": 9.748140335083008,
      "learning_rate": 1.2835344020272134e-05,
      "loss": 1.3444,
      "step": 46530
    },
    {
      "epoch": 7.690972939475315,
      "grad_norm": 17.1309757232666,
      "learning_rate": 1.2826162801373511e-05,
      "loss": 1.3255,
      "step": 46540
    },
    {
      "epoch": 7.692625490601116,
      "grad_norm": 10.874885559082031,
      "learning_rate": 1.281698158247489e-05,
      "loss": 1.3793,
      "step": 46550
    },
    {
      "epoch": 7.694278041726916,
      "grad_norm": 13.18276596069336,
      "learning_rate": 1.280780036357627e-05,
      "loss": 1.399,
      "step": 46560
    },
    {
      "epoch": 7.695930592852717,
      "grad_norm": 12.042601585388184,
      "learning_rate": 1.2798619144677649e-05,
      "loss": 1.3615,
      "step": 46570
    },
    {
      "epoch": 7.697583143978517,
      "grad_norm": 25.975971221923828,
      "learning_rate": 1.2789437925779027e-05,
      "loss": 1.5137,
      "step": 46580
    },
    {
      "epoch": 7.699235695104317,
      "grad_norm": 16.21059226989746,
      "learning_rate": 1.2780256706880406e-05,
      "loss": 1.3351,
      "step": 46590
    },
    {
      "epoch": 7.700888246230118,
      "grad_norm": 13.045463562011719,
      "learning_rate": 1.2771075487981785e-05,
      "loss": 1.3001,
      "step": 46600
    },
    {
      "epoch": 7.702540797355918,
      "grad_norm": 11.22043228149414,
      "learning_rate": 1.2761894269083163e-05,
      "loss": 1.4261,
      "step": 46610
    },
    {
      "epoch": 7.704193348481718,
      "grad_norm": 36.950538635253906,
      "learning_rate": 1.2752713050184542e-05,
      "loss": 1.3339,
      "step": 46620
    },
    {
      "epoch": 7.705845899607519,
      "grad_norm": 9.229752540588379,
      "learning_rate": 1.2743531831285921e-05,
      "loss": 1.3431,
      "step": 46630
    },
    {
      "epoch": 7.707498450733319,
      "grad_norm": 11.805033683776855,
      "learning_rate": 1.2734350612387302e-05,
      "loss": 1.3491,
      "step": 46640
    },
    {
      "epoch": 7.70915100185912,
      "grad_norm": 23.916683197021484,
      "learning_rate": 1.2725169393488681e-05,
      "loss": 1.3354,
      "step": 46650
    },
    {
      "epoch": 7.71080355298492,
      "grad_norm": 15.571776390075684,
      "learning_rate": 1.271598817459006e-05,
      "loss": 1.4053,
      "step": 46660
    },
    {
      "epoch": 7.712456104110721,
      "grad_norm": 16.37409782409668,
      "learning_rate": 1.2706806955691438e-05,
      "loss": 1.2922,
      "step": 46670
    },
    {
      "epoch": 7.7141086552365215,
      "grad_norm": 10.659138679504395,
      "learning_rate": 1.2697625736792817e-05,
      "loss": 1.3695,
      "step": 46680
    },
    {
      "epoch": 7.7157612063623215,
      "grad_norm": 20.0853328704834,
      "learning_rate": 1.2688444517894197e-05,
      "loss": 1.4891,
      "step": 46690
    },
    {
      "epoch": 7.7174137574881225,
      "grad_norm": 16.797853469848633,
      "learning_rate": 1.2679263298995576e-05,
      "loss": 1.4373,
      "step": 46700
    },
    {
      "epoch": 7.719066308613923,
      "grad_norm": 14.921588897705078,
      "learning_rate": 1.2670082080096953e-05,
      "loss": 1.4576,
      "step": 46710
    },
    {
      "epoch": 7.720718859739723,
      "grad_norm": 9.122919082641602,
      "learning_rate": 1.2660900861198333e-05,
      "loss": 1.3505,
      "step": 46720
    },
    {
      "epoch": 7.722371410865524,
      "grad_norm": 18.881772994995117,
      "learning_rate": 1.2651719642299712e-05,
      "loss": 1.5416,
      "step": 46730
    },
    {
      "epoch": 7.724023961991324,
      "grad_norm": 9.798796653747559,
      "learning_rate": 1.264253842340109e-05,
      "loss": 1.3828,
      "step": 46740
    },
    {
      "epoch": 7.725676513117125,
      "grad_norm": 21.321590423583984,
      "learning_rate": 1.2633357204502472e-05,
      "loss": 1.4042,
      "step": 46750
    },
    {
      "epoch": 7.727329064242925,
      "grad_norm": 16.886615753173828,
      "learning_rate": 1.2624175985603851e-05,
      "loss": 1.3997,
      "step": 46760
    },
    {
      "epoch": 7.728981615368726,
      "grad_norm": 16.772817611694336,
      "learning_rate": 1.2614994766705229e-05,
      "loss": 1.4211,
      "step": 46770
    },
    {
      "epoch": 7.730634166494526,
      "grad_norm": 19.1043758392334,
      "learning_rate": 1.2605813547806608e-05,
      "loss": 1.3403,
      "step": 46780
    },
    {
      "epoch": 7.732286717620326,
      "grad_norm": 17.032512664794922,
      "learning_rate": 1.2596632328907987e-05,
      "loss": 1.3278,
      "step": 46790
    },
    {
      "epoch": 7.733939268746127,
      "grad_norm": 9.3512544631958,
      "learning_rate": 1.2587451110009365e-05,
      "loss": 1.4196,
      "step": 46800
    },
    {
      "epoch": 7.735591819871927,
      "grad_norm": 14.732444763183594,
      "learning_rate": 1.2578269891110744e-05,
      "loss": 1.4069,
      "step": 46810
    },
    {
      "epoch": 7.737244370997728,
      "grad_norm": 19.343429565429688,
      "learning_rate": 1.2569088672212123e-05,
      "loss": 1.3892,
      "step": 46820
    },
    {
      "epoch": 7.738896922123528,
      "grad_norm": 15.403067588806152,
      "learning_rate": 1.2559907453313503e-05,
      "loss": 1.433,
      "step": 46830
    },
    {
      "epoch": 7.740549473249328,
      "grad_norm": 14.680621147155762,
      "learning_rate": 1.255072623441488e-05,
      "loss": 1.5072,
      "step": 46840
    },
    {
      "epoch": 7.742202024375129,
      "grad_norm": 21.64932632446289,
      "learning_rate": 1.254154501551626e-05,
      "loss": 1.4388,
      "step": 46850
    },
    {
      "epoch": 7.743854575500929,
      "grad_norm": 12.506025314331055,
      "learning_rate": 1.2532363796617639e-05,
      "loss": 1.3015,
      "step": 46860
    },
    {
      "epoch": 7.74550712662673,
      "grad_norm": 18.126253128051758,
      "learning_rate": 1.252318257771902e-05,
      "loss": 1.2921,
      "step": 46870
    },
    {
      "epoch": 7.7471596777525304,
      "grad_norm": 8.994416236877441,
      "learning_rate": 1.2514001358820399e-05,
      "loss": 1.3792,
      "step": 46880
    },
    {
      "epoch": 7.7488122288783305,
      "grad_norm": 14.529839515686035,
      "learning_rate": 1.2504820139921778e-05,
      "loss": 1.3501,
      "step": 46890
    },
    {
      "epoch": 7.7504647800041315,
      "grad_norm": 8.697839736938477,
      "learning_rate": 1.2495638921023156e-05,
      "loss": 1.3575,
      "step": 46900
    },
    {
      "epoch": 7.752117331129932,
      "grad_norm": 14.368610382080078,
      "learning_rate": 1.2486457702124535e-05,
      "loss": 1.3781,
      "step": 46910
    },
    {
      "epoch": 7.753769882255733,
      "grad_norm": 11.302849769592285,
      "learning_rate": 1.2477276483225914e-05,
      "loss": 1.4043,
      "step": 46920
    },
    {
      "epoch": 7.755422433381533,
      "grad_norm": 13.063495635986328,
      "learning_rate": 1.2468095264327292e-05,
      "loss": 1.3099,
      "step": 46930
    },
    {
      "epoch": 7.757074984507334,
      "grad_norm": 11.363663673400879,
      "learning_rate": 1.2458914045428671e-05,
      "loss": 1.2627,
      "step": 46940
    },
    {
      "epoch": 7.758727535633134,
      "grad_norm": 16.601898193359375,
      "learning_rate": 1.2449732826530052e-05,
      "loss": 1.2975,
      "step": 46950
    },
    {
      "epoch": 7.760380086758934,
      "grad_norm": 15.275944709777832,
      "learning_rate": 1.244055160763143e-05,
      "loss": 1.3798,
      "step": 46960
    },
    {
      "epoch": 7.762032637884735,
      "grad_norm": 15.088699340820312,
      "learning_rate": 1.2431370388732809e-05,
      "loss": 1.4236,
      "step": 46970
    },
    {
      "epoch": 7.763685189010535,
      "grad_norm": 49.174373626708984,
      "learning_rate": 1.2422189169834188e-05,
      "loss": 1.446,
      "step": 46980
    },
    {
      "epoch": 7.765337740136335,
      "grad_norm": 10.12486457824707,
      "learning_rate": 1.2413007950935567e-05,
      "loss": 1.3574,
      "step": 46990
    },
    {
      "epoch": 7.766990291262136,
      "grad_norm": 13.83552074432373,
      "learning_rate": 1.2403826732036945e-05,
      "loss": 1.3296,
      "step": 47000
    },
    {
      "epoch": 7.768642842387936,
      "grad_norm": 7.309926986694336,
      "learning_rate": 1.2394645513138326e-05,
      "loss": 1.2785,
      "step": 47010
    },
    {
      "epoch": 7.770295393513737,
      "grad_norm": 11.483392715454102,
      "learning_rate": 1.2385464294239705e-05,
      "loss": 1.3437,
      "step": 47020
    },
    {
      "epoch": 7.771947944639537,
      "grad_norm": 14.15250301361084,
      "learning_rate": 1.2376283075341082e-05,
      "loss": 1.1891,
      "step": 47030
    },
    {
      "epoch": 7.773600495765338,
      "grad_norm": 8.438793182373047,
      "learning_rate": 1.2367101856442462e-05,
      "loss": 1.2479,
      "step": 47040
    },
    {
      "epoch": 7.775253046891138,
      "grad_norm": 11.011880874633789,
      "learning_rate": 1.2357920637543841e-05,
      "loss": 1.2589,
      "step": 47050
    },
    {
      "epoch": 7.776905598016938,
      "grad_norm": 12.733864784240723,
      "learning_rate": 1.2348739418645219e-05,
      "loss": 1.3427,
      "step": 47060
    },
    {
      "epoch": 7.778558149142739,
      "grad_norm": 11.575126647949219,
      "learning_rate": 1.23395581997466e-05,
      "loss": 1.3099,
      "step": 47070
    },
    {
      "epoch": 7.7802107002685394,
      "grad_norm": 9.306955337524414,
      "learning_rate": 1.2330376980847979e-05,
      "loss": 1.3208,
      "step": 47080
    },
    {
      "epoch": 7.7818632513943395,
      "grad_norm": 21.35601234436035,
      "learning_rate": 1.2321195761949356e-05,
      "loss": 1.4267,
      "step": 47090
    },
    {
      "epoch": 7.7835158025201405,
      "grad_norm": 9.708170890808105,
      "learning_rate": 1.2312014543050736e-05,
      "loss": 1.3281,
      "step": 47100
    },
    {
      "epoch": 7.785168353645941,
      "grad_norm": 11.418980598449707,
      "learning_rate": 1.2302833324152115e-05,
      "loss": 1.4658,
      "step": 47110
    },
    {
      "epoch": 7.786820904771742,
      "grad_norm": 33.159698486328125,
      "learning_rate": 1.2293652105253494e-05,
      "loss": 1.4893,
      "step": 47120
    },
    {
      "epoch": 7.788473455897542,
      "grad_norm": 16.78141212463379,
      "learning_rate": 1.2284470886354873e-05,
      "loss": 1.5537,
      "step": 47130
    },
    {
      "epoch": 7.790126007023343,
      "grad_norm": 8.051682472229004,
      "learning_rate": 1.2275289667456253e-05,
      "loss": 1.3073,
      "step": 47140
    },
    {
      "epoch": 7.791778558149143,
      "grad_norm": 10.035185813903809,
      "learning_rate": 1.2266108448557632e-05,
      "loss": 1.3287,
      "step": 47150
    },
    {
      "epoch": 7.793431109274943,
      "grad_norm": 16.550947189331055,
      "learning_rate": 1.225692722965901e-05,
      "loss": 1.3027,
      "step": 47160
    },
    {
      "epoch": 7.795083660400744,
      "grad_norm": 13.783156394958496,
      "learning_rate": 1.2247746010760389e-05,
      "loss": 1.5014,
      "step": 47170
    },
    {
      "epoch": 7.796736211526544,
      "grad_norm": 11.391876220703125,
      "learning_rate": 1.223856479186177e-05,
      "loss": 1.3918,
      "step": 47180
    },
    {
      "epoch": 7.798388762652345,
      "grad_norm": 16.179780960083008,
      "learning_rate": 1.2229383572963147e-05,
      "loss": 1.3255,
      "step": 47190
    },
    {
      "epoch": 7.800041313778145,
      "grad_norm": 20.307268142700195,
      "learning_rate": 1.2220202354064526e-05,
      "loss": 1.3893,
      "step": 47200
    },
    {
      "epoch": 7.801693864903945,
      "grad_norm": 32.367427825927734,
      "learning_rate": 1.2211021135165906e-05,
      "loss": 1.4139,
      "step": 47210
    },
    {
      "epoch": 7.803346416029746,
      "grad_norm": 9.915732383728027,
      "learning_rate": 1.2201839916267283e-05,
      "loss": 1.5902,
      "step": 47220
    },
    {
      "epoch": 7.804998967155546,
      "grad_norm": 10.596333503723145,
      "learning_rate": 1.2192658697368662e-05,
      "loss": 1.4343,
      "step": 47230
    },
    {
      "epoch": 7.806651518281347,
      "grad_norm": 16.008501052856445,
      "learning_rate": 1.2183477478470043e-05,
      "loss": 1.3559,
      "step": 47240
    },
    {
      "epoch": 7.808304069407147,
      "grad_norm": 14.380460739135742,
      "learning_rate": 1.2174296259571421e-05,
      "loss": 1.4441,
      "step": 47250
    },
    {
      "epoch": 7.809956620532947,
      "grad_norm": 18.642454147338867,
      "learning_rate": 1.21651150406728e-05,
      "loss": 1.3636,
      "step": 47260
    },
    {
      "epoch": 7.811609171658748,
      "grad_norm": 10.425969123840332,
      "learning_rate": 1.215593382177418e-05,
      "loss": 1.4597,
      "step": 47270
    },
    {
      "epoch": 7.8132617227845484,
      "grad_norm": 15.50149917602539,
      "learning_rate": 1.2146752602875559e-05,
      "loss": 1.3136,
      "step": 47280
    },
    {
      "epoch": 7.814914273910349,
      "grad_norm": 7.004604816436768,
      "learning_rate": 1.2137571383976936e-05,
      "loss": 1.3023,
      "step": 47290
    },
    {
      "epoch": 7.8165668250361495,
      "grad_norm": 10.380492210388184,
      "learning_rate": 1.2128390165078317e-05,
      "loss": 1.2798,
      "step": 47300
    },
    {
      "epoch": 7.8182193761619505,
      "grad_norm": 13.373892784118652,
      "learning_rate": 1.2119208946179696e-05,
      "loss": 1.4054,
      "step": 47310
    },
    {
      "epoch": 7.819871927287751,
      "grad_norm": 18.053157806396484,
      "learning_rate": 1.2110027727281074e-05,
      "loss": 1.5367,
      "step": 47320
    },
    {
      "epoch": 7.821524478413551,
      "grad_norm": 8.14160442352295,
      "learning_rate": 1.2100846508382453e-05,
      "loss": 1.2874,
      "step": 47330
    },
    {
      "epoch": 7.823177029539352,
      "grad_norm": 13.822787284851074,
      "learning_rate": 1.2091665289483832e-05,
      "loss": 1.3485,
      "step": 47340
    },
    {
      "epoch": 7.824829580665152,
      "grad_norm": 44.869102478027344,
      "learning_rate": 1.2082484070585212e-05,
      "loss": 1.4024,
      "step": 47350
    },
    {
      "epoch": 7.826482131790952,
      "grad_norm": 12.257972717285156,
      "learning_rate": 1.2073302851686591e-05,
      "loss": 1.3067,
      "step": 47360
    },
    {
      "epoch": 7.828134682916753,
      "grad_norm": 13.056764602661133,
      "learning_rate": 1.206412163278797e-05,
      "loss": 1.3647,
      "step": 47370
    },
    {
      "epoch": 7.829787234042553,
      "grad_norm": 9.314446449279785,
      "learning_rate": 1.2054940413889348e-05,
      "loss": 1.4241,
      "step": 47380
    },
    {
      "epoch": 7.831439785168354,
      "grad_norm": 8.149978637695312,
      "learning_rate": 1.2045759194990727e-05,
      "loss": 1.3955,
      "step": 47390
    },
    {
      "epoch": 7.833092336294154,
      "grad_norm": 25.435317993164062,
      "learning_rate": 1.2036577976092106e-05,
      "loss": 1.4185,
      "step": 47400
    },
    {
      "epoch": 7.834744887419955,
      "grad_norm": 15.4861478805542,
      "learning_rate": 1.2027396757193485e-05,
      "loss": 1.4882,
      "step": 47410
    },
    {
      "epoch": 7.836397438545755,
      "grad_norm": 8.784843444824219,
      "learning_rate": 1.2018215538294865e-05,
      "loss": 1.33,
      "step": 47420
    },
    {
      "epoch": 7.838049989671555,
      "grad_norm": 14.153306007385254,
      "learning_rate": 1.2009034319396244e-05,
      "loss": 1.3612,
      "step": 47430
    },
    {
      "epoch": 7.839702540797356,
      "grad_norm": 10.80195426940918,
      "learning_rate": 1.1999853100497623e-05,
      "loss": 1.48,
      "step": 47440
    },
    {
      "epoch": 7.841355091923156,
      "grad_norm": 8.831650733947754,
      "learning_rate": 1.1990671881599e-05,
      "loss": 1.3938,
      "step": 47450
    },
    {
      "epoch": 7.843007643048956,
      "grad_norm": 8.44402027130127,
      "learning_rate": 1.198149066270038e-05,
      "loss": 1.4246,
      "step": 47460
    },
    {
      "epoch": 7.844660194174757,
      "grad_norm": 10.85715389251709,
      "learning_rate": 1.1972309443801761e-05,
      "loss": 1.3261,
      "step": 47470
    },
    {
      "epoch": 7.8463127453005574,
      "grad_norm": 17.19044303894043,
      "learning_rate": 1.1963128224903138e-05,
      "loss": 1.5413,
      "step": 47480
    },
    {
      "epoch": 7.847965296426358,
      "grad_norm": 7.725858211517334,
      "learning_rate": 1.1953947006004518e-05,
      "loss": 1.4338,
      "step": 47490
    },
    {
      "epoch": 7.8496178475521585,
      "grad_norm": 9.736605644226074,
      "learning_rate": 1.1944765787105897e-05,
      "loss": 1.2932,
      "step": 47500
    },
    {
      "epoch": 7.8512703986779595,
      "grad_norm": 14.61507511138916,
      "learning_rate": 1.1935584568207275e-05,
      "loss": 1.4587,
      "step": 47510
    },
    {
      "epoch": 7.85292294980376,
      "grad_norm": 12.031658172607422,
      "learning_rate": 1.1926403349308654e-05,
      "loss": 1.4079,
      "step": 47520
    },
    {
      "epoch": 7.85457550092956,
      "grad_norm": 16.91095542907715,
      "learning_rate": 1.1917222130410035e-05,
      "loss": 1.3452,
      "step": 47530
    },
    {
      "epoch": 7.856228052055361,
      "grad_norm": 15.591407775878906,
      "learning_rate": 1.1908040911511412e-05,
      "loss": 1.4318,
      "step": 47540
    },
    {
      "epoch": 7.857880603181161,
      "grad_norm": 15.299722671508789,
      "learning_rate": 1.1898859692612792e-05,
      "loss": 1.3371,
      "step": 47550
    },
    {
      "epoch": 7.859533154306962,
      "grad_norm": 8.688148498535156,
      "learning_rate": 1.188967847371417e-05,
      "loss": 1.4927,
      "step": 47560
    },
    {
      "epoch": 7.861185705432762,
      "grad_norm": 14.59116268157959,
      "learning_rate": 1.188049725481555e-05,
      "loss": 1.3456,
      "step": 47570
    },
    {
      "epoch": 7.862838256558562,
      "grad_norm": 30.086807250976562,
      "learning_rate": 1.187131603591693e-05,
      "loss": 1.2876,
      "step": 47580
    },
    {
      "epoch": 7.864490807684363,
      "grad_norm": 25.886877059936523,
      "learning_rate": 1.1862134817018308e-05,
      "loss": 1.491,
      "step": 47590
    },
    {
      "epoch": 7.866143358810163,
      "grad_norm": 11.027227401733398,
      "learning_rate": 1.1852953598119688e-05,
      "loss": 1.4246,
      "step": 47600
    },
    {
      "epoch": 7.867795909935964,
      "grad_norm": 10.795921325683594,
      "learning_rate": 1.1843772379221065e-05,
      "loss": 1.3966,
      "step": 47610
    },
    {
      "epoch": 7.869448461061764,
      "grad_norm": 9.404576301574707,
      "learning_rate": 1.1834591160322445e-05,
      "loss": 1.3629,
      "step": 47620
    },
    {
      "epoch": 7.871101012187564,
      "grad_norm": 11.95871353149414,
      "learning_rate": 1.1825409941423824e-05,
      "loss": 1.4068,
      "step": 47630
    },
    {
      "epoch": 7.872753563313365,
      "grad_norm": 11.290194511413574,
      "learning_rate": 1.1816228722525203e-05,
      "loss": 1.4184,
      "step": 47640
    },
    {
      "epoch": 7.874406114439165,
      "grad_norm": 32.461849212646484,
      "learning_rate": 1.1807047503626582e-05,
      "loss": 1.4621,
      "step": 47650
    },
    {
      "epoch": 7.876058665564966,
      "grad_norm": 10.396666526794434,
      "learning_rate": 1.1797866284727962e-05,
      "loss": 1.3957,
      "step": 47660
    },
    {
      "epoch": 7.877711216690766,
      "grad_norm": 12.307036399841309,
      "learning_rate": 1.1788685065829339e-05,
      "loss": 1.3876,
      "step": 47670
    },
    {
      "epoch": 7.879363767816566,
      "grad_norm": 10.1054048538208,
      "learning_rate": 1.1779503846930718e-05,
      "loss": 1.4137,
      "step": 47680
    },
    {
      "epoch": 7.881016318942367,
      "grad_norm": 16.99799346923828,
      "learning_rate": 1.1770322628032098e-05,
      "loss": 1.3882,
      "step": 47690
    },
    {
      "epoch": 7.8826688700681675,
      "grad_norm": 14.859070777893066,
      "learning_rate": 1.1761141409133477e-05,
      "loss": 1.3793,
      "step": 47700
    },
    {
      "epoch": 7.8843214211939685,
      "grad_norm": 13.93166446685791,
      "learning_rate": 1.1751960190234856e-05,
      "loss": 1.4492,
      "step": 47710
    },
    {
      "epoch": 7.885973972319769,
      "grad_norm": 12.025364875793457,
      "learning_rate": 1.1742778971336235e-05,
      "loss": 1.5158,
      "step": 47720
    },
    {
      "epoch": 7.887626523445569,
      "grad_norm": 15.031390190124512,
      "learning_rate": 1.1733597752437615e-05,
      "loss": 1.3162,
      "step": 47730
    },
    {
      "epoch": 7.88927907457137,
      "grad_norm": 13.956084251403809,
      "learning_rate": 1.1724416533538992e-05,
      "loss": 1.2951,
      "step": 47740
    },
    {
      "epoch": 7.89093162569717,
      "grad_norm": 12.361295700073242,
      "learning_rate": 1.1715235314640373e-05,
      "loss": 1.5406,
      "step": 47750
    },
    {
      "epoch": 7.892584176822971,
      "grad_norm": 8.774741172790527,
      "learning_rate": 1.1706054095741752e-05,
      "loss": 1.2487,
      "step": 47760
    },
    {
      "epoch": 7.894236727948771,
      "grad_norm": 31.390975952148438,
      "learning_rate": 1.169687287684313e-05,
      "loss": 1.3818,
      "step": 47770
    },
    {
      "epoch": 7.895889279074572,
      "grad_norm": 7.857132911682129,
      "learning_rate": 1.1687691657944509e-05,
      "loss": 1.4169,
      "step": 47780
    },
    {
      "epoch": 7.897541830200372,
      "grad_norm": 27.877517700195312,
      "learning_rate": 1.1678510439045888e-05,
      "loss": 1.4917,
      "step": 47790
    },
    {
      "epoch": 7.899194381326172,
      "grad_norm": 21.92901611328125,
      "learning_rate": 1.1669329220147266e-05,
      "loss": 1.4045,
      "step": 47800
    },
    {
      "epoch": 7.900846932451973,
      "grad_norm": 13.53458023071289,
      "learning_rate": 1.1660148001248647e-05,
      "loss": 1.4553,
      "step": 47810
    },
    {
      "epoch": 7.902499483577773,
      "grad_norm": 13.368227005004883,
      "learning_rate": 1.1650966782350026e-05,
      "loss": 1.3294,
      "step": 47820
    },
    {
      "epoch": 7.904152034703573,
      "grad_norm": 9.063852310180664,
      "learning_rate": 1.1641785563451404e-05,
      "loss": 1.4244,
      "step": 47830
    },
    {
      "epoch": 7.905804585829374,
      "grad_norm": 62.96859359741211,
      "learning_rate": 1.1632604344552783e-05,
      "loss": 1.3679,
      "step": 47840
    },
    {
      "epoch": 7.907457136955174,
      "grad_norm": 11.219415664672852,
      "learning_rate": 1.1623423125654162e-05,
      "loss": 1.4244,
      "step": 47850
    },
    {
      "epoch": 7.909109688080975,
      "grad_norm": 13.625591278076172,
      "learning_rate": 1.1614241906755541e-05,
      "loss": 1.4893,
      "step": 47860
    },
    {
      "epoch": 7.910762239206775,
      "grad_norm": 7.387325763702393,
      "learning_rate": 1.160506068785692e-05,
      "loss": 1.3991,
      "step": 47870
    },
    {
      "epoch": 7.912414790332576,
      "grad_norm": 11.21784782409668,
      "learning_rate": 1.15958794689583e-05,
      "loss": 1.3294,
      "step": 47880
    },
    {
      "epoch": 7.914067341458376,
      "grad_norm": 17.778669357299805,
      "learning_rate": 1.1586698250059679e-05,
      "loss": 1.4258,
      "step": 47890
    },
    {
      "epoch": 7.9157198925841765,
      "grad_norm": 12.979781150817871,
      "learning_rate": 1.1577517031161057e-05,
      "loss": 1.2635,
      "step": 47900
    },
    {
      "epoch": 7.9173724437099775,
      "grad_norm": 13.952737808227539,
      "learning_rate": 1.1568335812262436e-05,
      "loss": 1.4153,
      "step": 47910
    },
    {
      "epoch": 7.919024994835778,
      "grad_norm": 10.576740264892578,
      "learning_rate": 1.1559154593363815e-05,
      "loss": 1.455,
      "step": 47920
    },
    {
      "epoch": 7.920677545961578,
      "grad_norm": 9.399680137634277,
      "learning_rate": 1.1549973374465194e-05,
      "loss": 1.3299,
      "step": 47930
    },
    {
      "epoch": 7.922330097087379,
      "grad_norm": 18.13553237915039,
      "learning_rate": 1.1540792155566574e-05,
      "loss": 1.3659,
      "step": 47940
    },
    {
      "epoch": 7.923982648213179,
      "grad_norm": 17.09833335876465,
      "learning_rate": 1.1531610936667953e-05,
      "loss": 1.3799,
      "step": 47950
    },
    {
      "epoch": 7.92563519933898,
      "grad_norm": 11.579339027404785,
      "learning_rate": 1.152242971776933e-05,
      "loss": 1.4604,
      "step": 47960
    },
    {
      "epoch": 7.92728775046478,
      "grad_norm": 17.407424926757812,
      "learning_rate": 1.151324849887071e-05,
      "loss": 1.4101,
      "step": 47970
    },
    {
      "epoch": 7.928940301590581,
      "grad_norm": 17.47738265991211,
      "learning_rate": 1.150406727997209e-05,
      "loss": 1.3574,
      "step": 47980
    },
    {
      "epoch": 7.930592852716381,
      "grad_norm": 10.31318473815918,
      "learning_rate": 1.1494886061073468e-05,
      "loss": 1.457,
      "step": 47990
    },
    {
      "epoch": 7.932245403842181,
      "grad_norm": 14.748434066772461,
      "learning_rate": 1.1485704842174847e-05,
      "loss": 1.3569,
      "step": 48000
    },
    {
      "epoch": 7.933897954967982,
      "grad_norm": 10.245794296264648,
      "learning_rate": 1.1476523623276227e-05,
      "loss": 1.4359,
      "step": 48010
    },
    {
      "epoch": 7.935550506093782,
      "grad_norm": 14.848341941833496,
      "learning_rate": 1.1467342404377606e-05,
      "loss": 1.365,
      "step": 48020
    },
    {
      "epoch": 7.937203057219583,
      "grad_norm": 11.871755599975586,
      "learning_rate": 1.1458161185478984e-05,
      "loss": 1.4103,
      "step": 48030
    },
    {
      "epoch": 7.938855608345383,
      "grad_norm": 13.944067001342773,
      "learning_rate": 1.1448979966580364e-05,
      "loss": 1.4093,
      "step": 48040
    },
    {
      "epoch": 7.940508159471183,
      "grad_norm": 67.12018585205078,
      "learning_rate": 1.1439798747681744e-05,
      "loss": 1.3977,
      "step": 48050
    },
    {
      "epoch": 7.942160710596984,
      "grad_norm": 13.307299613952637,
      "learning_rate": 1.1430617528783121e-05,
      "loss": 1.2964,
      "step": 48060
    },
    {
      "epoch": 7.943813261722784,
      "grad_norm": 7.652238845825195,
      "learning_rate": 1.14214363098845e-05,
      "loss": 1.221,
      "step": 48070
    },
    {
      "epoch": 7.945465812848585,
      "grad_norm": 11.270803451538086,
      "learning_rate": 1.141225509098588e-05,
      "loss": 1.3174,
      "step": 48080
    },
    {
      "epoch": 7.947118363974385,
      "grad_norm": 16.249711990356445,
      "learning_rate": 1.1403073872087259e-05,
      "loss": 1.3507,
      "step": 48090
    },
    {
      "epoch": 7.9487709151001855,
      "grad_norm": 12.031152725219727,
      "learning_rate": 1.1393892653188638e-05,
      "loss": 1.4643,
      "step": 48100
    },
    {
      "epoch": 7.9504234662259865,
      "grad_norm": 9.80186939239502,
      "learning_rate": 1.1384711434290018e-05,
      "loss": 1.2806,
      "step": 48110
    },
    {
      "epoch": 7.952076017351787,
      "grad_norm": 8.662875175476074,
      "learning_rate": 1.1375530215391395e-05,
      "loss": 1.3472,
      "step": 48120
    },
    {
      "epoch": 7.953728568477588,
      "grad_norm": 10.309189796447754,
      "learning_rate": 1.1366348996492774e-05,
      "loss": 1.3387,
      "step": 48130
    },
    {
      "epoch": 7.955381119603388,
      "grad_norm": 16.453157424926758,
      "learning_rate": 1.1357167777594154e-05,
      "loss": 1.3604,
      "step": 48140
    },
    {
      "epoch": 7.957033670729189,
      "grad_norm": 16.2752628326416,
      "learning_rate": 1.1347986558695533e-05,
      "loss": 1.4941,
      "step": 48150
    },
    {
      "epoch": 7.958686221854989,
      "grad_norm": 9.665366172790527,
      "learning_rate": 1.1338805339796912e-05,
      "loss": 1.3735,
      "step": 48160
    },
    {
      "epoch": 7.960338772980789,
      "grad_norm": 12.39187240600586,
      "learning_rate": 1.1329624120898291e-05,
      "loss": 1.3077,
      "step": 48170
    },
    {
      "epoch": 7.96199132410659,
      "grad_norm": 12.81519603729248,
      "learning_rate": 1.132044290199967e-05,
      "loss": 1.3662,
      "step": 48180
    },
    {
      "epoch": 7.96364387523239,
      "grad_norm": 23.483640670776367,
      "learning_rate": 1.1311261683101048e-05,
      "loss": 1.4731,
      "step": 48190
    },
    {
      "epoch": 7.96529642635819,
      "grad_norm": 26.04778480529785,
      "learning_rate": 1.1302080464202427e-05,
      "loss": 1.5032,
      "step": 48200
    },
    {
      "epoch": 7.966948977483991,
      "grad_norm": 7.779325485229492,
      "learning_rate": 1.1292899245303808e-05,
      "loss": 1.412,
      "step": 48210
    },
    {
      "epoch": 7.968601528609791,
      "grad_norm": 9.814796447753906,
      "learning_rate": 1.1283718026405186e-05,
      "loss": 1.4677,
      "step": 48220
    },
    {
      "epoch": 7.970254079735592,
      "grad_norm": 12.185991287231445,
      "learning_rate": 1.1274536807506565e-05,
      "loss": 1.3061,
      "step": 48230
    },
    {
      "epoch": 7.971906630861392,
      "grad_norm": 15.006294250488281,
      "learning_rate": 1.1265355588607944e-05,
      "loss": 1.3275,
      "step": 48240
    },
    {
      "epoch": 7.973559181987193,
      "grad_norm": 12.14893627166748,
      "learning_rate": 1.1256174369709324e-05,
      "loss": 1.3091,
      "step": 48250
    },
    {
      "epoch": 7.975211733112993,
      "grad_norm": 14.654484748840332,
      "learning_rate": 1.1246993150810701e-05,
      "loss": 1.4328,
      "step": 48260
    },
    {
      "epoch": 7.976864284238793,
      "grad_norm": 13.88418960571289,
      "learning_rate": 1.1237811931912082e-05,
      "loss": 1.3212,
      "step": 48270
    },
    {
      "epoch": 7.978516835364594,
      "grad_norm": 8.910808563232422,
      "learning_rate": 1.122863071301346e-05,
      "loss": 1.2907,
      "step": 48280
    },
    {
      "epoch": 7.980169386490394,
      "grad_norm": 9.435023307800293,
      "learning_rate": 1.1219449494114839e-05,
      "loss": 1.3946,
      "step": 48290
    },
    {
      "epoch": 7.9818219376161945,
      "grad_norm": 23.680673599243164,
      "learning_rate": 1.1210268275216218e-05,
      "loss": 1.401,
      "step": 48300
    },
    {
      "epoch": 7.9834744887419955,
      "grad_norm": 14.382912635803223,
      "learning_rate": 1.1201087056317597e-05,
      "loss": 1.367,
      "step": 48310
    },
    {
      "epoch": 7.985127039867796,
      "grad_norm": 18.664398193359375,
      "learning_rate": 1.1191905837418977e-05,
      "loss": 1.4019,
      "step": 48320
    },
    {
      "epoch": 7.986779590993597,
      "grad_norm": 13.360762596130371,
      "learning_rate": 1.1182724618520356e-05,
      "loss": 1.5117,
      "step": 48330
    },
    {
      "epoch": 7.988432142119397,
      "grad_norm": 13.729475975036621,
      "learning_rate": 1.1173543399621735e-05,
      "loss": 1.5005,
      "step": 48340
    },
    {
      "epoch": 7.990084693245198,
      "grad_norm": 9.402355194091797,
      "learning_rate": 1.1164362180723113e-05,
      "loss": 1.4522,
      "step": 48350
    },
    {
      "epoch": 7.991737244370998,
      "grad_norm": 10.234284400939941,
      "learning_rate": 1.1155180961824492e-05,
      "loss": 1.4535,
      "step": 48360
    },
    {
      "epoch": 7.993389795496798,
      "grad_norm": 14.257830619812012,
      "learning_rate": 1.1145999742925871e-05,
      "loss": 1.3578,
      "step": 48370
    },
    {
      "epoch": 7.995042346622599,
      "grad_norm": 16.891815185546875,
      "learning_rate": 1.113681852402725e-05,
      "loss": 1.46,
      "step": 48380
    },
    {
      "epoch": 7.996694897748399,
      "grad_norm": 11.595968246459961,
      "learning_rate": 1.112763730512863e-05,
      "loss": 1.3274,
      "step": 48390
    },
    {
      "epoch": 7.9983474488742,
      "grad_norm": 12.209437370300293,
      "learning_rate": 1.1118456086230009e-05,
      "loss": 1.3268,
      "step": 48400
    },
    {
      "epoch": 8.0,
      "grad_norm": 10.217020988464355,
      "learning_rate": 1.1109274867331388e-05,
      "loss": 1.3077,
      "step": 48410
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.3359696410838417,
      "eval_loss": 2.232658863067627,
      "eval_runtime": 833.0289,
      "eval_samples_per_second": 33.848,
      "eval_steps_per_second": 8.462,
      "step": 48410
    },
    {
      "epoch": 8.001652551125801,
      "grad_norm": 13.642919540405273,
      "learning_rate": 1.1100093648432766e-05,
      "loss": 1.2788,
      "step": 48420
    },
    {
      "epoch": 8.0033051022516,
      "grad_norm": 11.345486640930176,
      "learning_rate": 1.1090912429534145e-05,
      "loss": 1.3731,
      "step": 48430
    },
    {
      "epoch": 8.004957653377401,
      "grad_norm": 13.825784683227539,
      "learning_rate": 1.1081731210635524e-05,
      "loss": 1.398,
      "step": 48440
    },
    {
      "epoch": 8.006610204503202,
      "grad_norm": 22.2542667388916,
      "learning_rate": 1.1072549991736903e-05,
      "loss": 1.3321,
      "step": 48450
    },
    {
      "epoch": 8.008262755629003,
      "grad_norm": 12.9199800491333,
      "learning_rate": 1.1063368772838283e-05,
      "loss": 1.3493,
      "step": 48460
    },
    {
      "epoch": 8.009915306754802,
      "grad_norm": 9.398885726928711,
      "learning_rate": 1.1054187553939662e-05,
      "loss": 1.292,
      "step": 48470
    },
    {
      "epoch": 8.011567857880603,
      "grad_norm": 17.111080169677734,
      "learning_rate": 1.104500633504104e-05,
      "loss": 1.3765,
      "step": 48480
    },
    {
      "epoch": 8.013220409006404,
      "grad_norm": 268.5393981933594,
      "learning_rate": 1.1035825116142419e-05,
      "loss": 1.3245,
      "step": 48490
    },
    {
      "epoch": 8.014872960132204,
      "grad_norm": 11.041481971740723,
      "learning_rate": 1.10266438972438e-05,
      "loss": 1.3954,
      "step": 48500
    },
    {
      "epoch": 8.016525511258004,
      "grad_norm": 13.948131561279297,
      "learning_rate": 1.1017462678345177e-05,
      "loss": 1.2804,
      "step": 48510
    },
    {
      "epoch": 8.018178062383805,
      "grad_norm": 10.275325775146484,
      "learning_rate": 1.1008281459446557e-05,
      "loss": 1.4867,
      "step": 48520
    },
    {
      "epoch": 8.019830613509605,
      "grad_norm": 9.993414878845215,
      "learning_rate": 1.0999100240547936e-05,
      "loss": 1.4039,
      "step": 48530
    },
    {
      "epoch": 8.021483164635406,
      "grad_norm": 10.115413665771484,
      "learning_rate": 1.0989919021649315e-05,
      "loss": 1.2562,
      "step": 48540
    },
    {
      "epoch": 8.023135715761207,
      "grad_norm": 12.741392135620117,
      "learning_rate": 1.0980737802750694e-05,
      "loss": 1.1463,
      "step": 48550
    },
    {
      "epoch": 8.024788266887008,
      "grad_norm": 21.065656661987305,
      "learning_rate": 1.0971556583852073e-05,
      "loss": 1.3402,
      "step": 48560
    },
    {
      "epoch": 8.026440818012807,
      "grad_norm": 13.77077865600586,
      "learning_rate": 1.0962375364953453e-05,
      "loss": 1.4906,
      "step": 48570
    },
    {
      "epoch": 8.028093369138608,
      "grad_norm": 10.071250915527344,
      "learning_rate": 1.095319414605483e-05,
      "loss": 1.2456,
      "step": 48580
    },
    {
      "epoch": 8.029745920264409,
      "grad_norm": 41.76918029785156,
      "learning_rate": 1.094401292715621e-05,
      "loss": 1.3994,
      "step": 48590
    },
    {
      "epoch": 8.031398471390208,
      "grad_norm": 12.07766342163086,
      "learning_rate": 1.0934831708257589e-05,
      "loss": 1.5174,
      "step": 48600
    },
    {
      "epoch": 8.033051022516009,
      "grad_norm": 12.890829086303711,
      "learning_rate": 1.0925650489358968e-05,
      "loss": 1.398,
      "step": 48610
    },
    {
      "epoch": 8.03470357364181,
      "grad_norm": 18.99813461303711,
      "learning_rate": 1.0916469270460347e-05,
      "loss": 1.4505,
      "step": 48620
    },
    {
      "epoch": 8.03635612476761,
      "grad_norm": 13.51767635345459,
      "learning_rate": 1.0907288051561727e-05,
      "loss": 1.2901,
      "step": 48630
    },
    {
      "epoch": 8.03800867589341,
      "grad_norm": 11.078587532043457,
      "learning_rate": 1.0898106832663104e-05,
      "loss": 1.273,
      "step": 48640
    },
    {
      "epoch": 8.039661227019211,
      "grad_norm": 9.125107765197754,
      "learning_rate": 1.0888925613764483e-05,
      "loss": 1.2565,
      "step": 48650
    },
    {
      "epoch": 8.041313778145012,
      "grad_norm": 17.25629997253418,
      "learning_rate": 1.0879744394865863e-05,
      "loss": 1.4344,
      "step": 48660
    },
    {
      "epoch": 8.042966329270811,
      "grad_norm": 10.981740951538086,
      "learning_rate": 1.0870563175967242e-05,
      "loss": 1.4435,
      "step": 48670
    },
    {
      "epoch": 8.044618880396612,
      "grad_norm": 24.005199432373047,
      "learning_rate": 1.0861381957068621e-05,
      "loss": 1.3133,
      "step": 48680
    },
    {
      "epoch": 8.046271431522413,
      "grad_norm": 10.016214370727539,
      "learning_rate": 1.085220073817e-05,
      "loss": 1.3351,
      "step": 48690
    },
    {
      "epoch": 8.047923982648213,
      "grad_norm": 19.678804397583008,
      "learning_rate": 1.084301951927138e-05,
      "loss": 1.5142,
      "step": 48700
    },
    {
      "epoch": 8.049576533774013,
      "grad_norm": 12.911602020263672,
      "learning_rate": 1.0833838300372757e-05,
      "loss": 1.3535,
      "step": 48710
    },
    {
      "epoch": 8.051229084899814,
      "grad_norm": 28.89436149597168,
      "learning_rate": 1.0824657081474138e-05,
      "loss": 1.3765,
      "step": 48720
    },
    {
      "epoch": 8.052881636025614,
      "grad_norm": 9.610832214355469,
      "learning_rate": 1.0815475862575517e-05,
      "loss": 1.3013,
      "step": 48730
    },
    {
      "epoch": 8.054534187151415,
      "grad_norm": 10.988263130187988,
      "learning_rate": 1.0806294643676895e-05,
      "loss": 1.4339,
      "step": 48740
    },
    {
      "epoch": 8.056186738277216,
      "grad_norm": 8.778138160705566,
      "learning_rate": 1.0797113424778274e-05,
      "loss": 1.4137,
      "step": 48750
    },
    {
      "epoch": 8.057839289403017,
      "grad_norm": 14.911858558654785,
      "learning_rate": 1.0787932205879653e-05,
      "loss": 1.4565,
      "step": 48760
    },
    {
      "epoch": 8.059491840528816,
      "grad_norm": 11.017518043518066,
      "learning_rate": 1.0778750986981031e-05,
      "loss": 1.2947,
      "step": 48770
    },
    {
      "epoch": 8.061144391654617,
      "grad_norm": 10.435925483703613,
      "learning_rate": 1.0769569768082412e-05,
      "loss": 1.2544,
      "step": 48780
    },
    {
      "epoch": 8.062796942780418,
      "grad_norm": 15.238306999206543,
      "learning_rate": 1.0760388549183791e-05,
      "loss": 1.4135,
      "step": 48790
    },
    {
      "epoch": 8.064449493906217,
      "grad_norm": 19.062891006469727,
      "learning_rate": 1.0751207330285169e-05,
      "loss": 1.375,
      "step": 48800
    },
    {
      "epoch": 8.066102045032018,
      "grad_norm": 32.4384651184082,
      "learning_rate": 1.0742026111386548e-05,
      "loss": 1.5354,
      "step": 48810
    },
    {
      "epoch": 8.067754596157819,
      "grad_norm": 12.400315284729004,
      "learning_rate": 1.0732844892487927e-05,
      "loss": 1.3206,
      "step": 48820
    },
    {
      "epoch": 8.06940714728362,
      "grad_norm": 21.19147491455078,
      "learning_rate": 1.0723663673589306e-05,
      "loss": 1.4587,
      "step": 48830
    },
    {
      "epoch": 8.07105969840942,
      "grad_norm": 14.054428100585938,
      "learning_rate": 1.0714482454690686e-05,
      "loss": 1.3973,
      "step": 48840
    },
    {
      "epoch": 8.07271224953522,
      "grad_norm": 9.821606636047363,
      "learning_rate": 1.0705301235792065e-05,
      "loss": 1.3624,
      "step": 48850
    },
    {
      "epoch": 8.074364800661021,
      "grad_norm": 17.55989646911621,
      "learning_rate": 1.0696120016893444e-05,
      "loss": 1.4144,
      "step": 48860
    },
    {
      "epoch": 8.07601735178682,
      "grad_norm": 14.98707103729248,
      "learning_rate": 1.0686938797994822e-05,
      "loss": 1.3011,
      "step": 48870
    },
    {
      "epoch": 8.077669902912621,
      "grad_norm": 20.7563419342041,
      "learning_rate": 1.0677757579096201e-05,
      "loss": 1.282,
      "step": 48880
    },
    {
      "epoch": 8.079322454038422,
      "grad_norm": 12.075446128845215,
      "learning_rate": 1.066857636019758e-05,
      "loss": 1.3504,
      "step": 48890
    },
    {
      "epoch": 8.080975005164222,
      "grad_norm": 41.509918212890625,
      "learning_rate": 1.065939514129896e-05,
      "loss": 1.2301,
      "step": 48900
    },
    {
      "epoch": 8.082627556290022,
      "grad_norm": 14.281842231750488,
      "learning_rate": 1.0650213922400339e-05,
      "loss": 1.3586,
      "step": 48910
    },
    {
      "epoch": 8.084280107415823,
      "grad_norm": 11.7456693649292,
      "learning_rate": 1.0641032703501718e-05,
      "loss": 1.3069,
      "step": 48920
    },
    {
      "epoch": 8.085932658541624,
      "grad_norm": 19.099252700805664,
      "learning_rate": 1.0631851484603095e-05,
      "loss": 1.4793,
      "step": 48930
    },
    {
      "epoch": 8.087585209667424,
      "grad_norm": 24.131771087646484,
      "learning_rate": 1.0622670265704475e-05,
      "loss": 1.2569,
      "step": 48940
    },
    {
      "epoch": 8.089237760793225,
      "grad_norm": 11.209100723266602,
      "learning_rate": 1.0613489046805856e-05,
      "loss": 1.2394,
      "step": 48950
    },
    {
      "epoch": 8.090890311919026,
      "grad_norm": 13.452157020568848,
      "learning_rate": 1.0604307827907233e-05,
      "loss": 1.2686,
      "step": 48960
    },
    {
      "epoch": 8.092542863044825,
      "grad_norm": 15.363450050354004,
      "learning_rate": 1.0595126609008612e-05,
      "loss": 1.4572,
      "step": 48970
    },
    {
      "epoch": 8.094195414170626,
      "grad_norm": 10.773612022399902,
      "learning_rate": 1.0585945390109992e-05,
      "loss": 1.3411,
      "step": 48980
    },
    {
      "epoch": 8.095847965296427,
      "grad_norm": 142.82435607910156,
      "learning_rate": 1.0576764171211371e-05,
      "loss": 1.3155,
      "step": 48990
    },
    {
      "epoch": 8.097500516422226,
      "grad_norm": 67.02763366699219,
      "learning_rate": 1.0567582952312749e-05,
      "loss": 1.39,
      "step": 49000
    },
    {
      "epoch": 8.099153067548027,
      "grad_norm": 38.37979507446289,
      "learning_rate": 1.055840173341413e-05,
      "loss": 1.3099,
      "step": 49010
    },
    {
      "epoch": 8.100805618673828,
      "grad_norm": 12.719371795654297,
      "learning_rate": 1.0549220514515509e-05,
      "loss": 1.3926,
      "step": 49020
    },
    {
      "epoch": 8.102458169799629,
      "grad_norm": 18.666179656982422,
      "learning_rate": 1.0540039295616886e-05,
      "loss": 1.3513,
      "step": 49030
    },
    {
      "epoch": 8.104110720925428,
      "grad_norm": 11.054634094238281,
      "learning_rate": 1.0530858076718266e-05,
      "loss": 1.3234,
      "step": 49040
    },
    {
      "epoch": 8.10576327205123,
      "grad_norm": 14.501553535461426,
      "learning_rate": 1.0521676857819645e-05,
      "loss": 1.352,
      "step": 49050
    },
    {
      "epoch": 8.10741582317703,
      "grad_norm": 12.57413101196289,
      "learning_rate": 1.0512495638921022e-05,
      "loss": 1.2829,
      "step": 49060
    },
    {
      "epoch": 8.10906837430283,
      "grad_norm": 19.41373062133789,
      "learning_rate": 1.0503314420022403e-05,
      "loss": 1.2808,
      "step": 49070
    },
    {
      "epoch": 8.11072092542863,
      "grad_norm": 12.489398956298828,
      "learning_rate": 1.0494133201123783e-05,
      "loss": 1.3782,
      "step": 49080
    },
    {
      "epoch": 8.112373476554431,
      "grad_norm": 17.002321243286133,
      "learning_rate": 1.048495198222516e-05,
      "loss": 1.4326,
      "step": 49090
    },
    {
      "epoch": 8.11402602768023,
      "grad_norm": 10.533157348632812,
      "learning_rate": 1.047577076332654e-05,
      "loss": 1.2274,
      "step": 49100
    },
    {
      "epoch": 8.115678578806031,
      "grad_norm": 14.494053840637207,
      "learning_rate": 1.0466589544427919e-05,
      "loss": 1.3649,
      "step": 49110
    },
    {
      "epoch": 8.117331129931832,
      "grad_norm": 8.390114784240723,
      "learning_rate": 1.0457408325529298e-05,
      "loss": 1.3052,
      "step": 49120
    },
    {
      "epoch": 8.118983681057633,
      "grad_norm": 23.97467803955078,
      "learning_rate": 1.0448227106630677e-05,
      "loss": 1.3362,
      "step": 49130
    },
    {
      "epoch": 8.120636232183433,
      "grad_norm": 18.547996520996094,
      "learning_rate": 1.0439045887732056e-05,
      "loss": 1.3471,
      "step": 49140
    },
    {
      "epoch": 8.122288783309234,
      "grad_norm": 17.564769744873047,
      "learning_rate": 1.0429864668833436e-05,
      "loss": 1.3934,
      "step": 49150
    },
    {
      "epoch": 8.123941334435035,
      "grad_norm": 25.809839248657227,
      "learning_rate": 1.0420683449934813e-05,
      "loss": 1.5178,
      "step": 49160
    },
    {
      "epoch": 8.125593885560834,
      "grad_norm": 24.299781799316406,
      "learning_rate": 1.0411502231036192e-05,
      "loss": 1.352,
      "step": 49170
    },
    {
      "epoch": 8.127246436686635,
      "grad_norm": 13.150251388549805,
      "learning_rate": 1.0402321012137573e-05,
      "loss": 1.4525,
      "step": 49180
    },
    {
      "epoch": 8.128898987812436,
      "grad_norm": 14.097673416137695,
      "learning_rate": 1.0393139793238951e-05,
      "loss": 1.4137,
      "step": 49190
    },
    {
      "epoch": 8.130551538938235,
      "grad_norm": 13.112128257751465,
      "learning_rate": 1.038395857434033e-05,
      "loss": 1.2533,
      "step": 49200
    },
    {
      "epoch": 8.132204090064036,
      "grad_norm": 12.603325843811035,
      "learning_rate": 1.037477735544171e-05,
      "loss": 1.4444,
      "step": 49210
    },
    {
      "epoch": 8.133856641189837,
      "grad_norm": 41.19782638549805,
      "learning_rate": 1.0365596136543087e-05,
      "loss": 1.45,
      "step": 49220
    },
    {
      "epoch": 8.135509192315638,
      "grad_norm": 15.660469055175781,
      "learning_rate": 1.0356414917644466e-05,
      "loss": 1.436,
      "step": 49230
    },
    {
      "epoch": 8.137161743441437,
      "grad_norm": 12.267075538635254,
      "learning_rate": 1.0347233698745847e-05,
      "loss": 1.4064,
      "step": 49240
    },
    {
      "epoch": 8.138814294567238,
      "grad_norm": 12.525456428527832,
      "learning_rate": 1.0338052479847225e-05,
      "loss": 1.3886,
      "step": 49250
    },
    {
      "epoch": 8.14046684569304,
      "grad_norm": 12.728673934936523,
      "learning_rate": 1.0328871260948604e-05,
      "loss": 1.3165,
      "step": 49260
    },
    {
      "epoch": 8.142119396818838,
      "grad_norm": 14.614742279052734,
      "learning_rate": 1.0319690042049983e-05,
      "loss": 1.2583,
      "step": 49270
    },
    {
      "epoch": 8.14377194794464,
      "grad_norm": 14.37644100189209,
      "learning_rate": 1.0310508823151362e-05,
      "loss": 1.2297,
      "step": 49280
    },
    {
      "epoch": 8.14542449907044,
      "grad_norm": 10.755502700805664,
      "learning_rate": 1.030132760425274e-05,
      "loss": 1.3726,
      "step": 49290
    },
    {
      "epoch": 8.147077050196241,
      "grad_norm": 156.68991088867188,
      "learning_rate": 1.0292146385354121e-05,
      "loss": 1.4638,
      "step": 49300
    },
    {
      "epoch": 8.14872960132204,
      "grad_norm": 16.2905330657959,
      "learning_rate": 1.02829651664555e-05,
      "loss": 1.3056,
      "step": 49310
    },
    {
      "epoch": 8.150382152447841,
      "grad_norm": 12.603002548217773,
      "learning_rate": 1.0273783947556878e-05,
      "loss": 1.4877,
      "step": 49320
    },
    {
      "epoch": 8.152034703573642,
      "grad_norm": 14.49899673461914,
      "learning_rate": 1.0264602728658257e-05,
      "loss": 1.3918,
      "step": 49330
    },
    {
      "epoch": 8.153687254699442,
      "grad_norm": 12.177115440368652,
      "learning_rate": 1.0255421509759636e-05,
      "loss": 1.2091,
      "step": 49340
    },
    {
      "epoch": 8.155339805825243,
      "grad_norm": 9.79865550994873,
      "learning_rate": 1.0246240290861015e-05,
      "loss": 1.2554,
      "step": 49350
    },
    {
      "epoch": 8.156992356951044,
      "grad_norm": 8.341238975524902,
      "learning_rate": 1.0237059071962395e-05,
      "loss": 1.2596,
      "step": 49360
    },
    {
      "epoch": 8.158644908076843,
      "grad_norm": 15.849066734313965,
      "learning_rate": 1.0227877853063774e-05,
      "loss": 1.2234,
      "step": 49370
    },
    {
      "epoch": 8.160297459202644,
      "grad_norm": 8.827754974365234,
      "learning_rate": 1.0218696634165151e-05,
      "loss": 1.3488,
      "step": 49380
    },
    {
      "epoch": 8.161950010328445,
      "grad_norm": 10.364538192749023,
      "learning_rate": 1.020951541526653e-05,
      "loss": 1.356,
      "step": 49390
    },
    {
      "epoch": 8.163602561454246,
      "grad_norm": 33.56261444091797,
      "learning_rate": 1.020033419636791e-05,
      "loss": 1.3118,
      "step": 49400
    },
    {
      "epoch": 8.165255112580045,
      "grad_norm": 21.711645126342773,
      "learning_rate": 1.019115297746929e-05,
      "loss": 1.2635,
      "step": 49410
    },
    {
      "epoch": 8.166907663705846,
      "grad_norm": 11.296979904174805,
      "learning_rate": 1.0181971758570668e-05,
      "loss": 1.3455,
      "step": 49420
    },
    {
      "epoch": 8.168560214831647,
      "grad_norm": 22.215770721435547,
      "learning_rate": 1.0172790539672048e-05,
      "loss": 1.29,
      "step": 49430
    },
    {
      "epoch": 8.170212765957446,
      "grad_norm": 16.198862075805664,
      "learning_rate": 1.0163609320773427e-05,
      "loss": 1.4448,
      "step": 49440
    },
    {
      "epoch": 8.171865317083247,
      "grad_norm": 15.688820838928223,
      "learning_rate": 1.0154428101874805e-05,
      "loss": 1.4095,
      "step": 49450
    },
    {
      "epoch": 8.173517868209048,
      "grad_norm": 18.601253509521484,
      "learning_rate": 1.0145246882976184e-05,
      "loss": 1.3167,
      "step": 49460
    },
    {
      "epoch": 8.175170419334847,
      "grad_norm": 14.552800178527832,
      "learning_rate": 1.0136065664077565e-05,
      "loss": 1.3402,
      "step": 49470
    },
    {
      "epoch": 8.176822970460648,
      "grad_norm": 25.716659545898438,
      "learning_rate": 1.0126884445178942e-05,
      "loss": 1.2686,
      "step": 49480
    },
    {
      "epoch": 8.17847552158645,
      "grad_norm": 14.601028442382812,
      "learning_rate": 1.0117703226280322e-05,
      "loss": 1.3386,
      "step": 49490
    },
    {
      "epoch": 8.18012807271225,
      "grad_norm": 19.567943572998047,
      "learning_rate": 1.01085220073817e-05,
      "loss": 1.3595,
      "step": 49500
    },
    {
      "epoch": 8.18178062383805,
      "grad_norm": 16.065189361572266,
      "learning_rate": 1.0099340788483078e-05,
      "loss": 1.4657,
      "step": 49510
    },
    {
      "epoch": 8.18343317496385,
      "grad_norm": 21.156858444213867,
      "learning_rate": 1.009015956958446e-05,
      "loss": 1.3717,
      "step": 49520
    },
    {
      "epoch": 8.185085726089651,
      "grad_norm": 9.819899559020996,
      "learning_rate": 1.0080978350685838e-05,
      "loss": 1.3775,
      "step": 49530
    },
    {
      "epoch": 8.18673827721545,
      "grad_norm": 20.710819244384766,
      "learning_rate": 1.0071797131787216e-05,
      "loss": 1.4204,
      "step": 49540
    },
    {
      "epoch": 8.188390828341252,
      "grad_norm": 8.26885986328125,
      "learning_rate": 1.0062615912888595e-05,
      "loss": 1.3977,
      "step": 49550
    },
    {
      "epoch": 8.190043379467053,
      "grad_norm": 23.709930419921875,
      "learning_rate": 1.0053434693989975e-05,
      "loss": 1.3835,
      "step": 49560
    },
    {
      "epoch": 8.191695930592854,
      "grad_norm": 13.372426986694336,
      "learning_rate": 1.0044253475091354e-05,
      "loss": 1.3282,
      "step": 49570
    },
    {
      "epoch": 8.193348481718653,
      "grad_norm": 14.474409103393555,
      "learning_rate": 1.0035072256192733e-05,
      "loss": 1.4683,
      "step": 49580
    },
    {
      "epoch": 8.195001032844454,
      "grad_norm": 19.034423828125,
      "learning_rate": 1.0025891037294112e-05,
      "loss": 1.5369,
      "step": 49590
    },
    {
      "epoch": 8.196653583970255,
      "grad_norm": 13.448150634765625,
      "learning_rate": 1.0016709818395492e-05,
      "loss": 1.3128,
      "step": 49600
    },
    {
      "epoch": 8.198306135096054,
      "grad_norm": 16.181913375854492,
      "learning_rate": 1.0007528599496869e-05,
      "loss": 1.2877,
      "step": 49610
    },
    {
      "epoch": 8.199958686221855,
      "grad_norm": 10.915853500366211,
      "learning_rate": 9.998347380598248e-06,
      "loss": 1.237,
      "step": 49620
    },
    {
      "epoch": 8.201611237347656,
      "grad_norm": 11.362007141113281,
      "learning_rate": 9.989166161699628e-06,
      "loss": 1.5008,
      "step": 49630
    },
    {
      "epoch": 8.203263788473455,
      "grad_norm": 11.269021034240723,
      "learning_rate": 9.979984942801007e-06,
      "loss": 1.2711,
      "step": 49640
    },
    {
      "epoch": 8.204916339599256,
      "grad_norm": 18.16535758972168,
      "learning_rate": 9.970803723902386e-06,
      "loss": 1.3525,
      "step": 49650
    },
    {
      "epoch": 8.206568890725057,
      "grad_norm": 16.32088279724121,
      "learning_rate": 9.961622505003765e-06,
      "loss": 1.2825,
      "step": 49660
    },
    {
      "epoch": 8.208221441850858,
      "grad_norm": 10.657417297363281,
      "learning_rate": 9.952441286105143e-06,
      "loss": 1.4107,
      "step": 49670
    },
    {
      "epoch": 8.209873992976657,
      "grad_norm": 8.144217491149902,
      "learning_rate": 9.943260067206522e-06,
      "loss": 1.3751,
      "step": 49680
    },
    {
      "epoch": 8.211526544102458,
      "grad_norm": 11.399767875671387,
      "learning_rate": 9.934078848307901e-06,
      "loss": 1.3849,
      "step": 49690
    },
    {
      "epoch": 8.21317909522826,
      "grad_norm": 12.942078590393066,
      "learning_rate": 9.92489762940928e-06,
      "loss": 1.3903,
      "step": 49700
    },
    {
      "epoch": 8.214831646354058,
      "grad_norm": 14.669422149658203,
      "learning_rate": 9.91571641051066e-06,
      "loss": 1.2849,
      "step": 49710
    },
    {
      "epoch": 8.21648419747986,
      "grad_norm": 14.220985412597656,
      "learning_rate": 9.906535191612039e-06,
      "loss": 1.2295,
      "step": 49720
    },
    {
      "epoch": 8.21813674860566,
      "grad_norm": 15.08796215057373,
      "learning_rate": 9.897353972713418e-06,
      "loss": 1.2791,
      "step": 49730
    },
    {
      "epoch": 8.21978929973146,
      "grad_norm": 19.543704986572266,
      "learning_rate": 9.888172753814796e-06,
      "loss": 1.2695,
      "step": 49740
    },
    {
      "epoch": 8.22144185085726,
      "grad_norm": 12.316389083862305,
      "learning_rate": 9.878991534916177e-06,
      "loss": 1.4851,
      "step": 49750
    },
    {
      "epoch": 8.223094401983062,
      "grad_norm": 15.752791404724121,
      "learning_rate": 9.869810316017556e-06,
      "loss": 1.4293,
      "step": 49760
    },
    {
      "epoch": 8.224746953108863,
      "grad_norm": 12.644379615783691,
      "learning_rate": 9.860629097118934e-06,
      "loss": 1.2978,
      "step": 49770
    },
    {
      "epoch": 8.226399504234662,
      "grad_norm": 22.362035751342773,
      "learning_rate": 9.851447878220313e-06,
      "loss": 1.3376,
      "step": 49780
    },
    {
      "epoch": 8.228052055360463,
      "grad_norm": 12.713665008544922,
      "learning_rate": 9.842266659321692e-06,
      "loss": 1.4388,
      "step": 49790
    },
    {
      "epoch": 8.229704606486264,
      "grad_norm": 12.972372055053711,
      "learning_rate": 9.83308544042307e-06,
      "loss": 1.5245,
      "step": 49800
    },
    {
      "epoch": 8.231357157612063,
      "grad_norm": 18.19927215576172,
      "learning_rate": 9.82390422152445e-06,
      "loss": 1.3927,
      "step": 49810
    },
    {
      "epoch": 8.233009708737864,
      "grad_norm": 36.67835998535156,
      "learning_rate": 9.81472300262583e-06,
      "loss": 1.3945,
      "step": 49820
    },
    {
      "epoch": 8.234662259863665,
      "grad_norm": 13.975602149963379,
      "learning_rate": 9.805541783727207e-06,
      "loss": 1.5244,
      "step": 49830
    },
    {
      "epoch": 8.236314810989464,
      "grad_norm": 11.509796142578125,
      "learning_rate": 9.796360564828587e-06,
      "loss": 1.3487,
      "step": 49840
    },
    {
      "epoch": 8.237967362115265,
      "grad_norm": 13.768054008483887,
      "learning_rate": 9.787179345929966e-06,
      "loss": 1.3423,
      "step": 49850
    },
    {
      "epoch": 8.239619913241066,
      "grad_norm": 24.76049041748047,
      "learning_rate": 9.777998127031345e-06,
      "loss": 1.3772,
      "step": 49860
    },
    {
      "epoch": 8.241272464366867,
      "grad_norm": 19.85248374938965,
      "learning_rate": 9.768816908132724e-06,
      "loss": 1.364,
      "step": 49870
    },
    {
      "epoch": 8.242925015492666,
      "grad_norm": 10.143357276916504,
      "learning_rate": 9.759635689234104e-06,
      "loss": 1.3958,
      "step": 49880
    },
    {
      "epoch": 8.244577566618467,
      "grad_norm": 11.001799583435059,
      "learning_rate": 9.750454470335483e-06,
      "loss": 1.3621,
      "step": 49890
    },
    {
      "epoch": 8.246230117744268,
      "grad_norm": 9.355253219604492,
      "learning_rate": 9.74127325143686e-06,
      "loss": 1.2735,
      "step": 49900
    },
    {
      "epoch": 8.247882668870067,
      "grad_norm": 31.860811233520508,
      "learning_rate": 9.73209203253824e-06,
      "loss": 1.3465,
      "step": 49910
    },
    {
      "epoch": 8.249535219995868,
      "grad_norm": 9.52219295501709,
      "learning_rate": 9.72291081363962e-06,
      "loss": 1.2591,
      "step": 49920
    },
    {
      "epoch": 8.25118777112167,
      "grad_norm": 16.92643165588379,
      "learning_rate": 9.713729594740998e-06,
      "loss": 1.5207,
      "step": 49930
    },
    {
      "epoch": 8.252840322247469,
      "grad_norm": 24.846473693847656,
      "learning_rate": 9.704548375842377e-06,
      "loss": 1.1995,
      "step": 49940
    },
    {
      "epoch": 8.25449287337327,
      "grad_norm": 18.698997497558594,
      "learning_rate": 9.695367156943757e-06,
      "loss": 1.409,
      "step": 49950
    },
    {
      "epoch": 8.25614542449907,
      "grad_norm": 9.128297805786133,
      "learning_rate": 9.686185938045134e-06,
      "loss": 1.4773,
      "step": 49960
    },
    {
      "epoch": 8.257797975624872,
      "grad_norm": 8.733552932739258,
      "learning_rate": 9.677004719146514e-06,
      "loss": 1.3389,
      "step": 49970
    },
    {
      "epoch": 8.25945052675067,
      "grad_norm": 14.439172744750977,
      "learning_rate": 9.667823500247894e-06,
      "loss": 1.4054,
      "step": 49980
    },
    {
      "epoch": 8.261103077876472,
      "grad_norm": 25.46586036682129,
      "learning_rate": 9.658642281349272e-06,
      "loss": 1.385,
      "step": 49990
    },
    {
      "epoch": 8.262755629002273,
      "grad_norm": 15.000371932983398,
      "learning_rate": 9.649461062450651e-06,
      "loss": 1.3654,
      "step": 50000
    },
    {
      "epoch": 8.264408180128072,
      "grad_norm": 15.230158805847168,
      "learning_rate": 9.64027984355203e-06,
      "loss": 1.2959,
      "step": 50010
    },
    {
      "epoch": 8.266060731253873,
      "grad_norm": 20.123403549194336,
      "learning_rate": 9.63109862465341e-06,
      "loss": 1.4118,
      "step": 50020
    },
    {
      "epoch": 8.267713282379674,
      "grad_norm": 9.338974952697754,
      "learning_rate": 9.621917405754787e-06,
      "loss": 1.2685,
      "step": 50030
    },
    {
      "epoch": 8.269365833505475,
      "grad_norm": 18.35788917541504,
      "learning_rate": 9.612736186856168e-06,
      "loss": 1.3901,
      "step": 50040
    },
    {
      "epoch": 8.271018384631274,
      "grad_norm": 22.194643020629883,
      "learning_rate": 9.603554967957548e-06,
      "loss": 1.278,
      "step": 50050
    },
    {
      "epoch": 8.272670935757075,
      "grad_norm": 10.933758735656738,
      "learning_rate": 9.594373749058925e-06,
      "loss": 1.4638,
      "step": 50060
    },
    {
      "epoch": 8.274323486882876,
      "grad_norm": 13.151747703552246,
      "learning_rate": 9.585192530160304e-06,
      "loss": 1.4557,
      "step": 50070
    },
    {
      "epoch": 8.275976038008675,
      "grad_norm": 16.250370025634766,
      "learning_rate": 9.576011311261684e-06,
      "loss": 1.3638,
      "step": 50080
    },
    {
      "epoch": 8.277628589134476,
      "grad_norm": 14.143457412719727,
      "learning_rate": 9.566830092363063e-06,
      "loss": 1.2704,
      "step": 50090
    },
    {
      "epoch": 8.279281140260277,
      "grad_norm": 15.831094741821289,
      "learning_rate": 9.557648873464442e-06,
      "loss": 1.4091,
      "step": 50100
    },
    {
      "epoch": 8.280933691386076,
      "grad_norm": 11.630001068115234,
      "learning_rate": 9.548467654565821e-06,
      "loss": 1.3446,
      "step": 50110
    },
    {
      "epoch": 8.282586242511877,
      "grad_norm": 9.2128324508667,
      "learning_rate": 9.539286435667199e-06,
      "loss": 1.2162,
      "step": 50120
    },
    {
      "epoch": 8.284238793637678,
      "grad_norm": 16.4133243560791,
      "learning_rate": 9.530105216768578e-06,
      "loss": 1.3735,
      "step": 50130
    },
    {
      "epoch": 8.28589134476348,
      "grad_norm": 8.486006736755371,
      "learning_rate": 9.520923997869957e-06,
      "loss": 1.2581,
      "step": 50140
    },
    {
      "epoch": 8.287543895889279,
      "grad_norm": 25.598995208740234,
      "learning_rate": 9.511742778971337e-06,
      "loss": 1.2991,
      "step": 50150
    },
    {
      "epoch": 8.28919644701508,
      "grad_norm": 13.834522247314453,
      "learning_rate": 9.502561560072716e-06,
      "loss": 1.4145,
      "step": 50160
    },
    {
      "epoch": 8.29084899814088,
      "grad_norm": 13.25239086151123,
      "learning_rate": 9.493380341174095e-06,
      "loss": 1.4416,
      "step": 50170
    },
    {
      "epoch": 8.29250154926668,
      "grad_norm": 15.901601791381836,
      "learning_rate": 9.484199122275474e-06,
      "loss": 1.4722,
      "step": 50180
    },
    {
      "epoch": 8.29415410039248,
      "grad_norm": 12.437292098999023,
      "learning_rate": 9.475017903376852e-06,
      "loss": 1.3519,
      "step": 50190
    },
    {
      "epoch": 8.295806651518282,
      "grad_norm": 13.174797058105469,
      "learning_rate": 9.465836684478231e-06,
      "loss": 1.3125,
      "step": 50200
    },
    {
      "epoch": 8.297459202644081,
      "grad_norm": 18.76152229309082,
      "learning_rate": 9.456655465579612e-06,
      "loss": 1.3622,
      "step": 50210
    },
    {
      "epoch": 8.299111753769882,
      "grad_norm": 13.118831634521484,
      "learning_rate": 9.44747424668099e-06,
      "loss": 1.3004,
      "step": 50220
    },
    {
      "epoch": 8.300764304895683,
      "grad_norm": 18.371763229370117,
      "learning_rate": 9.438293027782369e-06,
      "loss": 1.3731,
      "step": 50230
    },
    {
      "epoch": 8.302416856021484,
      "grad_norm": 12.924381256103516,
      "learning_rate": 9.429111808883748e-06,
      "loss": 1.3149,
      "step": 50240
    },
    {
      "epoch": 8.304069407147283,
      "grad_norm": 10.467041015625,
      "learning_rate": 9.419930589985126e-06,
      "loss": 1.3262,
      "step": 50250
    },
    {
      "epoch": 8.305721958273084,
      "grad_norm": 18.903898239135742,
      "learning_rate": 9.410749371086505e-06,
      "loss": 1.2989,
      "step": 50260
    },
    {
      "epoch": 8.307374509398885,
      "grad_norm": 13.446491241455078,
      "learning_rate": 9.401568152187886e-06,
      "loss": 1.4296,
      "step": 50270
    },
    {
      "epoch": 8.309027060524684,
      "grad_norm": 14.688339233398438,
      "learning_rate": 9.392386933289263e-06,
      "loss": 1.2989,
      "step": 50280
    },
    {
      "epoch": 8.310679611650485,
      "grad_norm": 17.203166961669922,
      "learning_rate": 9.383205714390643e-06,
      "loss": 1.3036,
      "step": 50290
    },
    {
      "epoch": 8.312332162776286,
      "grad_norm": 83.26315307617188,
      "learning_rate": 9.374024495492022e-06,
      "loss": 1.3526,
      "step": 50300
    },
    {
      "epoch": 8.313984713902085,
      "grad_norm": 23.8687686920166,
      "learning_rate": 9.364843276593401e-06,
      "loss": 1.2074,
      "step": 50310
    },
    {
      "epoch": 8.315637265027886,
      "grad_norm": 13.990741729736328,
      "learning_rate": 9.35566205769478e-06,
      "loss": 1.3218,
      "step": 50320
    },
    {
      "epoch": 8.317289816153687,
      "grad_norm": 34.39853286743164,
      "learning_rate": 9.34648083879616e-06,
      "loss": 1.338,
      "step": 50330
    },
    {
      "epoch": 8.318942367279488,
      "grad_norm": 22.453739166259766,
      "learning_rate": 9.337299619897539e-06,
      "loss": 1.4889,
      "step": 50340
    },
    {
      "epoch": 8.320594918405288,
      "grad_norm": 10.095982551574707,
      "learning_rate": 9.328118400998916e-06,
      "loss": 1.336,
      "step": 50350
    },
    {
      "epoch": 8.322247469531089,
      "grad_norm": 74.14069366455078,
      "learning_rate": 9.318937182100296e-06,
      "loss": 1.4469,
      "step": 50360
    },
    {
      "epoch": 8.32390002065689,
      "grad_norm": 23.241098403930664,
      "learning_rate": 9.309755963201675e-06,
      "loss": 1.4088,
      "step": 50370
    },
    {
      "epoch": 8.325552571782689,
      "grad_norm": 9.130888938903809,
      "learning_rate": 9.300574744303054e-06,
      "loss": 1.3539,
      "step": 50380
    },
    {
      "epoch": 8.32720512290849,
      "grad_norm": 28.220802307128906,
      "learning_rate": 9.291393525404433e-06,
      "loss": 1.4253,
      "step": 50390
    },
    {
      "epoch": 8.32885767403429,
      "grad_norm": 14.703250885009766,
      "learning_rate": 9.282212306505813e-06,
      "loss": 1.2751,
      "step": 50400
    },
    {
      "epoch": 8.33051022516009,
      "grad_norm": 17.532804489135742,
      "learning_rate": 9.27303108760719e-06,
      "loss": 1.3413,
      "step": 50410
    },
    {
      "epoch": 8.332162776285891,
      "grad_norm": 11.652085304260254,
      "learning_rate": 9.26384986870857e-06,
      "loss": 1.2881,
      "step": 50420
    },
    {
      "epoch": 8.333815327411692,
      "grad_norm": 16.633052825927734,
      "learning_rate": 9.254668649809949e-06,
      "loss": 1.3163,
      "step": 50430
    },
    {
      "epoch": 8.335467878537493,
      "grad_norm": 17.8756046295166,
      "learning_rate": 9.245487430911328e-06,
      "loss": 1.2875,
      "step": 50440
    },
    {
      "epoch": 8.337120429663292,
      "grad_norm": 14.60150146484375,
      "learning_rate": 9.236306212012707e-06,
      "loss": 1.3227,
      "step": 50450
    },
    {
      "epoch": 8.338772980789093,
      "grad_norm": 10.519417762756348,
      "learning_rate": 9.227124993114087e-06,
      "loss": 1.1588,
      "step": 50460
    },
    {
      "epoch": 8.340425531914894,
      "grad_norm": 12.479300498962402,
      "learning_rate": 9.217943774215466e-06,
      "loss": 1.2749,
      "step": 50470
    },
    {
      "epoch": 8.342078083040693,
      "grad_norm": 15.454779624938965,
      "learning_rate": 9.208762555316843e-06,
      "loss": 1.4351,
      "step": 50480
    },
    {
      "epoch": 8.343730634166494,
      "grad_norm": 22.32270050048828,
      "learning_rate": 9.199581336418224e-06,
      "loss": 1.242,
      "step": 50490
    },
    {
      "epoch": 8.345383185292295,
      "grad_norm": 12.450278282165527,
      "learning_rate": 9.190400117519603e-06,
      "loss": 1.3128,
      "step": 50500
    },
    {
      "epoch": 8.347035736418096,
      "grad_norm": 12.18801498413086,
      "learning_rate": 9.181218898620981e-06,
      "loss": 1.3458,
      "step": 50510
    },
    {
      "epoch": 8.348688287543895,
      "grad_norm": 11.80329704284668,
      "learning_rate": 9.17203767972236e-06,
      "loss": 1.3676,
      "step": 50520
    },
    {
      "epoch": 8.350340838669696,
      "grad_norm": 16.792072296142578,
      "learning_rate": 9.16285646082374e-06,
      "loss": 1.3725,
      "step": 50530
    },
    {
      "epoch": 8.351993389795497,
      "grad_norm": 20.232120513916016,
      "learning_rate": 9.153675241925117e-06,
      "loss": 1.2889,
      "step": 50540
    },
    {
      "epoch": 8.353645940921297,
      "grad_norm": 10.46639347076416,
      "learning_rate": 9.144494023026498e-06,
      "loss": 1.3758,
      "step": 50550
    },
    {
      "epoch": 8.355298492047098,
      "grad_norm": 11.493830680847168,
      "learning_rate": 9.135312804127877e-06,
      "loss": 1.3593,
      "step": 50560
    },
    {
      "epoch": 8.356951043172899,
      "grad_norm": 9.20073413848877,
      "learning_rate": 9.126131585229255e-06,
      "loss": 1.3475,
      "step": 50570
    },
    {
      "epoch": 8.358603594298698,
      "grad_norm": 19.450620651245117,
      "learning_rate": 9.116950366330634e-06,
      "loss": 1.4109,
      "step": 50580
    },
    {
      "epoch": 8.360256145424499,
      "grad_norm": 13.948092460632324,
      "learning_rate": 9.107769147432013e-06,
      "loss": 1.3133,
      "step": 50590
    },
    {
      "epoch": 8.3619086965503,
      "grad_norm": 26.458175659179688,
      "learning_rate": 9.098587928533393e-06,
      "loss": 1.5351,
      "step": 50600
    },
    {
      "epoch": 8.3635612476761,
      "grad_norm": 11.807082176208496,
      "learning_rate": 9.089406709634772e-06,
      "loss": 1.3635,
      "step": 50610
    },
    {
      "epoch": 8.3652137988019,
      "grad_norm": 13.108859062194824,
      "learning_rate": 9.080225490736151e-06,
      "loss": 1.411,
      "step": 50620
    },
    {
      "epoch": 8.366866349927701,
      "grad_norm": 14.530641555786133,
      "learning_rate": 9.07104427183753e-06,
      "loss": 1.4175,
      "step": 50630
    },
    {
      "epoch": 8.368518901053502,
      "grad_norm": 21.881153106689453,
      "learning_rate": 9.061863052938908e-06,
      "loss": 1.4125,
      "step": 50640
    },
    {
      "epoch": 8.370171452179301,
      "grad_norm": 15.157549858093262,
      "learning_rate": 9.052681834040287e-06,
      "loss": 1.2924,
      "step": 50650
    },
    {
      "epoch": 8.371824003305102,
      "grad_norm": 15.397344589233398,
      "learning_rate": 9.043500615141666e-06,
      "loss": 1.2743,
      "step": 50660
    },
    {
      "epoch": 8.373476554430903,
      "grad_norm": 13.956780433654785,
      "learning_rate": 9.034319396243046e-06,
      "loss": 1.2906,
      "step": 50670
    },
    {
      "epoch": 8.375129105556702,
      "grad_norm": 15.764337539672852,
      "learning_rate": 9.025138177344425e-06,
      "loss": 1.362,
      "step": 50680
    },
    {
      "epoch": 8.376781656682503,
      "grad_norm": 18.606061935424805,
      "learning_rate": 9.015956958445804e-06,
      "loss": 1.3092,
      "step": 50690
    },
    {
      "epoch": 8.378434207808304,
      "grad_norm": 17.503150939941406,
      "learning_rate": 9.006775739547182e-06,
      "loss": 1.3202,
      "step": 50700
    },
    {
      "epoch": 8.380086758934105,
      "grad_norm": 7.5358967781066895,
      "learning_rate": 8.997594520648561e-06,
      "loss": 1.3524,
      "step": 50710
    },
    {
      "epoch": 8.381739310059904,
      "grad_norm": 19.432466506958008,
      "learning_rate": 8.988413301749942e-06,
      "loss": 1.344,
      "step": 50720
    },
    {
      "epoch": 8.383391861185705,
      "grad_norm": 10.042672157287598,
      "learning_rate": 8.97923208285132e-06,
      "loss": 1.3504,
      "step": 50730
    },
    {
      "epoch": 8.385044412311506,
      "grad_norm": 14.225571632385254,
      "learning_rate": 8.970050863952699e-06,
      "loss": 1.2284,
      "step": 50740
    },
    {
      "epoch": 8.386696963437306,
      "grad_norm": 11.462421417236328,
      "learning_rate": 8.960869645054078e-06,
      "loss": 1.371,
      "step": 50750
    },
    {
      "epoch": 8.388349514563107,
      "grad_norm": 9.020613670349121,
      "learning_rate": 8.951688426155457e-06,
      "loss": 1.262,
      "step": 50760
    },
    {
      "epoch": 8.390002065688908,
      "grad_norm": 17.618812561035156,
      "learning_rate": 8.942507207256835e-06,
      "loss": 1.393,
      "step": 50770
    },
    {
      "epoch": 8.391654616814709,
      "grad_norm": 19.001405715942383,
      "learning_rate": 8.933325988358216e-06,
      "loss": 1.3542,
      "step": 50780
    },
    {
      "epoch": 8.393307167940508,
      "grad_norm": 19.39535140991211,
      "learning_rate": 8.924144769459595e-06,
      "loss": 1.4777,
      "step": 50790
    },
    {
      "epoch": 8.394959719066309,
      "grad_norm": 11.611642837524414,
      "learning_rate": 8.914963550560972e-06,
      "loss": 1.2758,
      "step": 50800
    },
    {
      "epoch": 8.39661227019211,
      "grad_norm": 25.300411224365234,
      "learning_rate": 8.905782331662352e-06,
      "loss": 1.4082,
      "step": 50810
    },
    {
      "epoch": 8.398264821317909,
      "grad_norm": 11.443560600280762,
      "learning_rate": 8.896601112763731e-06,
      "loss": 1.2994,
      "step": 50820
    },
    {
      "epoch": 8.39991737244371,
      "grad_norm": 11.911940574645996,
      "learning_rate": 8.887419893865109e-06,
      "loss": 1.4266,
      "step": 50830
    },
    {
      "epoch": 8.401569923569511,
      "grad_norm": 10.640558242797852,
      "learning_rate": 8.87823867496649e-06,
      "loss": 1.3858,
      "step": 50840
    },
    {
      "epoch": 8.40322247469531,
      "grad_norm": 14.38443660736084,
      "learning_rate": 8.869057456067869e-06,
      "loss": 1.3228,
      "step": 50850
    },
    {
      "epoch": 8.404875025821111,
      "grad_norm": 11.794403076171875,
      "learning_rate": 8.859876237169246e-06,
      "loss": 1.3956,
      "step": 50860
    },
    {
      "epoch": 8.406527576946912,
      "grad_norm": 10.348832130432129,
      "learning_rate": 8.850695018270625e-06,
      "loss": 1.4676,
      "step": 50870
    },
    {
      "epoch": 8.408180128072713,
      "grad_norm": 10.430915832519531,
      "learning_rate": 8.841513799372005e-06,
      "loss": 1.3324,
      "step": 50880
    },
    {
      "epoch": 8.409832679198512,
      "grad_norm": 11.54885482788086,
      "learning_rate": 8.832332580473384e-06,
      "loss": 1.4735,
      "step": 50890
    },
    {
      "epoch": 8.411485230324313,
      "grad_norm": 19.26331901550293,
      "learning_rate": 8.823151361574763e-06,
      "loss": 1.2549,
      "step": 50900
    },
    {
      "epoch": 8.413137781450114,
      "grad_norm": 12.017107963562012,
      "learning_rate": 8.813970142676142e-06,
      "loss": 1.3315,
      "step": 50910
    },
    {
      "epoch": 8.414790332575913,
      "grad_norm": 13.017470359802246,
      "learning_rate": 8.804788923777522e-06,
      "loss": 1.2567,
      "step": 50920
    },
    {
      "epoch": 8.416442883701714,
      "grad_norm": 14.95610523223877,
      "learning_rate": 8.7956077048789e-06,
      "loss": 1.3519,
      "step": 50930
    },
    {
      "epoch": 8.418095434827515,
      "grad_norm": 18.328262329101562,
      "learning_rate": 8.786426485980279e-06,
      "loss": 1.3535,
      "step": 50940
    },
    {
      "epoch": 8.419747985953315,
      "grad_norm": 11.61627197265625,
      "learning_rate": 8.77724526708166e-06,
      "loss": 1.4133,
      "step": 50950
    },
    {
      "epoch": 8.421400537079116,
      "grad_norm": 17.8658390045166,
      "learning_rate": 8.768064048183037e-06,
      "loss": 1.3322,
      "step": 50960
    },
    {
      "epoch": 8.423053088204917,
      "grad_norm": 12.533629417419434,
      "learning_rate": 8.758882829284416e-06,
      "loss": 1.2034,
      "step": 50970
    },
    {
      "epoch": 8.424705639330718,
      "grad_norm": 10.913763046264648,
      "learning_rate": 8.749701610385796e-06,
      "loss": 1.3928,
      "step": 50980
    },
    {
      "epoch": 8.426358190456517,
      "grad_norm": 12.7138090133667,
      "learning_rate": 8.740520391487173e-06,
      "loss": 1.2596,
      "step": 50990
    },
    {
      "epoch": 8.428010741582318,
      "grad_norm": 69.10160827636719,
      "learning_rate": 8.731339172588552e-06,
      "loss": 1.3902,
      "step": 51000
    },
    {
      "epoch": 8.429663292708119,
      "grad_norm": 18.809534072875977,
      "learning_rate": 8.722157953689933e-06,
      "loss": 1.343,
      "step": 51010
    },
    {
      "epoch": 8.431315843833918,
      "grad_norm": 17.767322540283203,
      "learning_rate": 8.71297673479131e-06,
      "loss": 1.241,
      "step": 51020
    },
    {
      "epoch": 8.432968394959719,
      "grad_norm": 18.070133209228516,
      "learning_rate": 8.70379551589269e-06,
      "loss": 1.2567,
      "step": 51030
    },
    {
      "epoch": 8.43462094608552,
      "grad_norm": 17.62630271911621,
      "learning_rate": 8.69461429699407e-06,
      "loss": 1.3031,
      "step": 51040
    },
    {
      "epoch": 8.43627349721132,
      "grad_norm": 19.49014663696289,
      "learning_rate": 8.685433078095449e-06,
      "loss": 1.4,
      "step": 51050
    },
    {
      "epoch": 8.43792604833712,
      "grad_norm": 22.65168571472168,
      "learning_rate": 8.676251859196826e-06,
      "loss": 1.3422,
      "step": 51060
    },
    {
      "epoch": 8.439578599462921,
      "grad_norm": 15.873695373535156,
      "learning_rate": 8.667070640298207e-06,
      "loss": 1.5412,
      "step": 51070
    },
    {
      "epoch": 8.441231150588722,
      "grad_norm": 20.69378089904785,
      "learning_rate": 8.657889421399586e-06,
      "loss": 1.2421,
      "step": 51080
    },
    {
      "epoch": 8.442883701714521,
      "grad_norm": 14.681087493896484,
      "learning_rate": 8.648708202500964e-06,
      "loss": 1.2936,
      "step": 51090
    },
    {
      "epoch": 8.444536252840322,
      "grad_norm": 23.839170455932617,
      "learning_rate": 8.639526983602343e-06,
      "loss": 1.3616,
      "step": 51100
    },
    {
      "epoch": 8.446188803966123,
      "grad_norm": 12.453177452087402,
      "learning_rate": 8.630345764703722e-06,
      "loss": 1.4633,
      "step": 51110
    },
    {
      "epoch": 8.447841355091922,
      "grad_norm": 10.4721040725708,
      "learning_rate": 8.621164545805102e-06,
      "loss": 1.3924,
      "step": 51120
    },
    {
      "epoch": 8.449493906217723,
      "grad_norm": 11.576587677001953,
      "learning_rate": 8.611983326906481e-06,
      "loss": 1.2731,
      "step": 51130
    },
    {
      "epoch": 8.451146457343524,
      "grad_norm": 16.914947509765625,
      "learning_rate": 8.60280210800786e-06,
      "loss": 1.4043,
      "step": 51140
    },
    {
      "epoch": 8.452799008469324,
      "grad_norm": 14.741705894470215,
      "learning_rate": 8.593620889109238e-06,
      "loss": 1.3298,
      "step": 51150
    },
    {
      "epoch": 8.454451559595125,
      "grad_norm": 10.166900634765625,
      "learning_rate": 8.584439670210617e-06,
      "loss": 1.3858,
      "step": 51160
    },
    {
      "epoch": 8.456104110720926,
      "grad_norm": 16.264799118041992,
      "learning_rate": 8.575258451311996e-06,
      "loss": 1.3903,
      "step": 51170
    },
    {
      "epoch": 8.457756661846727,
      "grad_norm": 11.90115737915039,
      "learning_rate": 8.566077232413375e-06,
      "loss": 1.3022,
      "step": 51180
    },
    {
      "epoch": 8.459409212972526,
      "grad_norm": 14.521602630615234,
      "learning_rate": 8.556896013514755e-06,
      "loss": 1.2753,
      "step": 51190
    },
    {
      "epoch": 8.461061764098327,
      "grad_norm": 29.15642547607422,
      "learning_rate": 8.547714794616134e-06,
      "loss": 1.3347,
      "step": 51200
    },
    {
      "epoch": 8.462714315224128,
      "grad_norm": 20.147611618041992,
      "learning_rate": 8.538533575717513e-06,
      "loss": 1.4416,
      "step": 51210
    },
    {
      "epoch": 8.464366866349927,
      "grad_norm": 30.82805633544922,
      "learning_rate": 8.52935235681889e-06,
      "loss": 1.3531,
      "step": 51220
    },
    {
      "epoch": 8.466019417475728,
      "grad_norm": 21.31821632385254,
      "learning_rate": 8.52017113792027e-06,
      "loss": 1.2286,
      "step": 51230
    },
    {
      "epoch": 8.467671968601529,
      "grad_norm": 19.59657096862793,
      "learning_rate": 8.510989919021651e-06,
      "loss": 1.3915,
      "step": 51240
    },
    {
      "epoch": 8.46932451972733,
      "grad_norm": 23.90947151184082,
      "learning_rate": 8.501808700123028e-06,
      "loss": 1.3517,
      "step": 51250
    },
    {
      "epoch": 8.47097707085313,
      "grad_norm": 22.817235946655273,
      "learning_rate": 8.492627481224408e-06,
      "loss": 1.3708,
      "step": 51260
    },
    {
      "epoch": 8.47262962197893,
      "grad_norm": 11.758684158325195,
      "learning_rate": 8.483446262325787e-06,
      "loss": 1.2679,
      "step": 51270
    },
    {
      "epoch": 8.474282173104731,
      "grad_norm": 9.888134002685547,
      "learning_rate": 8.474265043427164e-06,
      "loss": 1.3361,
      "step": 51280
    },
    {
      "epoch": 8.47593472423053,
      "grad_norm": 11.831820487976074,
      "learning_rate": 8.465083824528545e-06,
      "loss": 1.274,
      "step": 51290
    },
    {
      "epoch": 8.477587275356331,
      "grad_norm": 42.835636138916016,
      "learning_rate": 8.455902605629925e-06,
      "loss": 1.3636,
      "step": 51300
    },
    {
      "epoch": 8.479239826482132,
      "grad_norm": 10.630398750305176,
      "learning_rate": 8.446721386731302e-06,
      "loss": 1.1861,
      "step": 51310
    },
    {
      "epoch": 8.480892377607931,
      "grad_norm": 53.76140594482422,
      "learning_rate": 8.437540167832681e-06,
      "loss": 1.3159,
      "step": 51320
    },
    {
      "epoch": 8.482544928733732,
      "grad_norm": 25.955413818359375,
      "learning_rate": 8.42835894893406e-06,
      "loss": 1.2953,
      "step": 51330
    },
    {
      "epoch": 8.484197479859533,
      "grad_norm": 15.469341278076172,
      "learning_rate": 8.41917773003544e-06,
      "loss": 1.2313,
      "step": 51340
    },
    {
      "epoch": 8.485850030985334,
      "grad_norm": 15.408190727233887,
      "learning_rate": 8.40999651113682e-06,
      "loss": 1.4626,
      "step": 51350
    },
    {
      "epoch": 8.487502582111134,
      "grad_norm": 18.029531478881836,
      "learning_rate": 8.400815292238198e-06,
      "loss": 1.3921,
      "step": 51360
    },
    {
      "epoch": 8.489155133236935,
      "grad_norm": 12.669473648071289,
      "learning_rate": 8.391634073339578e-06,
      "loss": 1.203,
      "step": 51370
    },
    {
      "epoch": 8.490807684362736,
      "grad_norm": 20.233211517333984,
      "learning_rate": 8.382452854440955e-06,
      "loss": 1.5235,
      "step": 51380
    },
    {
      "epoch": 8.492460235488535,
      "grad_norm": 18.365198135375977,
      "learning_rate": 8.373271635542335e-06,
      "loss": 1.5077,
      "step": 51390
    },
    {
      "epoch": 8.494112786614336,
      "grad_norm": 9.256744384765625,
      "learning_rate": 8.364090416643714e-06,
      "loss": 1.328,
      "step": 51400
    },
    {
      "epoch": 8.495765337740137,
      "grad_norm": 13.671814918518066,
      "learning_rate": 8.354909197745093e-06,
      "loss": 1.4123,
      "step": 51410
    },
    {
      "epoch": 8.497417888865936,
      "grad_norm": 48.41630172729492,
      "learning_rate": 8.345727978846472e-06,
      "loss": 1.3484,
      "step": 51420
    },
    {
      "epoch": 8.499070439991737,
      "grad_norm": 19.150592803955078,
      "learning_rate": 8.336546759947852e-06,
      "loss": 1.3481,
      "step": 51430
    },
    {
      "epoch": 8.500722991117538,
      "grad_norm": 20.59627914428711,
      "learning_rate": 8.327365541049229e-06,
      "loss": 1.281,
      "step": 51440
    },
    {
      "epoch": 8.502375542243339,
      "grad_norm": 16.762386322021484,
      "learning_rate": 8.318184322150608e-06,
      "loss": 1.4107,
      "step": 51450
    },
    {
      "epoch": 8.504028093369138,
      "grad_norm": 15.31123161315918,
      "learning_rate": 8.309003103251988e-06,
      "loss": 1.3289,
      "step": 51460
    },
    {
      "epoch": 8.505680644494939,
      "grad_norm": 13.639659881591797,
      "learning_rate": 8.299821884353367e-06,
      "loss": 1.2842,
      "step": 51470
    },
    {
      "epoch": 8.50733319562074,
      "grad_norm": 15.621134757995605,
      "learning_rate": 8.290640665454746e-06,
      "loss": 1.3049,
      "step": 51480
    },
    {
      "epoch": 8.50898574674654,
      "grad_norm": 11.561347961425781,
      "learning_rate": 8.281459446556125e-06,
      "loss": 1.3786,
      "step": 51490
    },
    {
      "epoch": 8.51063829787234,
      "grad_norm": 6.681066989898682,
      "learning_rate": 8.272278227657505e-06,
      "loss": 1.1806,
      "step": 51500
    },
    {
      "epoch": 8.512290848998141,
      "grad_norm": 13.842896461486816,
      "learning_rate": 8.263097008758882e-06,
      "loss": 1.4607,
      "step": 51510
    },
    {
      "epoch": 8.513943400123942,
      "grad_norm": 16.961971282958984,
      "learning_rate": 8.253915789860263e-06,
      "loss": 1.2977,
      "step": 51520
    },
    {
      "epoch": 8.515595951249741,
      "grad_norm": 11.850485801696777,
      "learning_rate": 8.244734570961642e-06,
      "loss": 1.3939,
      "step": 51530
    },
    {
      "epoch": 8.517248502375542,
      "grad_norm": 20.252544403076172,
      "learning_rate": 8.23555335206302e-06,
      "loss": 1.4132,
      "step": 51540
    },
    {
      "epoch": 8.518901053501343,
      "grad_norm": 11.936491966247559,
      "learning_rate": 8.226372133164399e-06,
      "loss": 1.2569,
      "step": 51550
    },
    {
      "epoch": 8.520553604627143,
      "grad_norm": 15.113805770874023,
      "learning_rate": 8.217190914265778e-06,
      "loss": 1.3359,
      "step": 51560
    },
    {
      "epoch": 8.522206155752944,
      "grad_norm": 14.202859878540039,
      "learning_rate": 8.208009695367158e-06,
      "loss": 1.2987,
      "step": 51570
    },
    {
      "epoch": 8.523858706878745,
      "grad_norm": 22.607847213745117,
      "learning_rate": 8.198828476468537e-06,
      "loss": 1.3453,
      "step": 51580
    },
    {
      "epoch": 8.525511258004544,
      "grad_norm": 12.42171859741211,
      "learning_rate": 8.189647257569916e-06,
      "loss": 1.3669,
      "step": 51590
    },
    {
      "epoch": 8.527163809130345,
      "grad_norm": 19.58823585510254,
      "learning_rate": 8.180466038671294e-06,
      "loss": 1.2657,
      "step": 51600
    },
    {
      "epoch": 8.528816360256146,
      "grad_norm": 8.663179397583008,
      "learning_rate": 8.171284819772673e-06,
      "loss": 1.2291,
      "step": 51610
    },
    {
      "epoch": 8.530468911381945,
      "grad_norm": 52.90243148803711,
      "learning_rate": 8.162103600874052e-06,
      "loss": 1.2141,
      "step": 51620
    },
    {
      "epoch": 8.532121462507746,
      "grad_norm": 17.209096908569336,
      "learning_rate": 8.152922381975431e-06,
      "loss": 1.3594,
      "step": 51630
    },
    {
      "epoch": 8.533774013633547,
      "grad_norm": 14.850648880004883,
      "learning_rate": 8.14374116307681e-06,
      "loss": 1.3939,
      "step": 51640
    },
    {
      "epoch": 8.535426564759348,
      "grad_norm": 11.91596794128418,
      "learning_rate": 8.13455994417819e-06,
      "loss": 1.4166,
      "step": 51650
    },
    {
      "epoch": 8.537079115885147,
      "grad_norm": 19.062267303466797,
      "learning_rate": 8.125378725279569e-06,
      "loss": 1.2763,
      "step": 51660
    },
    {
      "epoch": 8.538731667010948,
      "grad_norm": 39.40817642211914,
      "learning_rate": 8.116197506380947e-06,
      "loss": 1.4173,
      "step": 51670
    },
    {
      "epoch": 8.540384218136749,
      "grad_norm": 9.898063659667969,
      "learning_rate": 8.107016287482326e-06,
      "loss": 1.2107,
      "step": 51680
    },
    {
      "epoch": 8.542036769262548,
      "grad_norm": 19.64806365966797,
      "learning_rate": 8.097835068583707e-06,
      "loss": 1.4224,
      "step": 51690
    },
    {
      "epoch": 8.54368932038835,
      "grad_norm": 12.104853630065918,
      "learning_rate": 8.088653849685084e-06,
      "loss": 1.3676,
      "step": 51700
    },
    {
      "epoch": 8.54534187151415,
      "grad_norm": 20.285003662109375,
      "learning_rate": 8.079472630786464e-06,
      "loss": 1.3047,
      "step": 51710
    },
    {
      "epoch": 8.546994422639951,
      "grad_norm": 18.16466522216797,
      "learning_rate": 8.070291411887843e-06,
      "loss": 1.494,
      "step": 51720
    },
    {
      "epoch": 8.54864697376575,
      "grad_norm": 14.75426197052002,
      "learning_rate": 8.061110192989222e-06,
      "loss": 1.3873,
      "step": 51730
    },
    {
      "epoch": 8.550299524891551,
      "grad_norm": 13.810311317443848,
      "learning_rate": 8.0519289740906e-06,
      "loss": 1.3675,
      "step": 51740
    },
    {
      "epoch": 8.551952076017352,
      "grad_norm": 10.323554992675781,
      "learning_rate": 8.04274775519198e-06,
      "loss": 1.3622,
      "step": 51750
    },
    {
      "epoch": 8.553604627143152,
      "grad_norm": 13.959274291992188,
      "learning_rate": 8.033566536293358e-06,
      "loss": 1.2956,
      "step": 51760
    },
    {
      "epoch": 8.555257178268953,
      "grad_norm": 12.601279258728027,
      "learning_rate": 8.024385317394737e-06,
      "loss": 1.2695,
      "step": 51770
    },
    {
      "epoch": 8.556909729394754,
      "grad_norm": 11.207502365112305,
      "learning_rate": 8.015204098496117e-06,
      "loss": 1.3684,
      "step": 51780
    },
    {
      "epoch": 8.558562280520553,
      "grad_norm": 28.540502548217773,
      "learning_rate": 8.006022879597496e-06,
      "loss": 1.319,
      "step": 51790
    },
    {
      "epoch": 8.560214831646354,
      "grad_norm": 15.684926986694336,
      "learning_rate": 7.996841660698874e-06,
      "loss": 1.3478,
      "step": 51800
    },
    {
      "epoch": 8.561867382772155,
      "grad_norm": 9.565122604370117,
      "learning_rate": 7.987660441800254e-06,
      "loss": 1.2809,
      "step": 51810
    },
    {
      "epoch": 8.563519933897956,
      "grad_norm": 12.698163986206055,
      "learning_rate": 7.978479222901634e-06,
      "loss": 1.2952,
      "step": 51820
    },
    {
      "epoch": 8.565172485023755,
      "grad_norm": 30.729351043701172,
      "learning_rate": 7.969298004003011e-06,
      "loss": 1.5078,
      "step": 51830
    },
    {
      "epoch": 8.566825036149556,
      "grad_norm": 14.974106788635254,
      "learning_rate": 7.96011678510439e-06,
      "loss": 1.3052,
      "step": 51840
    },
    {
      "epoch": 8.568477587275357,
      "grad_norm": 6.864882946014404,
      "learning_rate": 7.95093556620577e-06,
      "loss": 1.3145,
      "step": 51850
    },
    {
      "epoch": 8.570130138401156,
      "grad_norm": 17.064834594726562,
      "learning_rate": 7.941754347307149e-06,
      "loss": 1.3939,
      "step": 51860
    },
    {
      "epoch": 8.571782689526957,
      "grad_norm": 9.994977951049805,
      "learning_rate": 7.932573128408528e-06,
      "loss": 1.3526,
      "step": 51870
    },
    {
      "epoch": 8.573435240652758,
      "grad_norm": 14.07989501953125,
      "learning_rate": 7.923391909509907e-06,
      "loss": 1.3424,
      "step": 51880
    },
    {
      "epoch": 8.575087791778557,
      "grad_norm": 12.24575138092041,
      "learning_rate": 7.914210690611287e-06,
      "loss": 1.2845,
      "step": 51890
    },
    {
      "epoch": 8.576740342904358,
      "grad_norm": 13.51680850982666,
      "learning_rate": 7.905029471712664e-06,
      "loss": 1.3452,
      "step": 51900
    },
    {
      "epoch": 8.57839289403016,
      "grad_norm": 13.959500312805176,
      "learning_rate": 7.895848252814044e-06,
      "loss": 1.2149,
      "step": 51910
    },
    {
      "epoch": 8.58004544515596,
      "grad_norm": 28.394367218017578,
      "learning_rate": 7.886667033915423e-06,
      "loss": 1.463,
      "step": 51920
    },
    {
      "epoch": 8.58169799628176,
      "grad_norm": 14.02746868133545,
      "learning_rate": 7.877485815016802e-06,
      "loss": 1.4335,
      "step": 51930
    },
    {
      "epoch": 8.58335054740756,
      "grad_norm": 11.631068229675293,
      "learning_rate": 7.868304596118181e-06,
      "loss": 1.2435,
      "step": 51940
    },
    {
      "epoch": 8.585003098533361,
      "grad_norm": 13.741445541381836,
      "learning_rate": 7.85912337721956e-06,
      "loss": 1.2622,
      "step": 51950
    },
    {
      "epoch": 8.58665564965916,
      "grad_norm": 21.316083908081055,
      "learning_rate": 7.849942158320938e-06,
      "loss": 1.484,
      "step": 51960
    },
    {
      "epoch": 8.588308200784962,
      "grad_norm": 17.675111770629883,
      "learning_rate": 7.840760939422317e-06,
      "loss": 1.38,
      "step": 51970
    },
    {
      "epoch": 8.589960751910763,
      "grad_norm": 21.218908309936523,
      "learning_rate": 7.831579720523698e-06,
      "loss": 1.247,
      "step": 51980
    },
    {
      "epoch": 8.591613303036564,
      "grad_norm": 18.68570899963379,
      "learning_rate": 7.822398501625076e-06,
      "loss": 1.3275,
      "step": 51990
    },
    {
      "epoch": 8.593265854162363,
      "grad_norm": 12.028681755065918,
      "learning_rate": 7.813217282726455e-06,
      "loss": 1.2789,
      "step": 52000
    },
    {
      "epoch": 8.594918405288164,
      "grad_norm": 16.868160247802734,
      "learning_rate": 7.804036063827834e-06,
      "loss": 1.1928,
      "step": 52010
    },
    {
      "epoch": 8.596570956413965,
      "grad_norm": 14.895330429077148,
      "learning_rate": 7.794854844929214e-06,
      "loss": 1.3861,
      "step": 52020
    },
    {
      "epoch": 8.598223507539764,
      "grad_norm": 17.00758934020996,
      "learning_rate": 7.785673626030591e-06,
      "loss": 1.3761,
      "step": 52030
    },
    {
      "epoch": 8.599876058665565,
      "grad_norm": 21.712890625,
      "learning_rate": 7.776492407131972e-06,
      "loss": 1.2694,
      "step": 52040
    },
    {
      "epoch": 8.601528609791366,
      "grad_norm": 21.000192642211914,
      "learning_rate": 7.767311188233351e-06,
      "loss": 1.3491,
      "step": 52050
    },
    {
      "epoch": 8.603181160917165,
      "grad_norm": 19.828134536743164,
      "learning_rate": 7.758129969334729e-06,
      "loss": 1.2993,
      "step": 52060
    },
    {
      "epoch": 8.604833712042966,
      "grad_norm": 9.904111862182617,
      "learning_rate": 7.748948750436108e-06,
      "loss": 1.3062,
      "step": 52070
    },
    {
      "epoch": 8.606486263168767,
      "grad_norm": 19.431184768676758,
      "learning_rate": 7.739767531537487e-06,
      "loss": 1.3353,
      "step": 52080
    },
    {
      "epoch": 8.608138814294568,
      "grad_norm": 16.284452438354492,
      "learning_rate": 7.730586312638867e-06,
      "loss": 1.3189,
      "step": 52090
    },
    {
      "epoch": 8.609791365420367,
      "grad_norm": 20.377105712890625,
      "learning_rate": 7.721405093740246e-06,
      "loss": 1.2256,
      "step": 52100
    },
    {
      "epoch": 8.611443916546168,
      "grad_norm": 16.32294273376465,
      "learning_rate": 7.712223874841625e-06,
      "loss": 1.3875,
      "step": 52110
    },
    {
      "epoch": 8.61309646767197,
      "grad_norm": 17.102628707885742,
      "learning_rate": 7.703042655943003e-06,
      "loss": 1.3282,
      "step": 52120
    },
    {
      "epoch": 8.614749018797768,
      "grad_norm": 7.808797359466553,
      "learning_rate": 7.693861437044382e-06,
      "loss": 1.2726,
      "step": 52130
    },
    {
      "epoch": 8.61640156992357,
      "grad_norm": 44.669883728027344,
      "learning_rate": 7.684680218145761e-06,
      "loss": 1.3213,
      "step": 52140
    },
    {
      "epoch": 8.61805412104937,
      "grad_norm": 10.58360767364502,
      "learning_rate": 7.67549899924714e-06,
      "loss": 1.2849,
      "step": 52150
    },
    {
      "epoch": 8.61970667217517,
      "grad_norm": 11.978372573852539,
      "learning_rate": 7.66631778034852e-06,
      "loss": 1.392,
      "step": 52160
    },
    {
      "epoch": 8.62135922330097,
      "grad_norm": 19.33893585205078,
      "learning_rate": 7.657136561449899e-06,
      "loss": 1.3544,
      "step": 52170
    },
    {
      "epoch": 8.623011774426772,
      "grad_norm": 15.807573318481445,
      "learning_rate": 7.647955342551278e-06,
      "loss": 1.4272,
      "step": 52180
    },
    {
      "epoch": 8.624664325552573,
      "grad_norm": 22.08311653137207,
      "learning_rate": 7.638774123652656e-06,
      "loss": 1.3413,
      "step": 52190
    },
    {
      "epoch": 8.626316876678372,
      "grad_norm": 18.999937057495117,
      "learning_rate": 7.629592904754035e-06,
      "loss": 1.4418,
      "step": 52200
    },
    {
      "epoch": 8.627969427804173,
      "grad_norm": 18.321889877319336,
      "learning_rate": 7.620411685855415e-06,
      "loss": 1.4389,
      "step": 52210
    },
    {
      "epoch": 8.629621978929974,
      "grad_norm": 18.07795524597168,
      "learning_rate": 7.6112304669567934e-06,
      "loss": 1.2659,
      "step": 52220
    },
    {
      "epoch": 8.631274530055773,
      "grad_norm": 17.448719024658203,
      "learning_rate": 7.602049248058173e-06,
      "loss": 1.3933,
      "step": 52230
    },
    {
      "epoch": 8.632927081181574,
      "grad_norm": 10.875787734985352,
      "learning_rate": 7.592868029159551e-06,
      "loss": 1.313,
      "step": 52240
    },
    {
      "epoch": 8.634579632307375,
      "grad_norm": 17.80232810974121,
      "learning_rate": 7.58368681026093e-06,
      "loss": 1.3124,
      "step": 52250
    },
    {
      "epoch": 8.636232183433176,
      "grad_norm": 21.202545166015625,
      "learning_rate": 7.5745055913623104e-06,
      "loss": 1.4396,
      "step": 52260
    },
    {
      "epoch": 8.637884734558975,
      "grad_norm": 22.008880615234375,
      "learning_rate": 7.565324372463689e-06,
      "loss": 1.3628,
      "step": 52270
    },
    {
      "epoch": 8.639537285684776,
      "grad_norm": 31.674184799194336,
      "learning_rate": 7.556143153565068e-06,
      "loss": 1.3047,
      "step": 52280
    },
    {
      "epoch": 8.641189836810577,
      "grad_norm": 13.10922908782959,
      "learning_rate": 7.5469619346664465e-06,
      "loss": 1.3011,
      "step": 52290
    },
    {
      "epoch": 8.642842387936376,
      "grad_norm": 43.33464813232422,
      "learning_rate": 7.537780715767826e-06,
      "loss": 1.4427,
      "step": 52300
    },
    {
      "epoch": 8.644494939062177,
      "grad_norm": 11.043828964233398,
      "learning_rate": 7.528599496869204e-06,
      "loss": 1.3129,
      "step": 52310
    },
    {
      "epoch": 8.646147490187978,
      "grad_norm": 17.55091667175293,
      "learning_rate": 7.519418277970584e-06,
      "loss": 1.3026,
      "step": 52320
    },
    {
      "epoch": 8.647800041313777,
      "grad_norm": 11.531242370605469,
      "learning_rate": 7.5102370590719635e-06,
      "loss": 1.343,
      "step": 52330
    },
    {
      "epoch": 8.649452592439578,
      "grad_norm": 7.957118988037109,
      "learning_rate": 7.501055840173342e-06,
      "loss": 1.2908,
      "step": 52340
    },
    {
      "epoch": 8.65110514356538,
      "grad_norm": 15.16044807434082,
      "learning_rate": 7.49187462127472e-06,
      "loss": 1.4025,
      "step": 52350
    },
    {
      "epoch": 8.652757694691179,
      "grad_norm": 12.470380783081055,
      "learning_rate": 7.4826934023760995e-06,
      "loss": 1.2541,
      "step": 52360
    },
    {
      "epoch": 8.65441024581698,
      "grad_norm": 14.842360496520996,
      "learning_rate": 7.473512183477478e-06,
      "loss": 1.4249,
      "step": 52370
    },
    {
      "epoch": 8.65606279694278,
      "grad_norm": 10.60826587677002,
      "learning_rate": 7.464330964578858e-06,
      "loss": 1.2648,
      "step": 52380
    },
    {
      "epoch": 8.657715348068582,
      "grad_norm": 20.237024307250977,
      "learning_rate": 7.455149745680237e-06,
      "loss": 1.3097,
      "step": 52390
    },
    {
      "epoch": 8.65936789919438,
      "grad_norm": 15.583151817321777,
      "learning_rate": 7.445968526781616e-06,
      "loss": 1.3364,
      "step": 52400
    },
    {
      "epoch": 8.661020450320182,
      "grad_norm": 22.51787757873535,
      "learning_rate": 7.436787307882995e-06,
      "loss": 1.3017,
      "step": 52410
    },
    {
      "epoch": 8.662673001445983,
      "grad_norm": 8.777755737304688,
      "learning_rate": 7.427606088984373e-06,
      "loss": 1.2984,
      "step": 52420
    },
    {
      "epoch": 8.664325552571782,
      "grad_norm": 43.127593994140625,
      "learning_rate": 7.4184248700857526e-06,
      "loss": 1.3406,
      "step": 52430
    },
    {
      "epoch": 8.665978103697583,
      "grad_norm": 10.117231369018555,
      "learning_rate": 7.409243651187133e-06,
      "loss": 1.261,
      "step": 52440
    },
    {
      "epoch": 8.667630654823384,
      "grad_norm": 12.074843406677246,
      "learning_rate": 7.400062432288511e-06,
      "loss": 1.4873,
      "step": 52450
    },
    {
      "epoch": 8.669283205949185,
      "grad_norm": 13.140725135803223,
      "learning_rate": 7.39088121338989e-06,
      "loss": 1.3359,
      "step": 52460
    },
    {
      "epoch": 8.670935757074984,
      "grad_norm": 15.065147399902344,
      "learning_rate": 7.381699994491269e-06,
      "loss": 1.2755,
      "step": 52470
    },
    {
      "epoch": 8.672588308200785,
      "grad_norm": 10.053617477416992,
      "learning_rate": 7.372518775592648e-06,
      "loss": 1.3763,
      "step": 52480
    },
    {
      "epoch": 8.674240859326586,
      "grad_norm": 15.998075485229492,
      "learning_rate": 7.363337556694028e-06,
      "loss": 1.3625,
      "step": 52490
    },
    {
      "epoch": 8.675893410452385,
      "grad_norm": 11.436149597167969,
      "learning_rate": 7.3541563377954064e-06,
      "loss": 1.3587,
      "step": 52500
    },
    {
      "epoch": 8.677545961578186,
      "grad_norm": 12.246932983398438,
      "learning_rate": 7.344975118896785e-06,
      "loss": 1.3444,
      "step": 52510
    },
    {
      "epoch": 8.679198512703987,
      "grad_norm": 10.678253173828125,
      "learning_rate": 7.335793899998164e-06,
      "loss": 1.3673,
      "step": 52520
    },
    {
      "epoch": 8.680851063829786,
      "grad_norm": 11.65112018585205,
      "learning_rate": 7.3266126810995425e-06,
      "loss": 1.3642,
      "step": 52530
    },
    {
      "epoch": 8.682503614955587,
      "grad_norm": 17.31778335571289,
      "learning_rate": 7.317431462200922e-06,
      "loss": 1.3697,
      "step": 52540
    },
    {
      "epoch": 8.684156166081388,
      "grad_norm": 14.801467895507812,
      "learning_rate": 7.308250243302302e-06,
      "loss": 1.3962,
      "step": 52550
    },
    {
      "epoch": 8.68580871720719,
      "grad_norm": 11.221620559692383,
      "learning_rate": 7.29906902440368e-06,
      "loss": 1.2398,
      "step": 52560
    },
    {
      "epoch": 8.687461268332989,
      "grad_norm": 19.58458137512207,
      "learning_rate": 7.2898878055050595e-06,
      "loss": 1.3754,
      "step": 52570
    },
    {
      "epoch": 8.68911381945879,
      "grad_norm": 12.247591972351074,
      "learning_rate": 7.280706586606438e-06,
      "loss": 1.3815,
      "step": 52580
    },
    {
      "epoch": 8.69076637058459,
      "grad_norm": 9.798544883728027,
      "learning_rate": 7.271525367707817e-06,
      "loss": 1.3341,
      "step": 52590
    },
    {
      "epoch": 8.69241892171039,
      "grad_norm": 12.177785873413086,
      "learning_rate": 7.2623441488091955e-06,
      "loss": 1.2962,
      "step": 52600
    },
    {
      "epoch": 8.69407147283619,
      "grad_norm": 13.61369800567627,
      "learning_rate": 7.253162929910576e-06,
      "loss": 1.4009,
      "step": 52610
    },
    {
      "epoch": 8.695724023961992,
      "grad_norm": 11.679938316345215,
      "learning_rate": 7.243981711011955e-06,
      "loss": 1.2359,
      "step": 52620
    },
    {
      "epoch": 8.697376575087791,
      "grad_norm": 17.74927520751953,
      "learning_rate": 7.234800492113333e-06,
      "loss": 1.3443,
      "step": 52630
    },
    {
      "epoch": 8.699029126213592,
      "grad_norm": 20.92237091064453,
      "learning_rate": 7.2256192732147125e-06,
      "loss": 1.3608,
      "step": 52640
    },
    {
      "epoch": 8.700681677339393,
      "grad_norm": 14.271102905273438,
      "learning_rate": 7.216438054316091e-06,
      "loss": 1.2804,
      "step": 52650
    },
    {
      "epoch": 8.702334228465194,
      "grad_norm": 16.352445602416992,
      "learning_rate": 7.207256835417471e-06,
      "loss": 1.3786,
      "step": 52660
    },
    {
      "epoch": 8.703986779590993,
      "grad_norm": 11.873323440551758,
      "learning_rate": 7.198075616518849e-06,
      "loss": 1.3908,
      "step": 52670
    },
    {
      "epoch": 8.705639330716794,
      "grad_norm": 15.193153381347656,
      "learning_rate": 7.188894397620229e-06,
      "loss": 1.2402,
      "step": 52680
    },
    {
      "epoch": 8.707291881842595,
      "grad_norm": 22.531282424926758,
      "learning_rate": 7.179713178721607e-06,
      "loss": 1.2404,
      "step": 52690
    },
    {
      "epoch": 8.708944432968394,
      "grad_norm": 10.484509468078613,
      "learning_rate": 7.170531959822986e-06,
      "loss": 1.3803,
      "step": 52700
    },
    {
      "epoch": 8.710596984094195,
      "grad_norm": 21.465103149414062,
      "learning_rate": 7.161350740924365e-06,
      "loss": 1.458,
      "step": 52710
    },
    {
      "epoch": 8.712249535219996,
      "grad_norm": 16.147686004638672,
      "learning_rate": 7.152169522025745e-06,
      "loss": 1.3513,
      "step": 52720
    },
    {
      "epoch": 8.713902086345797,
      "grad_norm": 14.028088569641113,
      "learning_rate": 7.142988303127124e-06,
      "loss": 1.2286,
      "step": 52730
    },
    {
      "epoch": 8.715554637471596,
      "grad_norm": 15.7096586227417,
      "learning_rate": 7.1338070842285025e-06,
      "loss": 1.2122,
      "step": 52740
    },
    {
      "epoch": 8.717207188597397,
      "grad_norm": 21.39398956298828,
      "learning_rate": 7.124625865329882e-06,
      "loss": 1.274,
      "step": 52750
    },
    {
      "epoch": 8.718859739723198,
      "grad_norm": 15.905527114868164,
      "learning_rate": 7.11544464643126e-06,
      "loss": 1.323,
      "step": 52760
    },
    {
      "epoch": 8.720512290848998,
      "grad_norm": 13.975886344909668,
      "learning_rate": 7.106263427532639e-06,
      "loss": 1.5179,
      "step": 52770
    },
    {
      "epoch": 8.722164841974799,
      "grad_norm": 20.37264060974121,
      "learning_rate": 7.0970822086340194e-06,
      "loss": 1.2484,
      "step": 52780
    },
    {
      "epoch": 8.7238173931006,
      "grad_norm": 10.651167869567871,
      "learning_rate": 7.087900989735398e-06,
      "loss": 1.3944,
      "step": 52790
    },
    {
      "epoch": 8.725469944226399,
      "grad_norm": 12.802726745605469,
      "learning_rate": 7.078719770836777e-06,
      "loss": 1.4366,
      "step": 52800
    },
    {
      "epoch": 8.7271224953522,
      "grad_norm": 13.142681121826172,
      "learning_rate": 7.0695385519381555e-06,
      "loss": 1.3166,
      "step": 52810
    },
    {
      "epoch": 8.728775046478,
      "grad_norm": 14.891148567199707,
      "learning_rate": 7.060357333039534e-06,
      "loss": 1.3012,
      "step": 52820
    },
    {
      "epoch": 8.7304275976038,
      "grad_norm": 13.529729843139648,
      "learning_rate": 7.051176114140913e-06,
      "loss": 1.4384,
      "step": 52830
    },
    {
      "epoch": 8.732080148729601,
      "grad_norm": 14.417351722717285,
      "learning_rate": 7.041994895242293e-06,
      "loss": 1.4955,
      "step": 52840
    },
    {
      "epoch": 8.733732699855402,
      "grad_norm": 11.694726943969727,
      "learning_rate": 7.032813676343672e-06,
      "loss": 1.3628,
      "step": 52850
    },
    {
      "epoch": 8.735385250981203,
      "grad_norm": 8.318774223327637,
      "learning_rate": 7.023632457445051e-06,
      "loss": 1.2791,
      "step": 52860
    },
    {
      "epoch": 8.737037802107002,
      "grad_norm": 29.99515724182129,
      "learning_rate": 7.014451238546429e-06,
      "loss": 1.357,
      "step": 52870
    },
    {
      "epoch": 8.738690353232803,
      "grad_norm": 8.45457649230957,
      "learning_rate": 7.0052700196478085e-06,
      "loss": 1.2616,
      "step": 52880
    },
    {
      "epoch": 8.740342904358604,
      "grad_norm": 15.457688331604004,
      "learning_rate": 6.996088800749189e-06,
      "loss": 1.4261,
      "step": 52890
    },
    {
      "epoch": 8.741995455484403,
      "grad_norm": 23.088054656982422,
      "learning_rate": 6.986907581850567e-06,
      "loss": 1.3279,
      "step": 52900
    },
    {
      "epoch": 8.743648006610204,
      "grad_norm": 10.44580364227295,
      "learning_rate": 6.977726362951946e-06,
      "loss": 1.2399,
      "step": 52910
    },
    {
      "epoch": 8.745300557736005,
      "grad_norm": 15.399632453918457,
      "learning_rate": 6.968545144053325e-06,
      "loss": 1.3267,
      "step": 52920
    },
    {
      "epoch": 8.746953108861806,
      "grad_norm": 11.569892883300781,
      "learning_rate": 6.959363925154704e-06,
      "loss": 1.2217,
      "step": 52930
    },
    {
      "epoch": 8.748605659987605,
      "grad_norm": 13.182928085327148,
      "learning_rate": 6.950182706256082e-06,
      "loss": 1.3053,
      "step": 52940
    },
    {
      "epoch": 8.750258211113406,
      "grad_norm": 10.968775749206543,
      "learning_rate": 6.941001487357462e-06,
      "loss": 1.2109,
      "step": 52950
    },
    {
      "epoch": 8.751910762239207,
      "grad_norm": 28.66567611694336,
      "learning_rate": 6.931820268458842e-06,
      "loss": 1.2991,
      "step": 52960
    },
    {
      "epoch": 8.753563313365007,
      "grad_norm": 14.520941734313965,
      "learning_rate": 6.92263904956022e-06,
      "loss": 1.3108,
      "step": 52970
    },
    {
      "epoch": 8.755215864490808,
      "grad_norm": 18.797256469726562,
      "learning_rate": 6.9134578306615985e-06,
      "loss": 1.3105,
      "step": 52980
    },
    {
      "epoch": 8.756868415616609,
      "grad_norm": 21.74106788635254,
      "learning_rate": 6.904276611762978e-06,
      "loss": 1.4593,
      "step": 52990
    },
    {
      "epoch": 8.758520966742408,
      "grad_norm": 21.514694213867188,
      "learning_rate": 6.895095392864356e-06,
      "loss": 1.2157,
      "step": 53000
    },
    {
      "epoch": 8.760173517868209,
      "grad_norm": 15.12600040435791,
      "learning_rate": 6.885914173965736e-06,
      "loss": 1.4009,
      "step": 53010
    },
    {
      "epoch": 8.76182606899401,
      "grad_norm": 17.038434982299805,
      "learning_rate": 6.8767329550671155e-06,
      "loss": 1.3185,
      "step": 53020
    },
    {
      "epoch": 8.76347862011981,
      "grad_norm": 16.833236694335938,
      "learning_rate": 6.867551736168494e-06,
      "loss": 1.2934,
      "step": 53030
    },
    {
      "epoch": 8.76513117124561,
      "grad_norm": 13.726874351501465,
      "learning_rate": 6.858370517269873e-06,
      "loss": 1.4518,
      "step": 53040
    },
    {
      "epoch": 8.766783722371411,
      "grad_norm": 42.266380310058594,
      "learning_rate": 6.8491892983712515e-06,
      "loss": 1.3364,
      "step": 53050
    },
    {
      "epoch": 8.768436273497212,
      "grad_norm": 22.413843154907227,
      "learning_rate": 6.840008079472632e-06,
      "loss": 1.3972,
      "step": 53060
    },
    {
      "epoch": 8.770088824623011,
      "grad_norm": 19.03530502319336,
      "learning_rate": 6.830826860574011e-06,
      "loss": 1.322,
      "step": 53070
    },
    {
      "epoch": 8.771741375748812,
      "grad_norm": 22.070907592773438,
      "learning_rate": 6.821645641675389e-06,
      "loss": 1.314,
      "step": 53080
    },
    {
      "epoch": 8.773393926874613,
      "grad_norm": 11.455080032348633,
      "learning_rate": 6.8124644227767685e-06,
      "loss": 1.3339,
      "step": 53090
    },
    {
      "epoch": 8.775046478000412,
      "grad_norm": 9.250655174255371,
      "learning_rate": 6.803283203878147e-06,
      "loss": 1.3523,
      "step": 53100
    },
    {
      "epoch": 8.776699029126213,
      "grad_norm": 11.893872261047363,
      "learning_rate": 6.794101984979525e-06,
      "loss": 1.3746,
      "step": 53110
    },
    {
      "epoch": 8.778351580252014,
      "grad_norm": 16.446348190307617,
      "learning_rate": 6.784920766080906e-06,
      "loss": 1.3863,
      "step": 53120
    },
    {
      "epoch": 8.780004131377815,
      "grad_norm": 13.59485912322998,
      "learning_rate": 6.775739547182285e-06,
      "loss": 1.2766,
      "step": 53130
    },
    {
      "epoch": 8.781656682503614,
      "grad_norm": 11.817255020141602,
      "learning_rate": 6.766558328283663e-06,
      "loss": 1.3529,
      "step": 53140
    },
    {
      "epoch": 8.783309233629415,
      "grad_norm": 14.739175796508789,
      "learning_rate": 6.757377109385042e-06,
      "loss": 1.4092,
      "step": 53150
    },
    {
      "epoch": 8.784961784755216,
      "grad_norm": 16.03285789489746,
      "learning_rate": 6.748195890486421e-06,
      "loss": 1.4103,
      "step": 53160
    },
    {
      "epoch": 8.786614335881016,
      "grad_norm": 14.885622024536133,
      "learning_rate": 6.7390146715878e-06,
      "loss": 1.2313,
      "step": 53170
    },
    {
      "epoch": 8.788266887006817,
      "grad_norm": 17.540298461914062,
      "learning_rate": 6.72983345268918e-06,
      "loss": 1.2694,
      "step": 53180
    },
    {
      "epoch": 8.789919438132618,
      "grad_norm": 12.274869918823242,
      "learning_rate": 6.7206522337905584e-06,
      "loss": 1.3668,
      "step": 53190
    },
    {
      "epoch": 8.791571989258419,
      "grad_norm": 12.353447914123535,
      "learning_rate": 6.711471014891938e-06,
      "loss": 1.3208,
      "step": 53200
    },
    {
      "epoch": 8.793224540384218,
      "grad_norm": 10.738207817077637,
      "learning_rate": 6.702289795993316e-06,
      "loss": 1.1065,
      "step": 53210
    },
    {
      "epoch": 8.794877091510019,
      "grad_norm": 11.016992568969727,
      "learning_rate": 6.693108577094695e-06,
      "loss": 1.3443,
      "step": 53220
    },
    {
      "epoch": 8.79652964263582,
      "grad_norm": 15.448955535888672,
      "learning_rate": 6.683927358196074e-06,
      "loss": 1.3079,
      "step": 53230
    },
    {
      "epoch": 8.798182193761619,
      "grad_norm": 12.838058471679688,
      "learning_rate": 6.674746139297454e-06,
      "loss": 1.2739,
      "step": 53240
    },
    {
      "epoch": 8.79983474488742,
      "grad_norm": 6.943477630615234,
      "learning_rate": 6.665564920398833e-06,
      "loss": 1.2321,
      "step": 53250
    },
    {
      "epoch": 8.80148729601322,
      "grad_norm": 22.82369041442871,
      "learning_rate": 6.6563837015002115e-06,
      "loss": 1.4648,
      "step": 53260
    },
    {
      "epoch": 8.80313984713902,
      "grad_norm": 11.41329288482666,
      "learning_rate": 6.64720248260159e-06,
      "loss": 1.2926,
      "step": 53270
    },
    {
      "epoch": 8.804792398264821,
      "grad_norm": 16.04501724243164,
      "learning_rate": 6.638021263702969e-06,
      "loss": 1.3558,
      "step": 53280
    },
    {
      "epoch": 8.806444949390622,
      "grad_norm": 11.551048278808594,
      "learning_rate": 6.628840044804349e-06,
      "loss": 1.3032,
      "step": 53290
    },
    {
      "epoch": 8.808097500516423,
      "grad_norm": 16.122791290283203,
      "learning_rate": 6.619658825905728e-06,
      "loss": 1.4187,
      "step": 53300
    },
    {
      "epoch": 8.809750051642222,
      "grad_norm": 19.001909255981445,
      "learning_rate": 6.610477607007107e-06,
      "loss": 1.3634,
      "step": 53310
    },
    {
      "epoch": 8.811402602768023,
      "grad_norm": 16.42638397216797,
      "learning_rate": 6.601296388108485e-06,
      "loss": 1.2488,
      "step": 53320
    },
    {
      "epoch": 8.813055153893824,
      "grad_norm": 18.04165267944336,
      "learning_rate": 6.5921151692098645e-06,
      "loss": 1.3113,
      "step": 53330
    },
    {
      "epoch": 8.814707705019623,
      "grad_norm": 14.503774642944336,
      "learning_rate": 6.582933950311243e-06,
      "loss": 1.2873,
      "step": 53340
    },
    {
      "epoch": 8.816360256145424,
      "grad_norm": 12.262187957763672,
      "learning_rate": 6.573752731412623e-06,
      "loss": 1.3056,
      "step": 53350
    },
    {
      "epoch": 8.818012807271225,
      "grad_norm": 16.321773529052734,
      "learning_rate": 6.564571512514002e-06,
      "loss": 1.2377,
      "step": 53360
    },
    {
      "epoch": 8.819665358397025,
      "grad_norm": 8.0360107421875,
      "learning_rate": 6.555390293615381e-06,
      "loss": 1.3937,
      "step": 53370
    },
    {
      "epoch": 8.821317909522826,
      "grad_norm": 9.701162338256836,
      "learning_rate": 6.54620907471676e-06,
      "loss": 1.2723,
      "step": 53380
    },
    {
      "epoch": 8.822970460648627,
      "grad_norm": 63.11784362792969,
      "learning_rate": 6.537027855818138e-06,
      "loss": 1.4564,
      "step": 53390
    },
    {
      "epoch": 8.824623011774428,
      "grad_norm": 13.213398933410645,
      "learning_rate": 6.527846636919517e-06,
      "loss": 1.3917,
      "step": 53400
    },
    {
      "epoch": 8.826275562900227,
      "grad_norm": 29.3795166015625,
      "learning_rate": 6.518665418020898e-06,
      "loss": 1.1673,
      "step": 53410
    },
    {
      "epoch": 8.827928114026028,
      "grad_norm": 20.529861450195312,
      "learning_rate": 6.509484199122276e-06,
      "loss": 1.3965,
      "step": 53420
    },
    {
      "epoch": 8.829580665151829,
      "grad_norm": 12.665575981140137,
      "learning_rate": 6.5003029802236544e-06,
      "loss": 1.2904,
      "step": 53430
    },
    {
      "epoch": 8.831233216277628,
      "grad_norm": 23.68876838684082,
      "learning_rate": 6.491121761325034e-06,
      "loss": 1.4163,
      "step": 53440
    },
    {
      "epoch": 8.832885767403429,
      "grad_norm": 13.767816543579102,
      "learning_rate": 6.481940542426412e-06,
      "loss": 1.2772,
      "step": 53450
    },
    {
      "epoch": 8.83453831852923,
      "grad_norm": 38.02272415161133,
      "learning_rate": 6.472759323527792e-06,
      "loss": 1.3729,
      "step": 53460
    },
    {
      "epoch": 8.83619086965503,
      "grad_norm": 18.688528060913086,
      "learning_rate": 6.4635781046291714e-06,
      "loss": 1.4323,
      "step": 53470
    },
    {
      "epoch": 8.83784342078083,
      "grad_norm": 8.972434043884277,
      "learning_rate": 6.45439688573055e-06,
      "loss": 1.2365,
      "step": 53480
    },
    {
      "epoch": 8.839495971906631,
      "grad_norm": 12.999137878417969,
      "learning_rate": 6.445215666831929e-06,
      "loss": 1.2838,
      "step": 53490
    },
    {
      "epoch": 8.841148523032432,
      "grad_norm": 12.123946189880371,
      "learning_rate": 6.4360344479333075e-06,
      "loss": 1.3943,
      "step": 53500
    },
    {
      "epoch": 8.842801074158231,
      "grad_norm": 8.562018394470215,
      "learning_rate": 6.426853229034687e-06,
      "loss": 1.4237,
      "step": 53510
    },
    {
      "epoch": 8.844453625284032,
      "grad_norm": 19.84341049194336,
      "learning_rate": 6.417672010136067e-06,
      "loss": 1.3901,
      "step": 53520
    },
    {
      "epoch": 8.846106176409833,
      "grad_norm": 11.601472854614258,
      "learning_rate": 6.408490791237445e-06,
      "loss": 1.2849,
      "step": 53530
    },
    {
      "epoch": 8.847758727535632,
      "grad_norm": 16.48011589050293,
      "learning_rate": 6.3993095723388245e-06,
      "loss": 1.4206,
      "step": 53540
    },
    {
      "epoch": 8.849411278661433,
      "grad_norm": 10.797418594360352,
      "learning_rate": 6.390128353440203e-06,
      "loss": 1.516,
      "step": 53550
    },
    {
      "epoch": 8.851063829787234,
      "grad_norm": 13.68311595916748,
      "learning_rate": 6.380947134541581e-06,
      "loss": 1.3379,
      "step": 53560
    },
    {
      "epoch": 8.852716380913034,
      "grad_norm": 11.746673583984375,
      "learning_rate": 6.3717659156429605e-06,
      "loss": 1.3683,
      "step": 53570
    },
    {
      "epoch": 8.854368932038835,
      "grad_norm": 16.722267150878906,
      "learning_rate": 6.362584696744341e-06,
      "loss": 1.4114,
      "step": 53580
    },
    {
      "epoch": 8.856021483164636,
      "grad_norm": 13.57430362701416,
      "learning_rate": 6.353403477845719e-06,
      "loss": 1.3794,
      "step": 53590
    },
    {
      "epoch": 8.857674034290437,
      "grad_norm": 19.50122833251953,
      "learning_rate": 6.344222258947098e-06,
      "loss": 1.3773,
      "step": 53600
    },
    {
      "epoch": 8.859326585416236,
      "grad_norm": 13.851290702819824,
      "learning_rate": 6.335041040048477e-06,
      "loss": 1.2712,
      "step": 53610
    },
    {
      "epoch": 8.860979136542037,
      "grad_norm": 11.362812995910645,
      "learning_rate": 6.325859821149856e-06,
      "loss": 1.2669,
      "step": 53620
    },
    {
      "epoch": 8.862631687667838,
      "grad_norm": 13.138988494873047,
      "learning_rate": 6.316678602251236e-06,
      "loss": 1.3556,
      "step": 53630
    },
    {
      "epoch": 8.864284238793637,
      "grad_norm": 21.056747436523438,
      "learning_rate": 6.307497383352614e-06,
      "loss": 1.4155,
      "step": 53640
    },
    {
      "epoch": 8.865936789919438,
      "grad_norm": 16.05164909362793,
      "learning_rate": 6.298316164453994e-06,
      "loss": 1.4202,
      "step": 53650
    },
    {
      "epoch": 8.867589341045239,
      "grad_norm": 30.52538299560547,
      "learning_rate": 6.289134945555372e-06,
      "loss": 1.2314,
      "step": 53660
    },
    {
      "epoch": 8.86924189217104,
      "grad_norm": 27.325618743896484,
      "learning_rate": 6.279953726656751e-06,
      "loss": 1.3327,
      "step": 53670
    },
    {
      "epoch": 8.870894443296839,
      "grad_norm": 14.13979434967041,
      "learning_rate": 6.27077250775813e-06,
      "loss": 1.4423,
      "step": 53680
    },
    {
      "epoch": 8.87254699442264,
      "grad_norm": 9.841453552246094,
      "learning_rate": 6.26159128885951e-06,
      "loss": 1.4869,
      "step": 53690
    },
    {
      "epoch": 8.874199545548441,
      "grad_norm": 10.473819732666016,
      "learning_rate": 6.252410069960889e-06,
      "loss": 1.2138,
      "step": 53700
    },
    {
      "epoch": 8.87585209667424,
      "grad_norm": 20.32805824279785,
      "learning_rate": 6.2432288510622675e-06,
      "loss": 1.3251,
      "step": 53710
    },
    {
      "epoch": 8.877504647800041,
      "grad_norm": 26.516998291015625,
      "learning_rate": 6.234047632163646e-06,
      "loss": 1.3455,
      "step": 53720
    },
    {
      "epoch": 8.879157198925842,
      "grad_norm": 10.170158386230469,
      "learning_rate": 6.224866413265026e-06,
      "loss": 1.2815,
      "step": 53730
    },
    {
      "epoch": 8.880809750051641,
      "grad_norm": 13.213897705078125,
      "learning_rate": 6.215685194366404e-06,
      "loss": 1.3635,
      "step": 53740
    },
    {
      "epoch": 8.882462301177442,
      "grad_norm": 15.690320014953613,
      "learning_rate": 6.206503975467784e-06,
      "loss": 1.3606,
      "step": 53750
    },
    {
      "epoch": 8.884114852303243,
      "grad_norm": 8.988179206848145,
      "learning_rate": 6.197322756569163e-06,
      "loss": 1.3001,
      "step": 53760
    },
    {
      "epoch": 8.885767403429044,
      "grad_norm": 15.613768577575684,
      "learning_rate": 6.188141537670541e-06,
      "loss": 1.2319,
      "step": 53770
    },
    {
      "epoch": 8.887419954554844,
      "grad_norm": 22.44986343383789,
      "learning_rate": 6.1789603187719205e-06,
      "loss": 1.2765,
      "step": 53780
    },
    {
      "epoch": 8.889072505680645,
      "grad_norm": 19.037084579467773,
      "learning_rate": 6.1697790998733e-06,
      "loss": 1.3022,
      "step": 53790
    },
    {
      "epoch": 8.890725056806446,
      "grad_norm": 20.137554168701172,
      "learning_rate": 6.160597880974678e-06,
      "loss": 1.4232,
      "step": 53800
    },
    {
      "epoch": 8.892377607932245,
      "grad_norm": 10.645371437072754,
      "learning_rate": 6.151416662076057e-06,
      "loss": 1.2907,
      "step": 53810
    },
    {
      "epoch": 8.894030159058046,
      "grad_norm": 11.387445449829102,
      "learning_rate": 6.142235443177437e-06,
      "loss": 1.2386,
      "step": 53820
    },
    {
      "epoch": 8.895682710183847,
      "grad_norm": 13.088994979858398,
      "learning_rate": 6.133054224278816e-06,
      "loss": 1.379,
      "step": 53830
    },
    {
      "epoch": 8.897335261309646,
      "grad_norm": 17.280851364135742,
      "learning_rate": 6.123873005380194e-06,
      "loss": 1.2731,
      "step": 53840
    },
    {
      "epoch": 8.898987812435447,
      "grad_norm": 13.71379566192627,
      "learning_rate": 6.1146917864815735e-06,
      "loss": 1.3099,
      "step": 53850
    },
    {
      "epoch": 8.900640363561248,
      "grad_norm": 27.96197509765625,
      "learning_rate": 6.105510567582953e-06,
      "loss": 1.3606,
      "step": 53860
    },
    {
      "epoch": 8.902292914687049,
      "grad_norm": 14.15559196472168,
      "learning_rate": 6.096329348684331e-06,
      "loss": 1.3055,
      "step": 53870
    },
    {
      "epoch": 8.903945465812848,
      "grad_norm": 14.177240371704102,
      "learning_rate": 6.0871481297857104e-06,
      "loss": 1.3035,
      "step": 53880
    },
    {
      "epoch": 8.905598016938649,
      "grad_norm": 12.994119644165039,
      "learning_rate": 6.07796691088709e-06,
      "loss": 1.3279,
      "step": 53890
    },
    {
      "epoch": 8.90725056806445,
      "grad_norm": 15.457225799560547,
      "learning_rate": 6.068785691988468e-06,
      "loss": 1.3741,
      "step": 53900
    },
    {
      "epoch": 8.90890311919025,
      "grad_norm": 9.61684799194336,
      "learning_rate": 6.059604473089848e-06,
      "loss": 1.3281,
      "step": 53910
    },
    {
      "epoch": 8.91055567031605,
      "grad_norm": 10.564475059509277,
      "learning_rate": 6.0504232541912266e-06,
      "loss": 1.3453,
      "step": 53920
    },
    {
      "epoch": 8.912208221441851,
      "grad_norm": 12.618019104003906,
      "learning_rate": 6.041242035292606e-06,
      "loss": 1.2849,
      "step": 53930
    },
    {
      "epoch": 8.913860772567652,
      "grad_norm": 19.173622131347656,
      "learning_rate": 6.032060816393985e-06,
      "loss": 1.4845,
      "step": 53940
    },
    {
      "epoch": 8.915513323693451,
      "grad_norm": 11.999756813049316,
      "learning_rate": 6.0228795974953635e-06,
      "loss": 1.3252,
      "step": 53950
    },
    {
      "epoch": 8.917165874819252,
      "grad_norm": 25.721511840820312,
      "learning_rate": 6.013698378596743e-06,
      "loss": 1.3001,
      "step": 53960
    },
    {
      "epoch": 8.918818425945053,
      "grad_norm": 14.201374053955078,
      "learning_rate": 6.004517159698122e-06,
      "loss": 1.4462,
      "step": 53970
    },
    {
      "epoch": 8.920470977070853,
      "grad_norm": 14.782429695129395,
      "learning_rate": 5.9953359407995e-06,
      "loss": 1.5245,
      "step": 53980
    },
    {
      "epoch": 8.922123528196654,
      "grad_norm": 13.691256523132324,
      "learning_rate": 5.9861547219008805e-06,
      "loss": 1.3555,
      "step": 53990
    },
    {
      "epoch": 8.923776079322455,
      "grad_norm": 18.531856536865234,
      "learning_rate": 5.976973503002259e-06,
      "loss": 1.3047,
      "step": 54000
    },
    {
      "epoch": 8.925428630448254,
      "grad_norm": 11.382560729980469,
      "learning_rate": 5.967792284103637e-06,
      "loss": 1.2516,
      "step": 54010
    },
    {
      "epoch": 8.927081181574055,
      "grad_norm": 12.062386512756348,
      "learning_rate": 5.958611065205017e-06,
      "loss": 1.3224,
      "step": 54020
    },
    {
      "epoch": 8.928733732699856,
      "grad_norm": 26.916439056396484,
      "learning_rate": 5.949429846306396e-06,
      "loss": 1.3987,
      "step": 54030
    },
    {
      "epoch": 8.930386283825655,
      "grad_norm": 13.287517547607422,
      "learning_rate": 5.940248627407775e-06,
      "loss": 1.339,
      "step": 54040
    },
    {
      "epoch": 8.932038834951456,
      "grad_norm": 27.684982299804688,
      "learning_rate": 5.931067408509154e-06,
      "loss": 1.2144,
      "step": 54050
    },
    {
      "epoch": 8.933691386077257,
      "grad_norm": 12.099579811096191,
      "learning_rate": 5.921886189610533e-06,
      "loss": 1.2829,
      "step": 54060
    },
    {
      "epoch": 8.935343937203058,
      "grad_norm": 12.295833587646484,
      "learning_rate": 5.912704970711912e-06,
      "loss": 1.3906,
      "step": 54070
    },
    {
      "epoch": 8.936996488328857,
      "grad_norm": 15.061843872070312,
      "learning_rate": 5.903523751813291e-06,
      "loss": 1.4865,
      "step": 54080
    },
    {
      "epoch": 8.938649039454658,
      "grad_norm": 10.409856796264648,
      "learning_rate": 5.8943425329146695e-06,
      "loss": 1.4299,
      "step": 54090
    },
    {
      "epoch": 8.940301590580459,
      "grad_norm": 7.9486212730407715,
      "learning_rate": 5.885161314016049e-06,
      "loss": 1.2411,
      "step": 54100
    },
    {
      "epoch": 8.941954141706258,
      "grad_norm": 11.08385944366455,
      "learning_rate": 5.875980095117428e-06,
      "loss": 1.3165,
      "step": 54110
    },
    {
      "epoch": 8.94360669283206,
      "grad_norm": 15.459141731262207,
      "learning_rate": 5.866798876218807e-06,
      "loss": 1.431,
      "step": 54120
    },
    {
      "epoch": 8.94525924395786,
      "grad_norm": 12.070756912231445,
      "learning_rate": 5.8576176573201865e-06,
      "loss": 1.3148,
      "step": 54130
    },
    {
      "epoch": 8.946911795083661,
      "grad_norm": 15.25113582611084,
      "learning_rate": 5.848436438421565e-06,
      "loss": 1.4825,
      "step": 54140
    },
    {
      "epoch": 8.94856434620946,
      "grad_norm": 14.48801326751709,
      "learning_rate": 5.839255219522944e-06,
      "loss": 1.3299,
      "step": 54150
    },
    {
      "epoch": 8.950216897335261,
      "grad_norm": 40.4534797668457,
      "learning_rate": 5.8300740006243234e-06,
      "loss": 1.2875,
      "step": 54160
    },
    {
      "epoch": 8.951869448461062,
      "grad_norm": 10.24796199798584,
      "learning_rate": 5.820892781725702e-06,
      "loss": 1.296,
      "step": 54170
    },
    {
      "epoch": 8.953521999586862,
      "grad_norm": 12.344043731689453,
      "learning_rate": 5.811711562827081e-06,
      "loss": 1.4524,
      "step": 54180
    },
    {
      "epoch": 8.955174550712663,
      "grad_norm": 16.066959381103516,
      "learning_rate": 5.80253034392846e-06,
      "loss": 1.2011,
      "step": 54190
    },
    {
      "epoch": 8.956827101838464,
      "grad_norm": 16.067623138427734,
      "learning_rate": 5.7933491250298396e-06,
      "loss": 1.4874,
      "step": 54200
    },
    {
      "epoch": 8.958479652964265,
      "grad_norm": 15.692643165588379,
      "learning_rate": 5.784167906131218e-06,
      "loss": 1.2578,
      "step": 54210
    },
    {
      "epoch": 8.960132204090064,
      "grad_norm": 29.125408172607422,
      "learning_rate": 5.774986687232597e-06,
      "loss": 1.4732,
      "step": 54220
    },
    {
      "epoch": 8.961784755215865,
      "grad_norm": 14.437169075012207,
      "learning_rate": 5.7658054683339765e-06,
      "loss": 1.3453,
      "step": 54230
    },
    {
      "epoch": 8.963437306341666,
      "grad_norm": 14.426401138305664,
      "learning_rate": 5.756624249435355e-06,
      "loss": 1.3487,
      "step": 54240
    },
    {
      "epoch": 8.965089857467465,
      "grad_norm": 17.095333099365234,
      "learning_rate": 5.747443030536734e-06,
      "loss": 1.3297,
      "step": 54250
    },
    {
      "epoch": 8.966742408593266,
      "grad_norm": 15.15270709991455,
      "learning_rate": 5.738261811638113e-06,
      "loss": 1.2081,
      "step": 54260
    },
    {
      "epoch": 8.968394959719067,
      "grad_norm": 17.099987030029297,
      "learning_rate": 5.729080592739492e-06,
      "loss": 1.2349,
      "step": 54270
    },
    {
      "epoch": 8.970047510844866,
      "grad_norm": 19.881269454956055,
      "learning_rate": 5.719899373840872e-06,
      "loss": 1.3661,
      "step": 54280
    },
    {
      "epoch": 8.971700061970667,
      "grad_norm": 21.131669998168945,
      "learning_rate": 5.71071815494225e-06,
      "loss": 1.2879,
      "step": 54290
    },
    {
      "epoch": 8.973352613096468,
      "grad_norm": 19.318918228149414,
      "learning_rate": 5.7015369360436295e-06,
      "loss": 1.2724,
      "step": 54300
    },
    {
      "epoch": 8.975005164222267,
      "grad_norm": 22.20404624938965,
      "learning_rate": 5.692355717145009e-06,
      "loss": 1.3058,
      "step": 54310
    },
    {
      "epoch": 8.976657715348068,
      "grad_norm": 9.28625202178955,
      "learning_rate": 5.683174498246387e-06,
      "loss": 1.3284,
      "step": 54320
    },
    {
      "epoch": 8.97831026647387,
      "grad_norm": 18.054636001586914,
      "learning_rate": 5.673993279347766e-06,
      "loss": 1.2838,
      "step": 54330
    },
    {
      "epoch": 8.97996281759967,
      "grad_norm": 13.409037590026855,
      "learning_rate": 5.664812060449146e-06,
      "loss": 1.2346,
      "step": 54340
    },
    {
      "epoch": 8.98161536872547,
      "grad_norm": 15.619589805603027,
      "learning_rate": 5.655630841550524e-06,
      "loss": 1.3766,
      "step": 54350
    },
    {
      "epoch": 8.98326791985127,
      "grad_norm": 11.095884323120117,
      "learning_rate": 5.646449622651904e-06,
      "loss": 1.3703,
      "step": 54360
    },
    {
      "epoch": 8.984920470977071,
      "grad_norm": 14.479853630065918,
      "learning_rate": 5.6372684037532826e-06,
      "loss": 1.3841,
      "step": 54370
    },
    {
      "epoch": 8.98657302210287,
      "grad_norm": 12.613340377807617,
      "learning_rate": 5.628087184854662e-06,
      "loss": 1.2455,
      "step": 54380
    },
    {
      "epoch": 8.988225573228672,
      "grad_norm": 9.666672706604004,
      "learning_rate": 5.618905965956041e-06,
      "loss": 1.1269,
      "step": 54390
    },
    {
      "epoch": 8.989878124354473,
      "grad_norm": 32.712371826171875,
      "learning_rate": 5.6097247470574194e-06,
      "loss": 1.4111,
      "step": 54400
    },
    {
      "epoch": 8.991530675480274,
      "grad_norm": 13.545608520507812,
      "learning_rate": 5.600543528158799e-06,
      "loss": 1.3404,
      "step": 54410
    },
    {
      "epoch": 8.993183226606073,
      "grad_norm": 12.60755729675293,
      "learning_rate": 5.591362309260178e-06,
      "loss": 1.2749,
      "step": 54420
    },
    {
      "epoch": 8.994835777731874,
      "grad_norm": 17.72956657409668,
      "learning_rate": 5.582181090361556e-06,
      "loss": 1.3757,
      "step": 54430
    },
    {
      "epoch": 8.996488328857675,
      "grad_norm": 8.734169006347656,
      "learning_rate": 5.572999871462936e-06,
      "loss": 1.2355,
      "step": 54440
    },
    {
      "epoch": 8.998140879983474,
      "grad_norm": 10.11967945098877,
      "learning_rate": 5.563818652564315e-06,
      "loss": 1.1991,
      "step": 54450
    },
    {
      "epoch": 8.999793431109275,
      "grad_norm": 13.364051818847656,
      "learning_rate": 5.554637433665694e-06,
      "loss": 1.3775,
      "step": 54460
    },
    {
      "epoch": 8.999958686221856,
      "eval_accuracy": 0.33997730174492835,
      "eval_loss": 2.3859617710113525,
      "eval_runtime": 823.2799,
      "eval_samples_per_second": 34.248,
      "eval_steps_per_second": 8.562,
      "step": 54461
    },
    {
      "epoch": 9.001445982235076,
      "grad_norm": 18.1331844329834,
      "learning_rate": 5.5454562147670725e-06,
      "loss": 1.433,
      "step": 54470
    },
    {
      "epoch": 9.003098533360875,
      "grad_norm": 19.45754051208496,
      "learning_rate": 5.536274995868452e-06,
      "loss": 1.199,
      "step": 54480
    },
    {
      "epoch": 9.004751084486676,
      "grad_norm": 6.5627970695495605,
      "learning_rate": 5.527093776969831e-06,
      "loss": 1.2614,
      "step": 54490
    },
    {
      "epoch": 9.006403635612477,
      "grad_norm": 13.369138717651367,
      "learning_rate": 5.517912558071209e-06,
      "loss": 1.3375,
      "step": 54500
    },
    {
      "epoch": 9.008056186738278,
      "grad_norm": 21.747480392456055,
      "learning_rate": 5.508731339172589e-06,
      "loss": 1.2666,
      "step": 54510
    },
    {
      "epoch": 9.009708737864077,
      "grad_norm": 13.323081970214844,
      "learning_rate": 5.499550120273968e-06,
      "loss": 1.3027,
      "step": 54520
    },
    {
      "epoch": 9.011361288989878,
      "grad_norm": 8.494196891784668,
      "learning_rate": 5.490368901375347e-06,
      "loss": 1.1626,
      "step": 54530
    },
    {
      "epoch": 9.01301384011568,
      "grad_norm": 8.807066917419434,
      "learning_rate": 5.481187682476726e-06,
      "loss": 1.142,
      "step": 54540
    },
    {
      "epoch": 9.014666391241478,
      "grad_norm": 21.88263702392578,
      "learning_rate": 5.472006463578105e-06,
      "loss": 1.3858,
      "step": 54550
    },
    {
      "epoch": 9.01631894236728,
      "grad_norm": 16.97997283935547,
      "learning_rate": 5.462825244679484e-06,
      "loss": 1.4277,
      "step": 54560
    },
    {
      "epoch": 9.01797149349308,
      "grad_norm": 17.69019889831543,
      "learning_rate": 5.453644025780863e-06,
      "loss": 1.4564,
      "step": 54570
    },
    {
      "epoch": 9.01962404461888,
      "grad_norm": 21.331026077270508,
      "learning_rate": 5.444462806882242e-06,
      "loss": 1.2666,
      "step": 54580
    },
    {
      "epoch": 9.02127659574468,
      "grad_norm": 21.188798904418945,
      "learning_rate": 5.435281587983621e-06,
      "loss": 1.3853,
      "step": 54590
    },
    {
      "epoch": 9.022929146870482,
      "grad_norm": 17.578445434570312,
      "learning_rate": 5.426100369085e-06,
      "loss": 1.2737,
      "step": 54600
    },
    {
      "epoch": 9.024581697996283,
      "grad_norm": 21.36452293395996,
      "learning_rate": 5.4169191501863786e-06,
      "loss": 1.319,
      "step": 54610
    },
    {
      "epoch": 9.026234249122082,
      "grad_norm": 23.47857666015625,
      "learning_rate": 5.407737931287759e-06,
      "loss": 1.3926,
      "step": 54620
    },
    {
      "epoch": 9.027886800247883,
      "grad_norm": 11.379927635192871,
      "learning_rate": 5.398556712389137e-06,
      "loss": 1.2875,
      "step": 54630
    },
    {
      "epoch": 9.029539351373684,
      "grad_norm": 12.81651496887207,
      "learning_rate": 5.3893754934905155e-06,
      "loss": 1.1859,
      "step": 54640
    },
    {
      "epoch": 9.031191902499483,
      "grad_norm": 13.319893836975098,
      "learning_rate": 5.3801942745918956e-06,
      "loss": 1.3343,
      "step": 54650
    },
    {
      "epoch": 9.032844453625284,
      "grad_norm": 10.670450210571289,
      "learning_rate": 5.371013055693274e-06,
      "loss": 1.3357,
      "step": 54660
    },
    {
      "epoch": 9.034497004751085,
      "grad_norm": 12.764323234558105,
      "learning_rate": 5.361831836794653e-06,
      "loss": 1.3346,
      "step": 54670
    },
    {
      "epoch": 9.036149555876884,
      "grad_norm": 11.50831127166748,
      "learning_rate": 5.3526506178960325e-06,
      "loss": 1.2738,
      "step": 54680
    },
    {
      "epoch": 9.037802107002685,
      "grad_norm": 14.689154624938965,
      "learning_rate": 5.343469398997411e-06,
      "loss": 1.3225,
      "step": 54690
    },
    {
      "epoch": 9.039454658128486,
      "grad_norm": 12.108161926269531,
      "learning_rate": 5.33428818009879e-06,
      "loss": 1.2694,
      "step": 54700
    },
    {
      "epoch": 9.041107209254287,
      "grad_norm": 12.230198860168457,
      "learning_rate": 5.325106961200169e-06,
      "loss": 1.336,
      "step": 54710
    },
    {
      "epoch": 9.042759760380086,
      "grad_norm": 16.08524513244629,
      "learning_rate": 5.315925742301548e-06,
      "loss": 1.2942,
      "step": 54720
    },
    {
      "epoch": 9.044412311505887,
      "grad_norm": 36.26041030883789,
      "learning_rate": 5.306744523402928e-06,
      "loss": 1.2881,
      "step": 54730
    },
    {
      "epoch": 9.046064862631688,
      "grad_norm": 10.63325309753418,
      "learning_rate": 5.297563304504306e-06,
      "loss": 1.533,
      "step": 54740
    },
    {
      "epoch": 9.047717413757487,
      "grad_norm": 20.543127059936523,
      "learning_rate": 5.2883820856056855e-06,
      "loss": 1.429,
      "step": 54750
    },
    {
      "epoch": 9.049369964883288,
      "grad_norm": 16.287906646728516,
      "learning_rate": 5.279200866707065e-06,
      "loss": 1.3495,
      "step": 54760
    },
    {
      "epoch": 9.05102251600909,
      "grad_norm": 18.479795455932617,
      "learning_rate": 5.270019647808443e-06,
      "loss": 1.3766,
      "step": 54770
    },
    {
      "epoch": 9.05267506713489,
      "grad_norm": 12.131267547607422,
      "learning_rate": 5.260838428909822e-06,
      "loss": 1.388,
      "step": 54780
    },
    {
      "epoch": 9.05432761826069,
      "grad_norm": 17.166824340820312,
      "learning_rate": 5.251657210011202e-06,
      "loss": 1.3733,
      "step": 54790
    },
    {
      "epoch": 9.05598016938649,
      "grad_norm": 14.11315631866455,
      "learning_rate": 5.24247599111258e-06,
      "loss": 1.3204,
      "step": 54800
    },
    {
      "epoch": 9.057632720512292,
      "grad_norm": 11.092327117919922,
      "learning_rate": 5.233294772213959e-06,
      "loss": 1.2731,
      "step": 54810
    },
    {
      "epoch": 9.05928527163809,
      "grad_norm": 11.508016586303711,
      "learning_rate": 5.2241135533153385e-06,
      "loss": 1.3763,
      "step": 54820
    },
    {
      "epoch": 9.060937822763892,
      "grad_norm": 13.264908790588379,
      "learning_rate": 5.214932334416718e-06,
      "loss": 1.1948,
      "step": 54830
    },
    {
      "epoch": 9.062590373889693,
      "grad_norm": 16.212055206298828,
      "learning_rate": 5.205751115518096e-06,
      "loss": 1.318,
      "step": 54840
    },
    {
      "epoch": 9.064242925015492,
      "grad_norm": 33.09046173095703,
      "learning_rate": 5.1965698966194754e-06,
      "loss": 1.28,
      "step": 54850
    },
    {
      "epoch": 9.065895476141293,
      "grad_norm": 12.761731147766113,
      "learning_rate": 5.187388677720855e-06,
      "loss": 1.2515,
      "step": 54860
    },
    {
      "epoch": 9.067548027267094,
      "grad_norm": 11.305936813354492,
      "learning_rate": 5.178207458822233e-06,
      "loss": 1.1792,
      "step": 54870
    },
    {
      "epoch": 9.069200578392895,
      "grad_norm": 19.43776512145996,
      "learning_rate": 5.169026239923612e-06,
      "loss": 1.3766,
      "step": 54880
    },
    {
      "epoch": 9.070853129518694,
      "grad_norm": 11.29062557220459,
      "learning_rate": 5.1598450210249916e-06,
      "loss": 1.2883,
      "step": 54890
    },
    {
      "epoch": 9.072505680644495,
      "grad_norm": 12.61872673034668,
      "learning_rate": 5.15066380212637e-06,
      "loss": 1.245,
      "step": 54900
    },
    {
      "epoch": 9.074158231770296,
      "grad_norm": 10.980611801147461,
      "learning_rate": 5.14148258322775e-06,
      "loss": 1.42,
      "step": 54910
    },
    {
      "epoch": 9.075810782896095,
      "grad_norm": 13.058303833007812,
      "learning_rate": 5.1323013643291285e-06,
      "loss": 1.21,
      "step": 54920
    },
    {
      "epoch": 9.077463334021896,
      "grad_norm": 21.789306640625,
      "learning_rate": 5.123120145430508e-06,
      "loss": 1.3207,
      "step": 54930
    },
    {
      "epoch": 9.079115885147697,
      "grad_norm": 10.95693588256836,
      "learning_rate": 5.113938926531887e-06,
      "loss": 1.3505,
      "step": 54940
    },
    {
      "epoch": 9.080768436273496,
      "grad_norm": 21.274913787841797,
      "learning_rate": 5.104757707633265e-06,
      "loss": 1.2573,
      "step": 54950
    },
    {
      "epoch": 9.082420987399297,
      "grad_norm": 17.98501205444336,
      "learning_rate": 5.095576488734645e-06,
      "loss": 1.2962,
      "step": 54960
    },
    {
      "epoch": 9.084073538525098,
      "grad_norm": 21.35386848449707,
      "learning_rate": 5.086395269836024e-06,
      "loss": 1.3419,
      "step": 54970
    },
    {
      "epoch": 9.0857260896509,
      "grad_norm": 19.58464241027832,
      "learning_rate": 5.077214050937402e-06,
      "loss": 1.4037,
      "step": 54980
    },
    {
      "epoch": 9.087378640776699,
      "grad_norm": 19.767635345458984,
      "learning_rate": 5.068032832038782e-06,
      "loss": 1.2972,
      "step": 54990
    },
    {
      "epoch": 9.0890311919025,
      "grad_norm": 10.111676216125488,
      "learning_rate": 5.058851613140161e-06,
      "loss": 1.1995,
      "step": 55000
    },
    {
      "epoch": 9.0906837430283,
      "grad_norm": 11.100005149841309,
      "learning_rate": 5.049670394241539e-06,
      "loss": 1.2342,
      "step": 55010
    },
    {
      "epoch": 9.0923362941541,
      "grad_norm": 13.454795837402344,
      "learning_rate": 5.040489175342919e-06,
      "loss": 1.2795,
      "step": 55020
    },
    {
      "epoch": 9.0939888452799,
      "grad_norm": 16.521162033081055,
      "learning_rate": 5.031307956444298e-06,
      "loss": 1.3299,
      "step": 55030
    },
    {
      "epoch": 9.095641396405702,
      "grad_norm": 12.513435363769531,
      "learning_rate": 5.022126737545677e-06,
      "loss": 1.2241,
      "step": 55040
    },
    {
      "epoch": 9.097293947531501,
      "grad_norm": 33.393882751464844,
      "learning_rate": 5.012945518647056e-06,
      "loss": 1.3831,
      "step": 55050
    },
    {
      "epoch": 9.098946498657302,
      "grad_norm": 12.040635108947754,
      "learning_rate": 5.0037642997484345e-06,
      "loss": 1.2954,
      "step": 55060
    },
    {
      "epoch": 9.100599049783103,
      "grad_norm": 16.64748764038086,
      "learning_rate": 4.994583080849814e-06,
      "loss": 1.2867,
      "step": 55070
    },
    {
      "epoch": 9.102251600908904,
      "grad_norm": 20.921030044555664,
      "learning_rate": 4.985401861951193e-06,
      "loss": 1.3136,
      "step": 55080
    },
    {
      "epoch": 9.103904152034703,
      "grad_norm": 17.621395111083984,
      "learning_rate": 4.9762206430525714e-06,
      "loss": 1.2214,
      "step": 55090
    },
    {
      "epoch": 9.105556703160504,
      "grad_norm": 13.227643013000488,
      "learning_rate": 4.967039424153951e-06,
      "loss": 1.3465,
      "step": 55100
    },
    {
      "epoch": 9.107209254286305,
      "grad_norm": 12.755581855773926,
      "learning_rate": 4.95785820525533e-06,
      "loss": 1.3946,
      "step": 55110
    },
    {
      "epoch": 9.108861805412104,
      "grad_norm": 19.701791763305664,
      "learning_rate": 4.948676986356709e-06,
      "loss": 1.2807,
      "step": 55120
    },
    {
      "epoch": 9.110514356537905,
      "grad_norm": 16.122726440429688,
      "learning_rate": 4.9394957674580884e-06,
      "loss": 1.3864,
      "step": 55130
    },
    {
      "epoch": 9.112166907663706,
      "grad_norm": 14.194814682006836,
      "learning_rate": 4.930314548559467e-06,
      "loss": 1.3725,
      "step": 55140
    },
    {
      "epoch": 9.113819458789507,
      "grad_norm": 16.94427490234375,
      "learning_rate": 4.921133329660846e-06,
      "loss": 1.311,
      "step": 55150
    },
    {
      "epoch": 9.115472009915306,
      "grad_norm": 41.79347229003906,
      "learning_rate": 4.911952110762225e-06,
      "loss": 1.498,
      "step": 55160
    },
    {
      "epoch": 9.117124561041107,
      "grad_norm": 26.90255355834961,
      "learning_rate": 4.902770891863604e-06,
      "loss": 1.2616,
      "step": 55170
    },
    {
      "epoch": 9.118777112166908,
      "grad_norm": 15.550649642944336,
      "learning_rate": 4.893589672964983e-06,
      "loss": 1.2621,
      "step": 55180
    },
    {
      "epoch": 9.120429663292708,
      "grad_norm": 25.298446655273438,
      "learning_rate": 4.884408454066362e-06,
      "loss": 1.2517,
      "step": 55190
    },
    {
      "epoch": 9.122082214418509,
      "grad_norm": 19.14631462097168,
      "learning_rate": 4.8752272351677415e-06,
      "loss": 1.353,
      "step": 55200
    },
    {
      "epoch": 9.12373476554431,
      "grad_norm": 16.164554595947266,
      "learning_rate": 4.86604601626912e-06,
      "loss": 1.2734,
      "step": 55210
    },
    {
      "epoch": 9.125387316670109,
      "grad_norm": 16.599605560302734,
      "learning_rate": 4.856864797370499e-06,
      "loss": 1.2069,
      "step": 55220
    },
    {
      "epoch": 9.12703986779591,
      "grad_norm": 22.704082489013672,
      "learning_rate": 4.847683578471878e-06,
      "loss": 1.3562,
      "step": 55230
    },
    {
      "epoch": 9.12869241892171,
      "grad_norm": 9.706205368041992,
      "learning_rate": 4.838502359573257e-06,
      "loss": 1.239,
      "step": 55240
    },
    {
      "epoch": 9.130344970047512,
      "grad_norm": 19.193674087524414,
      "learning_rate": 4.829321140674636e-06,
      "loss": 1.3882,
      "step": 55250
    },
    {
      "epoch": 9.13199752117331,
      "grad_norm": 9.436002731323242,
      "learning_rate": 4.820139921776015e-06,
      "loss": 1.3104,
      "step": 55260
    },
    {
      "epoch": 9.133650072299112,
      "grad_norm": 12.764290809631348,
      "learning_rate": 4.810958702877394e-06,
      "loss": 1.238,
      "step": 55270
    },
    {
      "epoch": 9.135302623424913,
      "grad_norm": 13.602463722229004,
      "learning_rate": 4.801777483978774e-06,
      "loss": 1.4013,
      "step": 55280
    },
    {
      "epoch": 9.136955174550712,
      "grad_norm": 18.19413185119629,
      "learning_rate": 4.792596265080152e-06,
      "loss": 1.2348,
      "step": 55290
    },
    {
      "epoch": 9.138607725676513,
      "grad_norm": 11.962725639343262,
      "learning_rate": 4.783415046181531e-06,
      "loss": 1.1874,
      "step": 55300
    },
    {
      "epoch": 9.140260276802314,
      "grad_norm": 14.87602710723877,
      "learning_rate": 4.774233827282911e-06,
      "loss": 1.3184,
      "step": 55310
    },
    {
      "epoch": 9.141912827928113,
      "grad_norm": 17.802339553833008,
      "learning_rate": 4.765052608384289e-06,
      "loss": 1.2428,
      "step": 55320
    },
    {
      "epoch": 9.143565379053914,
      "grad_norm": 23.761425018310547,
      "learning_rate": 4.755871389485668e-06,
      "loss": 1.2894,
      "step": 55330
    },
    {
      "epoch": 9.145217930179715,
      "grad_norm": 10.590774536132812,
      "learning_rate": 4.7466901705870475e-06,
      "loss": 1.2307,
      "step": 55340
    },
    {
      "epoch": 9.146870481305516,
      "grad_norm": 11.037496566772461,
      "learning_rate": 4.737508951688426e-06,
      "loss": 1.2873,
      "step": 55350
    },
    {
      "epoch": 9.148523032431315,
      "grad_norm": 14.964154243469238,
      "learning_rate": 4.728327732789806e-06,
      "loss": 1.295,
      "step": 55360
    },
    {
      "epoch": 9.150175583557116,
      "grad_norm": 15.096177101135254,
      "learning_rate": 4.7191465138911844e-06,
      "loss": 1.2046,
      "step": 55370
    },
    {
      "epoch": 9.151828134682917,
      "grad_norm": 8.96634578704834,
      "learning_rate": 4.709965294992563e-06,
      "loss": 1.382,
      "step": 55380
    },
    {
      "epoch": 9.153480685808717,
      "grad_norm": 16.687162399291992,
      "learning_rate": 4.700784076093943e-06,
      "loss": 1.317,
      "step": 55390
    },
    {
      "epoch": 9.155133236934518,
      "grad_norm": 15.783407211303711,
      "learning_rate": 4.691602857195321e-06,
      "loss": 1.3151,
      "step": 55400
    },
    {
      "epoch": 9.156785788060319,
      "grad_norm": 18.55613136291504,
      "learning_rate": 4.682421638296701e-06,
      "loss": 1.402,
      "step": 55410
    },
    {
      "epoch": 9.158438339186118,
      "grad_norm": 11.569503784179688,
      "learning_rate": 4.67324041939808e-06,
      "loss": 1.3886,
      "step": 55420
    },
    {
      "epoch": 9.160090890311919,
      "grad_norm": 49.667762756347656,
      "learning_rate": 4.664059200499458e-06,
      "loss": 1.0859,
      "step": 55430
    },
    {
      "epoch": 9.16174344143772,
      "grad_norm": 11.705397605895996,
      "learning_rate": 4.6548779816008375e-06,
      "loss": 1.2273,
      "step": 55440
    },
    {
      "epoch": 9.16339599256352,
      "grad_norm": 15.881692886352539,
      "learning_rate": 4.645696762702217e-06,
      "loss": 1.388,
      "step": 55450
    },
    {
      "epoch": 9.16504854368932,
      "grad_norm": 18.31934928894043,
      "learning_rate": 4.636515543803595e-06,
      "loss": 1.3677,
      "step": 55460
    },
    {
      "epoch": 9.16670109481512,
      "grad_norm": 14.108166694641113,
      "learning_rate": 4.627334324904974e-06,
      "loss": 1.2068,
      "step": 55470
    },
    {
      "epoch": 9.168353645940922,
      "grad_norm": 10.275944709777832,
      "learning_rate": 4.618153106006354e-06,
      "loss": 1.3015,
      "step": 55480
    },
    {
      "epoch": 9.170006197066721,
      "grad_norm": 11.239326477050781,
      "learning_rate": 4.608971887107733e-06,
      "loss": 1.3207,
      "step": 55490
    },
    {
      "epoch": 9.171658748192522,
      "grad_norm": 16.125225067138672,
      "learning_rate": 4.599790668209112e-06,
      "loss": 1.2972,
      "step": 55500
    },
    {
      "epoch": 9.173311299318323,
      "grad_norm": 15.761225700378418,
      "learning_rate": 4.5906094493104905e-06,
      "loss": 1.2291,
      "step": 55510
    },
    {
      "epoch": 9.174963850444122,
      "grad_norm": 12.649540901184082,
      "learning_rate": 4.58142823041187e-06,
      "loss": 1.2749,
      "step": 55520
    },
    {
      "epoch": 9.176616401569923,
      "grad_norm": 17.501537322998047,
      "learning_rate": 4.572247011513249e-06,
      "loss": 1.3203,
      "step": 55530
    },
    {
      "epoch": 9.178268952695724,
      "grad_norm": 18.077924728393555,
      "learning_rate": 4.563065792614627e-06,
      "loss": 1.397,
      "step": 55540
    },
    {
      "epoch": 9.179921503821525,
      "grad_norm": 16.778873443603516,
      "learning_rate": 4.553884573716007e-06,
      "loss": 1.2976,
      "step": 55550
    },
    {
      "epoch": 9.181574054947324,
      "grad_norm": 13.572968482971191,
      "learning_rate": 4.544703354817386e-06,
      "loss": 1.2331,
      "step": 55560
    },
    {
      "epoch": 9.183226606073125,
      "grad_norm": 13.623981475830078,
      "learning_rate": 4.535522135918765e-06,
      "loss": 1.3081,
      "step": 55570
    },
    {
      "epoch": 9.184879157198926,
      "grad_norm": 11.336007118225098,
      "learning_rate": 4.5263409170201436e-06,
      "loss": 1.2609,
      "step": 55580
    },
    {
      "epoch": 9.186531708324726,
      "grad_norm": 14.808525085449219,
      "learning_rate": 4.517159698121523e-06,
      "loss": 1.2853,
      "step": 55590
    },
    {
      "epoch": 9.188184259450527,
      "grad_norm": 11.729154586791992,
      "learning_rate": 4.507978479222902e-06,
      "loss": 1.3402,
      "step": 55600
    },
    {
      "epoch": 9.189836810576328,
      "grad_norm": 25.362701416015625,
      "learning_rate": 4.4987972603242805e-06,
      "loss": 1.5593,
      "step": 55610
    },
    {
      "epoch": 9.191489361702128,
      "grad_norm": 10.813699722290039,
      "learning_rate": 4.48961604142566e-06,
      "loss": 1.233,
      "step": 55620
    },
    {
      "epoch": 9.193141912827928,
      "grad_norm": 11.491537094116211,
      "learning_rate": 4.480434822527039e-06,
      "loss": 1.3878,
      "step": 55630
    },
    {
      "epoch": 9.194794463953729,
      "grad_norm": 17.993120193481445,
      "learning_rate": 4.471253603628417e-06,
      "loss": 1.2426,
      "step": 55640
    },
    {
      "epoch": 9.19644701507953,
      "grad_norm": 14.781805038452148,
      "learning_rate": 4.4620723847297975e-06,
      "loss": 1.2122,
      "step": 55650
    },
    {
      "epoch": 9.198099566205329,
      "grad_norm": 12.504164695739746,
      "learning_rate": 4.452891165831176e-06,
      "loss": 1.3508,
      "step": 55660
    },
    {
      "epoch": 9.19975211733113,
      "grad_norm": 14.548686027526855,
      "learning_rate": 4.443709946932554e-06,
      "loss": 1.3797,
      "step": 55670
    },
    {
      "epoch": 9.20140466845693,
      "grad_norm": 14.047411918640137,
      "learning_rate": 4.434528728033934e-06,
      "loss": 1.1652,
      "step": 55680
    },
    {
      "epoch": 9.20305721958273,
      "grad_norm": 27.93585205078125,
      "learning_rate": 4.425347509135313e-06,
      "loss": 1.2276,
      "step": 55690
    },
    {
      "epoch": 9.204709770708531,
      "grad_norm": 14.484628677368164,
      "learning_rate": 4.416166290236692e-06,
      "loss": 1.2675,
      "step": 55700
    },
    {
      "epoch": 9.206362321834332,
      "grad_norm": 23.827726364135742,
      "learning_rate": 4.406985071338071e-06,
      "loss": 1.306,
      "step": 55710
    },
    {
      "epoch": 9.208014872960133,
      "grad_norm": 22.634979248046875,
      "learning_rate": 4.39780385243945e-06,
      "loss": 1.4343,
      "step": 55720
    },
    {
      "epoch": 9.209667424085932,
      "grad_norm": 17.955224990844727,
      "learning_rate": 4.38862263354083e-06,
      "loss": 1.3497,
      "step": 55730
    },
    {
      "epoch": 9.211319975211733,
      "grad_norm": 12.252691268920898,
      "learning_rate": 4.379441414642208e-06,
      "loss": 1.2658,
      "step": 55740
    },
    {
      "epoch": 9.212972526337534,
      "grad_norm": 29.16194725036621,
      "learning_rate": 4.3702601957435865e-06,
      "loss": 1.1545,
      "step": 55750
    },
    {
      "epoch": 9.214625077463333,
      "grad_norm": 19.66663932800293,
      "learning_rate": 4.361078976844967e-06,
      "loss": 1.3138,
      "step": 55760
    },
    {
      "epoch": 9.216277628589134,
      "grad_norm": 11.54080867767334,
      "learning_rate": 4.351897757946345e-06,
      "loss": 1.2657,
      "step": 55770
    },
    {
      "epoch": 9.217930179714935,
      "grad_norm": 17.858362197875977,
      "learning_rate": 4.342716539047724e-06,
      "loss": 1.3622,
      "step": 55780
    },
    {
      "epoch": 9.219582730840735,
      "grad_norm": 10.797994613647461,
      "learning_rate": 4.3335353201491035e-06,
      "loss": 1.4025,
      "step": 55790
    },
    {
      "epoch": 9.221235281966536,
      "grad_norm": 12.868209838867188,
      "learning_rate": 4.324354101250482e-06,
      "loss": 1.3193,
      "step": 55800
    },
    {
      "epoch": 9.222887833092337,
      "grad_norm": 12.669856071472168,
      "learning_rate": 4.315172882351861e-06,
      "loss": 1.3091,
      "step": 55810
    },
    {
      "epoch": 9.224540384218137,
      "grad_norm": 14.927458763122559,
      "learning_rate": 4.3059916634532404e-06,
      "loss": 1.3628,
      "step": 55820
    },
    {
      "epoch": 9.226192935343937,
      "grad_norm": 13.37768840789795,
      "learning_rate": 4.296810444554619e-06,
      "loss": 1.3404,
      "step": 55830
    },
    {
      "epoch": 9.227845486469738,
      "grad_norm": 21.847307205200195,
      "learning_rate": 4.287629225655998e-06,
      "loss": 1.306,
      "step": 55840
    },
    {
      "epoch": 9.229498037595539,
      "grad_norm": 40.5103759765625,
      "learning_rate": 4.278448006757377e-06,
      "loss": 1.3563,
      "step": 55850
    },
    {
      "epoch": 9.231150588721338,
      "grad_norm": 9.23901653289795,
      "learning_rate": 4.2692667878587566e-06,
      "loss": 1.1886,
      "step": 55860
    },
    {
      "epoch": 9.232803139847139,
      "grad_norm": 21.267831802368164,
      "learning_rate": 4.260085568960135e-06,
      "loss": 1.3202,
      "step": 55870
    },
    {
      "epoch": 9.23445569097294,
      "grad_norm": 32.90507888793945,
      "learning_rate": 4.250904350061514e-06,
      "loss": 1.3244,
      "step": 55880
    },
    {
      "epoch": 9.23610824209874,
      "grad_norm": 44.84469223022461,
      "learning_rate": 4.2417231311628935e-06,
      "loss": 1.3374,
      "step": 55890
    },
    {
      "epoch": 9.23776079322454,
      "grad_norm": 21.209402084350586,
      "learning_rate": 4.232541912264273e-06,
      "loss": 1.3268,
      "step": 55900
    },
    {
      "epoch": 9.239413344350341,
      "grad_norm": 10.94561767578125,
      "learning_rate": 4.223360693365651e-06,
      "loss": 1.3878,
      "step": 55910
    },
    {
      "epoch": 9.241065895476142,
      "grad_norm": 24.05000114440918,
      "learning_rate": 4.21417947446703e-06,
      "loss": 1.2205,
      "step": 55920
    },
    {
      "epoch": 9.242718446601941,
      "grad_norm": 17.849794387817383,
      "learning_rate": 4.20499825556841e-06,
      "loss": 1.2642,
      "step": 55930
    },
    {
      "epoch": 9.244370997727742,
      "grad_norm": 22.153186798095703,
      "learning_rate": 4.195817036669789e-06,
      "loss": 1.2733,
      "step": 55940
    },
    {
      "epoch": 9.246023548853543,
      "grad_norm": 15.071115493774414,
      "learning_rate": 4.186635817771167e-06,
      "loss": 1.2561,
      "step": 55950
    },
    {
      "epoch": 9.247676099979342,
      "grad_norm": 22.48716926574707,
      "learning_rate": 4.1774545988725465e-06,
      "loss": 1.3194,
      "step": 55960
    },
    {
      "epoch": 9.249328651105143,
      "grad_norm": 18.10262680053711,
      "learning_rate": 4.168273379973926e-06,
      "loss": 1.3629,
      "step": 55970
    },
    {
      "epoch": 9.250981202230944,
      "grad_norm": 14.13875961303711,
      "learning_rate": 4.159092161075304e-06,
      "loss": 1.286,
      "step": 55980
    },
    {
      "epoch": 9.252633753356745,
      "grad_norm": 16.74396324157715,
      "learning_rate": 4.149910942176683e-06,
      "loss": 1.2687,
      "step": 55990
    },
    {
      "epoch": 9.254286304482545,
      "grad_norm": 33.51347732543945,
      "learning_rate": 4.140729723278063e-06,
      "loss": 1.4942,
      "step": 56000
    },
    {
      "epoch": 9.255938855608346,
      "grad_norm": 15.115519523620605,
      "learning_rate": 4.131548504379441e-06,
      "loss": 1.36,
      "step": 56010
    },
    {
      "epoch": 9.257591406734146,
      "grad_norm": 25.56139373779297,
      "learning_rate": 4.122367285480821e-06,
      "loss": 1.4605,
      "step": 56020
    },
    {
      "epoch": 9.259243957859946,
      "grad_norm": 20.03956413269043,
      "learning_rate": 4.1131860665821995e-06,
      "loss": 1.3127,
      "step": 56030
    },
    {
      "epoch": 9.260896508985747,
      "grad_norm": 14.782129287719727,
      "learning_rate": 4.104004847683579e-06,
      "loss": 1.4149,
      "step": 56040
    },
    {
      "epoch": 9.262549060111548,
      "grad_norm": 16.4676570892334,
      "learning_rate": 4.094823628784958e-06,
      "loss": 1.3314,
      "step": 56050
    },
    {
      "epoch": 9.264201611237347,
      "grad_norm": 18.199886322021484,
      "learning_rate": 4.0856424098863364e-06,
      "loss": 1.3737,
      "step": 56060
    },
    {
      "epoch": 9.265854162363148,
      "grad_norm": 19.445627212524414,
      "learning_rate": 4.076461190987716e-06,
      "loss": 1.3664,
      "step": 56070
    },
    {
      "epoch": 9.267506713488949,
      "grad_norm": 14.629560470581055,
      "learning_rate": 4.067279972089095e-06,
      "loss": 1.3397,
      "step": 56080
    },
    {
      "epoch": 9.26915926461475,
      "grad_norm": 22.30562400817871,
      "learning_rate": 4.058098753190473e-06,
      "loss": 1.3311,
      "step": 56090
    },
    {
      "epoch": 9.270811815740549,
      "grad_norm": 9.496840476989746,
      "learning_rate": 4.0489175342918534e-06,
      "loss": 1.1589,
      "step": 56100
    },
    {
      "epoch": 9.27246436686635,
      "grad_norm": 13.570216178894043,
      "learning_rate": 4.039736315393232e-06,
      "loss": 1.3121,
      "step": 56110
    },
    {
      "epoch": 9.274116917992151,
      "grad_norm": 23.2502384185791,
      "learning_rate": 4.030555096494611e-06,
      "loss": 1.3267,
      "step": 56120
    },
    {
      "epoch": 9.27576946911795,
      "grad_norm": 15.193344116210938,
      "learning_rate": 4.02137387759599e-06,
      "loss": 1.308,
      "step": 56130
    },
    {
      "epoch": 9.277422020243751,
      "grad_norm": 18.408641815185547,
      "learning_rate": 4.012192658697369e-06,
      "loss": 1.3684,
      "step": 56140
    },
    {
      "epoch": 9.279074571369552,
      "grad_norm": 16.26186180114746,
      "learning_rate": 4.003011439798748e-06,
      "loss": 1.3184,
      "step": 56150
    },
    {
      "epoch": 9.280727122495351,
      "grad_norm": 15.509997367858887,
      "learning_rate": 3.993830220900127e-06,
      "loss": 1.3755,
      "step": 56160
    },
    {
      "epoch": 9.282379673621152,
      "grad_norm": 10.246735572814941,
      "learning_rate": 3.984649002001506e-06,
      "loss": 1.4041,
      "step": 56170
    },
    {
      "epoch": 9.284032224746953,
      "grad_norm": 14.112131118774414,
      "learning_rate": 3.975467783102885e-06,
      "loss": 1.2511,
      "step": 56180
    },
    {
      "epoch": 9.285684775872754,
      "grad_norm": 14.345402717590332,
      "learning_rate": 3.966286564204264e-06,
      "loss": 1.3432,
      "step": 56190
    },
    {
      "epoch": 9.287337326998554,
      "grad_norm": 13.019506454467773,
      "learning_rate": 3.957105345305643e-06,
      "loss": 1.4026,
      "step": 56200
    },
    {
      "epoch": 9.288989878124355,
      "grad_norm": 10.218276023864746,
      "learning_rate": 3.947924126407022e-06,
      "loss": 1.3788,
      "step": 56210
    },
    {
      "epoch": 9.290642429250155,
      "grad_norm": 10.781867980957031,
      "learning_rate": 3.938742907508401e-06,
      "loss": 1.2091,
      "step": 56220
    },
    {
      "epoch": 9.292294980375955,
      "grad_norm": 8.732402801513672,
      "learning_rate": 3.92956168860978e-06,
      "loss": 1.2941,
      "step": 56230
    },
    {
      "epoch": 9.293947531501756,
      "grad_norm": 19.886987686157227,
      "learning_rate": 3.920380469711159e-06,
      "loss": 1.2725,
      "step": 56240
    },
    {
      "epoch": 9.295600082627557,
      "grad_norm": 13.052376747131348,
      "learning_rate": 3.911199250812538e-06,
      "loss": 1.2563,
      "step": 56250
    },
    {
      "epoch": 9.297252633753356,
      "grad_norm": 9.23311710357666,
      "learning_rate": 3.902018031913917e-06,
      "loss": 1.2777,
      "step": 56260
    },
    {
      "epoch": 9.298905184879157,
      "grad_norm": 17.674257278442383,
      "learning_rate": 3.8928368130152956e-06,
      "loss": 1.1415,
      "step": 56270
    },
    {
      "epoch": 9.300557736004958,
      "grad_norm": 13.015941619873047,
      "learning_rate": 3.883655594116676e-06,
      "loss": 1.2718,
      "step": 56280
    },
    {
      "epoch": 9.302210287130759,
      "grad_norm": 13.032365798950195,
      "learning_rate": 3.874474375218054e-06,
      "loss": 1.3148,
      "step": 56290
    },
    {
      "epoch": 9.303862838256558,
      "grad_norm": 13.440291404724121,
      "learning_rate": 3.865293156319433e-06,
      "loss": 1.2976,
      "step": 56300
    },
    {
      "epoch": 9.305515389382359,
      "grad_norm": 20.601293563842773,
      "learning_rate": 3.8561119374208125e-06,
      "loss": 1.2242,
      "step": 56310
    },
    {
      "epoch": 9.30716794050816,
      "grad_norm": 14.70266342163086,
      "learning_rate": 3.846930718522191e-06,
      "loss": 1.3573,
      "step": 56320
    },
    {
      "epoch": 9.30882049163396,
      "grad_norm": 18.35739517211914,
      "learning_rate": 3.83774949962357e-06,
      "loss": 1.2484,
      "step": 56330
    },
    {
      "epoch": 9.31047304275976,
      "grad_norm": 14.594882011413574,
      "learning_rate": 3.8285682807249494e-06,
      "loss": 1.3821,
      "step": 56340
    },
    {
      "epoch": 9.312125593885561,
      "grad_norm": 17.77569007873535,
      "learning_rate": 3.819387061826328e-06,
      "loss": 1.2919,
      "step": 56350
    },
    {
      "epoch": 9.313778145011362,
      "grad_norm": 15.527255058288574,
      "learning_rate": 3.8102058429277075e-06,
      "loss": 1.29,
      "step": 56360
    },
    {
      "epoch": 9.315430696137161,
      "grad_norm": 15.60554313659668,
      "learning_rate": 3.8010246240290863e-06,
      "loss": 1.2916,
      "step": 56370
    },
    {
      "epoch": 9.317083247262962,
      "grad_norm": 14.84151554107666,
      "learning_rate": 3.791843405130465e-06,
      "loss": 1.2193,
      "step": 56380
    },
    {
      "epoch": 9.318735798388763,
      "grad_norm": 15.908535957336426,
      "learning_rate": 3.7826621862318444e-06,
      "loss": 1.3779,
      "step": 56390
    },
    {
      "epoch": 9.320388349514563,
      "grad_norm": 20.001300811767578,
      "learning_rate": 3.7734809673332232e-06,
      "loss": 1.3879,
      "step": 56400
    },
    {
      "epoch": 9.322040900640364,
      "grad_norm": 25.633216857910156,
      "learning_rate": 3.764299748434602e-06,
      "loss": 1.4137,
      "step": 56410
    },
    {
      "epoch": 9.323693451766164,
      "grad_norm": 14.804731369018555,
      "learning_rate": 3.7551185295359817e-06,
      "loss": 1.3807,
      "step": 56420
    },
    {
      "epoch": 9.325346002891964,
      "grad_norm": 17.46161651611328,
      "learning_rate": 3.74593731063736e-06,
      "loss": 1.2958,
      "step": 56430
    },
    {
      "epoch": 9.326998554017765,
      "grad_norm": 29.73974609375,
      "learning_rate": 3.736756091738739e-06,
      "loss": 1.3127,
      "step": 56440
    },
    {
      "epoch": 9.328651105143566,
      "grad_norm": 14.454090118408203,
      "learning_rate": 3.7275748728401186e-06,
      "loss": 1.3697,
      "step": 56450
    },
    {
      "epoch": 9.330303656269367,
      "grad_norm": 16.23573875427246,
      "learning_rate": 3.7183936539414975e-06,
      "loss": 1.3834,
      "step": 56460
    },
    {
      "epoch": 9.331956207395166,
      "grad_norm": 31.776811599731445,
      "learning_rate": 3.7092124350428763e-06,
      "loss": 1.3624,
      "step": 56470
    },
    {
      "epoch": 9.333608758520967,
      "grad_norm": 17.408411026000977,
      "learning_rate": 3.7000312161442555e-06,
      "loss": 1.13,
      "step": 56480
    },
    {
      "epoch": 9.335261309646768,
      "grad_norm": 14.618858337402344,
      "learning_rate": 3.6908499972456343e-06,
      "loss": 1.2647,
      "step": 56490
    },
    {
      "epoch": 9.336913860772567,
      "grad_norm": 15.05654239654541,
      "learning_rate": 3.681668778347014e-06,
      "loss": 1.2852,
      "step": 56500
    },
    {
      "epoch": 9.338566411898368,
      "grad_norm": 12.420156478881836,
      "learning_rate": 3.6724875594483924e-06,
      "loss": 1.43,
      "step": 56510
    },
    {
      "epoch": 9.340218963024169,
      "grad_norm": 15.04156494140625,
      "learning_rate": 3.6633063405497712e-06,
      "loss": 1.207,
      "step": 56520
    },
    {
      "epoch": 9.341871514149968,
      "grad_norm": 8.728355407714844,
      "learning_rate": 3.654125121651151e-06,
      "loss": 1.2025,
      "step": 56530
    },
    {
      "epoch": 9.34352406527577,
      "grad_norm": 21.201353073120117,
      "learning_rate": 3.6449439027525297e-06,
      "loss": 1.0984,
      "step": 56540
    },
    {
      "epoch": 9.34517661640157,
      "grad_norm": 11.79217529296875,
      "learning_rate": 3.6357626838539086e-06,
      "loss": 1.2759,
      "step": 56550
    },
    {
      "epoch": 9.346829167527371,
      "grad_norm": 13.807501792907715,
      "learning_rate": 3.626581464955288e-06,
      "loss": 1.2994,
      "step": 56560
    },
    {
      "epoch": 9.34848171865317,
      "grad_norm": 19.914011001586914,
      "learning_rate": 3.6174002460566666e-06,
      "loss": 1.3291,
      "step": 56570
    },
    {
      "epoch": 9.350134269778971,
      "grad_norm": 16.1053409576416,
      "learning_rate": 3.6082190271580455e-06,
      "loss": 1.2639,
      "step": 56580
    },
    {
      "epoch": 9.351786820904772,
      "grad_norm": 17.01410484313965,
      "learning_rate": 3.5990378082594247e-06,
      "loss": 1.4563,
      "step": 56590
    },
    {
      "epoch": 9.353439372030572,
      "grad_norm": 12.761663436889648,
      "learning_rate": 3.5898565893608035e-06,
      "loss": 1.4376,
      "step": 56600
    },
    {
      "epoch": 9.355091923156373,
      "grad_norm": 12.494538307189941,
      "learning_rate": 3.5806753704621824e-06,
      "loss": 1.4416,
      "step": 56610
    },
    {
      "epoch": 9.356744474282173,
      "grad_norm": 15.235856056213379,
      "learning_rate": 3.571494151563562e-06,
      "loss": 1.2971,
      "step": 56620
    },
    {
      "epoch": 9.358397025407974,
      "grad_norm": 22.249408721923828,
      "learning_rate": 3.562312932664941e-06,
      "loss": 1.3123,
      "step": 56630
    },
    {
      "epoch": 9.360049576533774,
      "grad_norm": 13.691697120666504,
      "learning_rate": 3.5531317137663197e-06,
      "loss": 1.4081,
      "step": 56640
    },
    {
      "epoch": 9.361702127659575,
      "grad_norm": 14.822437286376953,
      "learning_rate": 3.543950494867699e-06,
      "loss": 1.2504,
      "step": 56650
    },
    {
      "epoch": 9.363354678785376,
      "grad_norm": 13.317008018493652,
      "learning_rate": 3.5347692759690777e-06,
      "loss": 1.2689,
      "step": 56660
    },
    {
      "epoch": 9.365007229911175,
      "grad_norm": 19.28122901916504,
      "learning_rate": 3.5255880570704566e-06,
      "loss": 1.3562,
      "step": 56670
    },
    {
      "epoch": 9.366659781036976,
      "grad_norm": 14.012068748474121,
      "learning_rate": 3.516406838171836e-06,
      "loss": 1.3528,
      "step": 56680
    },
    {
      "epoch": 9.368312332162777,
      "grad_norm": 16.407472610473633,
      "learning_rate": 3.5072256192732146e-06,
      "loss": 1.1581,
      "step": 56690
    },
    {
      "epoch": 9.369964883288576,
      "grad_norm": 13.19459342956543,
      "learning_rate": 3.4980444003745943e-06,
      "loss": 1.2654,
      "step": 56700
    },
    {
      "epoch": 9.371617434414377,
      "grad_norm": 11.365413665771484,
      "learning_rate": 3.488863181475973e-06,
      "loss": 1.3031,
      "step": 56710
    },
    {
      "epoch": 9.373269985540178,
      "grad_norm": 11.131091117858887,
      "learning_rate": 3.479681962577352e-06,
      "loss": 1.3596,
      "step": 56720
    },
    {
      "epoch": 9.374922536665977,
      "grad_norm": 13.381197929382324,
      "learning_rate": 3.470500743678731e-06,
      "loss": 1.3861,
      "step": 56730
    },
    {
      "epoch": 9.376575087791778,
      "grad_norm": 9.68941879272461,
      "learning_rate": 3.46131952478011e-06,
      "loss": 1.3179,
      "step": 56740
    },
    {
      "epoch": 9.37822763891758,
      "grad_norm": 15.212055206298828,
      "learning_rate": 3.452138305881489e-06,
      "loss": 1.2758,
      "step": 56750
    },
    {
      "epoch": 9.37988019004338,
      "grad_norm": 17.098337173461914,
      "learning_rate": 3.442957086982868e-06,
      "loss": 1.1934,
      "step": 56760
    },
    {
      "epoch": 9.38153274116918,
      "grad_norm": 10.168761253356934,
      "learning_rate": 3.433775868084247e-06,
      "loss": 1.3831,
      "step": 56770
    },
    {
      "epoch": 9.38318529229498,
      "grad_norm": 50.241641998291016,
      "learning_rate": 3.4245946491856258e-06,
      "loss": 1.4488,
      "step": 56780
    },
    {
      "epoch": 9.384837843420781,
      "grad_norm": 18.76384162902832,
      "learning_rate": 3.4154134302870054e-06,
      "loss": 1.3276,
      "step": 56790
    },
    {
      "epoch": 9.38649039454658,
      "grad_norm": 16.12409782409668,
      "learning_rate": 3.4062322113883842e-06,
      "loss": 1.2093,
      "step": 56800
    },
    {
      "epoch": 9.388142945672382,
      "grad_norm": 16.562042236328125,
      "learning_rate": 3.3970509924897626e-06,
      "loss": 1.1709,
      "step": 56810
    },
    {
      "epoch": 9.389795496798182,
      "grad_norm": 26.49726104736328,
      "learning_rate": 3.3878697735911423e-06,
      "loss": 1.3279,
      "step": 56820
    },
    {
      "epoch": 9.391448047923983,
      "grad_norm": 13.345320701599121,
      "learning_rate": 3.378688554692521e-06,
      "loss": 1.2046,
      "step": 56830
    },
    {
      "epoch": 9.393100599049783,
      "grad_norm": 16.027231216430664,
      "learning_rate": 3.3695073357939e-06,
      "loss": 1.3354,
      "step": 56840
    },
    {
      "epoch": 9.394753150175584,
      "grad_norm": 10.626627922058105,
      "learning_rate": 3.3603261168952792e-06,
      "loss": 1.168,
      "step": 56850
    },
    {
      "epoch": 9.396405701301385,
      "grad_norm": 18.5012264251709,
      "learning_rate": 3.351144897996658e-06,
      "loss": 1.4278,
      "step": 56860
    },
    {
      "epoch": 9.398058252427184,
      "grad_norm": 21.47645378112793,
      "learning_rate": 3.341963679098037e-06,
      "loss": 1.3033,
      "step": 56870
    },
    {
      "epoch": 9.399710803552985,
      "grad_norm": 17.78287696838379,
      "learning_rate": 3.3327824601994165e-06,
      "loss": 1.2744,
      "step": 56880
    },
    {
      "epoch": 9.401363354678786,
      "grad_norm": 14.139810562133789,
      "learning_rate": 3.323601241300795e-06,
      "loss": 1.4479,
      "step": 56890
    },
    {
      "epoch": 9.403015905804585,
      "grad_norm": 28.748075485229492,
      "learning_rate": 3.3144200224021746e-06,
      "loss": 1.2902,
      "step": 56900
    },
    {
      "epoch": 9.404668456930386,
      "grad_norm": 46.95508575439453,
      "learning_rate": 3.3052388035035534e-06,
      "loss": 1.4943,
      "step": 56910
    },
    {
      "epoch": 9.406321008056187,
      "grad_norm": 15.258516311645508,
      "learning_rate": 3.2960575846049323e-06,
      "loss": 1.1904,
      "step": 56920
    },
    {
      "epoch": 9.407973559181988,
      "grad_norm": 23.977991104125977,
      "learning_rate": 3.2868763657063115e-06,
      "loss": 1.2751,
      "step": 56930
    },
    {
      "epoch": 9.409626110307787,
      "grad_norm": 12.30024242401123,
      "learning_rate": 3.2776951468076903e-06,
      "loss": 1.2913,
      "step": 56940
    },
    {
      "epoch": 9.411278661433588,
      "grad_norm": 11.82300853729248,
      "learning_rate": 3.268513927909069e-06,
      "loss": 1.4331,
      "step": 56950
    },
    {
      "epoch": 9.41293121255939,
      "grad_norm": 17.451065063476562,
      "learning_rate": 3.259332709010449e-06,
      "loss": 1.2351,
      "step": 56960
    },
    {
      "epoch": 9.414583763685188,
      "grad_norm": 13.529333114624023,
      "learning_rate": 3.2501514901118272e-06,
      "loss": 1.1954,
      "step": 56970
    },
    {
      "epoch": 9.41623631481099,
      "grad_norm": 14.515869140625,
      "learning_rate": 3.240970271213206e-06,
      "loss": 1.3682,
      "step": 56980
    },
    {
      "epoch": 9.41788886593679,
      "grad_norm": 11.694517135620117,
      "learning_rate": 3.2317890523145857e-06,
      "loss": 1.3659,
      "step": 56990
    },
    {
      "epoch": 9.41954141706259,
      "grad_norm": 27.488210678100586,
      "learning_rate": 3.2226078334159645e-06,
      "loss": 1.1186,
      "step": 57000
    },
    {
      "epoch": 9.42119396818839,
      "grad_norm": 15.681142807006836,
      "learning_rate": 3.2134266145173434e-06,
      "loss": 1.2438,
      "step": 57010
    },
    {
      "epoch": 9.422846519314191,
      "grad_norm": 19.548660278320312,
      "learning_rate": 3.2042453956187226e-06,
      "loss": 1.3773,
      "step": 57020
    },
    {
      "epoch": 9.424499070439992,
      "grad_norm": 23.549972534179688,
      "learning_rate": 3.1950641767201014e-06,
      "loss": 1.308,
      "step": 57030
    },
    {
      "epoch": 9.426151621565792,
      "grad_norm": 19.313648223876953,
      "learning_rate": 3.1858829578214803e-06,
      "loss": 1.2348,
      "step": 57040
    },
    {
      "epoch": 9.427804172691593,
      "grad_norm": 32.71291732788086,
      "learning_rate": 3.1767017389228595e-06,
      "loss": 1.1344,
      "step": 57050
    },
    {
      "epoch": 9.429456723817394,
      "grad_norm": 14.130952835083008,
      "learning_rate": 3.1675205200242383e-06,
      "loss": 1.3653,
      "step": 57060
    },
    {
      "epoch": 9.431109274943193,
      "grad_norm": 17.445035934448242,
      "learning_rate": 3.158339301125618e-06,
      "loss": 1.3794,
      "step": 57070
    },
    {
      "epoch": 9.432761826068994,
      "grad_norm": 14.413025856018066,
      "learning_rate": 3.149158082226997e-06,
      "loss": 1.3055,
      "step": 57080
    },
    {
      "epoch": 9.434414377194795,
      "grad_norm": 17.203859329223633,
      "learning_rate": 3.1399768633283757e-06,
      "loss": 1.3073,
      "step": 57090
    },
    {
      "epoch": 9.436066928320596,
      "grad_norm": 13.873775482177734,
      "learning_rate": 3.130795644429755e-06,
      "loss": 1.2085,
      "step": 57100
    },
    {
      "epoch": 9.437719479446395,
      "grad_norm": 15.375999450683594,
      "learning_rate": 3.1216144255311337e-06,
      "loss": 1.326,
      "step": 57110
    },
    {
      "epoch": 9.439372030572196,
      "grad_norm": 17.9966983795166,
      "learning_rate": 3.112433206632513e-06,
      "loss": 1.2816,
      "step": 57120
    },
    {
      "epoch": 9.441024581697997,
      "grad_norm": 19.590351104736328,
      "learning_rate": 3.103251987733892e-06,
      "loss": 1.2491,
      "step": 57130
    },
    {
      "epoch": 9.442677132823796,
      "grad_norm": 19.17728614807129,
      "learning_rate": 3.0940707688352706e-06,
      "loss": 1.3321,
      "step": 57140
    },
    {
      "epoch": 9.444329683949597,
      "grad_norm": 17.96892547607422,
      "learning_rate": 3.08488954993665e-06,
      "loss": 1.2419,
      "step": 57150
    },
    {
      "epoch": 9.445982235075398,
      "grad_norm": 19.752483367919922,
      "learning_rate": 3.0757083310380287e-06,
      "loss": 1.3689,
      "step": 57160
    },
    {
      "epoch": 9.447634786201197,
      "grad_norm": 16.997968673706055,
      "learning_rate": 3.066527112139408e-06,
      "loss": 1.3388,
      "step": 57170
    },
    {
      "epoch": 9.449287337326998,
      "grad_norm": 13.03810977935791,
      "learning_rate": 3.0573458932407868e-06,
      "loss": 1.2395,
      "step": 57180
    },
    {
      "epoch": 9.4509398884528,
      "grad_norm": 21.81022071838379,
      "learning_rate": 3.0481646743421656e-06,
      "loss": 1.2308,
      "step": 57190
    },
    {
      "epoch": 9.4525924395786,
      "grad_norm": 19.233882904052734,
      "learning_rate": 3.038983455443545e-06,
      "loss": 1.135,
      "step": 57200
    },
    {
      "epoch": 9.4542449907044,
      "grad_norm": 19.100370407104492,
      "learning_rate": 3.029802236544924e-06,
      "loss": 1.1726,
      "step": 57210
    },
    {
      "epoch": 9.4558975418302,
      "grad_norm": 10.6184720993042,
      "learning_rate": 3.020621017646303e-06,
      "loss": 1.2419,
      "step": 57220
    },
    {
      "epoch": 9.457550092956001,
      "grad_norm": 23.119884490966797,
      "learning_rate": 3.0114397987476817e-06,
      "loss": 1.253,
      "step": 57230
    },
    {
      "epoch": 9.4592026440818,
      "grad_norm": 15.129219055175781,
      "learning_rate": 3.002258579849061e-06,
      "loss": 1.3053,
      "step": 57240
    },
    {
      "epoch": 9.460855195207602,
      "grad_norm": 16.50132179260254,
      "learning_rate": 2.9930773609504402e-06,
      "loss": 1.2244,
      "step": 57250
    },
    {
      "epoch": 9.462507746333403,
      "grad_norm": 12.94861125946045,
      "learning_rate": 2.9838961420518186e-06,
      "loss": 1.3387,
      "step": 57260
    },
    {
      "epoch": 9.464160297459202,
      "grad_norm": 17.04841423034668,
      "learning_rate": 2.974714923153198e-06,
      "loss": 1.3652,
      "step": 57270
    },
    {
      "epoch": 9.465812848585003,
      "grad_norm": 10.876708984375,
      "learning_rate": 2.965533704254577e-06,
      "loss": 1.2098,
      "step": 57280
    },
    {
      "epoch": 9.467465399710804,
      "grad_norm": 11.755070686340332,
      "learning_rate": 2.956352485355956e-06,
      "loss": 1.299,
      "step": 57290
    },
    {
      "epoch": 9.469117950836605,
      "grad_norm": 10.822213172912598,
      "learning_rate": 2.9471712664573348e-06,
      "loss": 1.2336,
      "step": 57300
    },
    {
      "epoch": 9.470770501962404,
      "grad_norm": 19.71744155883789,
      "learning_rate": 2.937990047558714e-06,
      "loss": 1.3242,
      "step": 57310
    },
    {
      "epoch": 9.472423053088205,
      "grad_norm": 21.902231216430664,
      "learning_rate": 2.9288088286600933e-06,
      "loss": 1.4042,
      "step": 57320
    },
    {
      "epoch": 9.474075604214006,
      "grad_norm": 28.398181915283203,
      "learning_rate": 2.919627609761472e-06,
      "loss": 1.292,
      "step": 57330
    },
    {
      "epoch": 9.475728155339805,
      "grad_norm": 17.9105281829834,
      "learning_rate": 2.910446390862851e-06,
      "loss": 1.4072,
      "step": 57340
    },
    {
      "epoch": 9.477380706465606,
      "grad_norm": 24.03833770751953,
      "learning_rate": 2.90126517196423e-06,
      "loss": 1.3059,
      "step": 57350
    },
    {
      "epoch": 9.479033257591407,
      "grad_norm": 18.128108978271484,
      "learning_rate": 2.892083953065609e-06,
      "loss": 1.3698,
      "step": 57360
    },
    {
      "epoch": 9.480685808717206,
      "grad_norm": 15.403632164001465,
      "learning_rate": 2.8829027341669882e-06,
      "loss": 1.2558,
      "step": 57370
    },
    {
      "epoch": 9.482338359843007,
      "grad_norm": 12.787527084350586,
      "learning_rate": 2.873721515268367e-06,
      "loss": 1.3369,
      "step": 57380
    },
    {
      "epoch": 9.483990910968808,
      "grad_norm": 13.58300495147705,
      "learning_rate": 2.864540296369746e-06,
      "loss": 1.2305,
      "step": 57390
    },
    {
      "epoch": 9.48564346209461,
      "grad_norm": 8.516387939453125,
      "learning_rate": 2.855359077471125e-06,
      "loss": 1.184,
      "step": 57400
    },
    {
      "epoch": 9.487296013220408,
      "grad_norm": 12.774700164794922,
      "learning_rate": 2.8461778585725044e-06,
      "loss": 1.3685,
      "step": 57410
    },
    {
      "epoch": 9.48894856434621,
      "grad_norm": 18.586366653442383,
      "learning_rate": 2.836996639673883e-06,
      "loss": 1.3613,
      "step": 57420
    },
    {
      "epoch": 9.49060111547201,
      "grad_norm": 19.410852432250977,
      "learning_rate": 2.827815420775262e-06,
      "loss": 1.3254,
      "step": 57430
    },
    {
      "epoch": 9.49225366659781,
      "grad_norm": 37.648040771484375,
      "learning_rate": 2.8186342018766413e-06,
      "loss": 1.3388,
      "step": 57440
    },
    {
      "epoch": 9.49390621772361,
      "grad_norm": 11.228830337524414,
      "learning_rate": 2.8094529829780205e-06,
      "loss": 1.2954,
      "step": 57450
    },
    {
      "epoch": 9.495558768849412,
      "grad_norm": 17.186487197875977,
      "learning_rate": 2.8002717640793993e-06,
      "loss": 1.3855,
      "step": 57460
    },
    {
      "epoch": 9.49721131997521,
      "grad_norm": 14.050568580627441,
      "learning_rate": 2.791090545180778e-06,
      "loss": 1.3872,
      "step": 57470
    },
    {
      "epoch": 9.498863871101012,
      "grad_norm": 57.6312370300293,
      "learning_rate": 2.7819093262821574e-06,
      "loss": 1.3453,
      "step": 57480
    },
    {
      "epoch": 9.500516422226813,
      "grad_norm": 16.523967742919922,
      "learning_rate": 2.7727281073835362e-06,
      "loss": 1.3147,
      "step": 57490
    },
    {
      "epoch": 9.502168973352614,
      "grad_norm": 18.444759368896484,
      "learning_rate": 2.7635468884849155e-06,
      "loss": 1.3901,
      "step": 57500
    },
    {
      "epoch": 9.503821524478413,
      "grad_norm": 10.994654655456543,
      "learning_rate": 2.7543656695862943e-06,
      "loss": 1.3167,
      "step": 57510
    },
    {
      "epoch": 9.505474075604214,
      "grad_norm": 12.720823287963867,
      "learning_rate": 2.7451844506876736e-06,
      "loss": 1.2899,
      "step": 57520
    },
    {
      "epoch": 9.507126626730015,
      "grad_norm": 16.59524917602539,
      "learning_rate": 2.7360032317890524e-06,
      "loss": 1.25,
      "step": 57530
    },
    {
      "epoch": 9.508779177855814,
      "grad_norm": 14.175232887268066,
      "learning_rate": 2.7268220128904316e-06,
      "loss": 1.3248,
      "step": 57540
    },
    {
      "epoch": 9.510431728981615,
      "grad_norm": 16.468807220458984,
      "learning_rate": 2.7176407939918105e-06,
      "loss": 1.3374,
      "step": 57550
    },
    {
      "epoch": 9.512084280107416,
      "grad_norm": 26.731430053710938,
      "learning_rate": 2.7084595750931893e-06,
      "loss": 1.2489,
      "step": 57560
    },
    {
      "epoch": 9.513736831233217,
      "grad_norm": 9.305521011352539,
      "learning_rate": 2.6992783561945685e-06,
      "loss": 1.3049,
      "step": 57570
    },
    {
      "epoch": 9.515389382359016,
      "grad_norm": 12.240220069885254,
      "learning_rate": 2.6900971372959478e-06,
      "loss": 1.3189,
      "step": 57580
    },
    {
      "epoch": 9.517041933484817,
      "grad_norm": 76.72401428222656,
      "learning_rate": 2.6809159183973266e-06,
      "loss": 1.3185,
      "step": 57590
    },
    {
      "epoch": 9.518694484610618,
      "grad_norm": 13.349190711975098,
      "learning_rate": 2.6717346994987054e-06,
      "loss": 1.4297,
      "step": 57600
    },
    {
      "epoch": 9.520347035736417,
      "grad_norm": 21.3134765625,
      "learning_rate": 2.6625534806000847e-06,
      "loss": 1.1601,
      "step": 57610
    },
    {
      "epoch": 9.521999586862218,
      "grad_norm": 11.682439804077148,
      "learning_rate": 2.653372261701464e-06,
      "loss": 1.2994,
      "step": 57620
    },
    {
      "epoch": 9.52365213798802,
      "grad_norm": 17.29874038696289,
      "learning_rate": 2.6441910428028427e-06,
      "loss": 1.3704,
      "step": 57630
    },
    {
      "epoch": 9.525304689113819,
      "grad_norm": 9.633611679077148,
      "learning_rate": 2.6350098239042216e-06,
      "loss": 1.2424,
      "step": 57640
    },
    {
      "epoch": 9.52695724023962,
      "grad_norm": 73.63542175292969,
      "learning_rate": 2.625828605005601e-06,
      "loss": 1.3749,
      "step": 57650
    },
    {
      "epoch": 9.52860979136542,
      "grad_norm": 20.629478454589844,
      "learning_rate": 2.6166473861069796e-06,
      "loss": 1.2985,
      "step": 57660
    },
    {
      "epoch": 9.530262342491222,
      "grad_norm": 15.100749969482422,
      "learning_rate": 2.607466167208359e-06,
      "loss": 1.272,
      "step": 57670
    },
    {
      "epoch": 9.53191489361702,
      "grad_norm": 16.76715850830078,
      "learning_rate": 2.5982849483097377e-06,
      "loss": 1.283,
      "step": 57680
    },
    {
      "epoch": 9.533567444742822,
      "grad_norm": 23.314260482788086,
      "learning_rate": 2.5891037294111165e-06,
      "loss": 1.3745,
      "step": 57690
    },
    {
      "epoch": 9.535219995868623,
      "grad_norm": 25.0333194732666,
      "learning_rate": 2.5799225105124958e-06,
      "loss": 1.1921,
      "step": 57700
    },
    {
      "epoch": 9.536872546994422,
      "grad_norm": 14.260005950927734,
      "learning_rate": 2.570741291613875e-06,
      "loss": 1.2287,
      "step": 57710
    },
    {
      "epoch": 9.538525098120223,
      "grad_norm": 20.597248077392578,
      "learning_rate": 2.561560072715254e-06,
      "loss": 1.3879,
      "step": 57720
    },
    {
      "epoch": 9.540177649246024,
      "grad_norm": 15.62580680847168,
      "learning_rate": 2.5523788538166327e-06,
      "loss": 1.2893,
      "step": 57730
    },
    {
      "epoch": 9.541830200371823,
      "grad_norm": 11.68388557434082,
      "learning_rate": 2.543197634918012e-06,
      "loss": 1.2051,
      "step": 57740
    },
    {
      "epoch": 9.543482751497624,
      "grad_norm": 15.208742141723633,
      "learning_rate": 2.534016416019391e-06,
      "loss": 1.3074,
      "step": 57750
    },
    {
      "epoch": 9.545135302623425,
      "grad_norm": 15.40304946899414,
      "learning_rate": 2.5248351971207696e-06,
      "loss": 1.3454,
      "step": 57760
    },
    {
      "epoch": 9.546787853749226,
      "grad_norm": 21.844449996948242,
      "learning_rate": 2.515653978222149e-06,
      "loss": 1.1692,
      "step": 57770
    },
    {
      "epoch": 9.548440404875025,
      "grad_norm": 16.35841178894043,
      "learning_rate": 2.506472759323528e-06,
      "loss": 1.386,
      "step": 57780
    },
    {
      "epoch": 9.550092956000826,
      "grad_norm": 15.275668144226074,
      "learning_rate": 2.497291540424907e-06,
      "loss": 1.3673,
      "step": 57790
    },
    {
      "epoch": 9.551745507126627,
      "grad_norm": 13.76074504852295,
      "learning_rate": 2.4881103215262857e-06,
      "loss": 1.3039,
      "step": 57800
    },
    {
      "epoch": 9.553398058252426,
      "grad_norm": 14.345163345336914,
      "learning_rate": 2.478929102627665e-06,
      "loss": 1.4154,
      "step": 57810
    },
    {
      "epoch": 9.555050609378227,
      "grad_norm": 12.346527099609375,
      "learning_rate": 2.4697478837290442e-06,
      "loss": 1.2726,
      "step": 57820
    },
    {
      "epoch": 9.556703160504028,
      "grad_norm": 19.1199893951416,
      "learning_rate": 2.460566664830423e-06,
      "loss": 1.263,
      "step": 57830
    },
    {
      "epoch": 9.55835571162983,
      "grad_norm": 14.129487037658691,
      "learning_rate": 2.451385445931802e-06,
      "loss": 1.3791,
      "step": 57840
    },
    {
      "epoch": 9.560008262755629,
      "grad_norm": 15.209273338317871,
      "learning_rate": 2.442204227033181e-06,
      "loss": 1.2314,
      "step": 57850
    },
    {
      "epoch": 9.56166081388143,
      "grad_norm": 15.329292297363281,
      "learning_rate": 2.43302300813456e-06,
      "loss": 1.2872,
      "step": 57860
    },
    {
      "epoch": 9.56331336500723,
      "grad_norm": 16.666093826293945,
      "learning_rate": 2.423841789235939e-06,
      "loss": 1.3597,
      "step": 57870
    },
    {
      "epoch": 9.56496591613303,
      "grad_norm": 13.654024124145508,
      "learning_rate": 2.414660570337318e-06,
      "loss": 1.2514,
      "step": 57880
    },
    {
      "epoch": 9.56661846725883,
      "grad_norm": 12.923748016357422,
      "learning_rate": 2.405479351438697e-06,
      "loss": 1.1992,
      "step": 57890
    },
    {
      "epoch": 9.568271018384632,
      "grad_norm": 12.820647239685059,
      "learning_rate": 2.396298132540076e-06,
      "loss": 1.3571,
      "step": 57900
    },
    {
      "epoch": 9.569923569510431,
      "grad_norm": 16.327159881591797,
      "learning_rate": 2.3871169136414553e-06,
      "loss": 1.3175,
      "step": 57910
    },
    {
      "epoch": 9.571576120636232,
      "grad_norm": 24.59794807434082,
      "learning_rate": 2.377935694742834e-06,
      "loss": 1.3611,
      "step": 57920
    },
    {
      "epoch": 9.573228671762033,
      "grad_norm": 12.740336418151855,
      "learning_rate": 2.368754475844213e-06,
      "loss": 1.344,
      "step": 57930
    },
    {
      "epoch": 9.574881222887832,
      "grad_norm": 17.685501098632812,
      "learning_rate": 2.3595732569455922e-06,
      "loss": 1.394,
      "step": 57940
    },
    {
      "epoch": 9.576533774013633,
      "grad_norm": 19.866283416748047,
      "learning_rate": 2.3503920380469715e-06,
      "loss": 1.1501,
      "step": 57950
    },
    {
      "epoch": 9.578186325139434,
      "grad_norm": 14.622546195983887,
      "learning_rate": 2.3412108191483503e-06,
      "loss": 1.2882,
      "step": 57960
    },
    {
      "epoch": 9.579838876265235,
      "grad_norm": 16.59194564819336,
      "learning_rate": 2.332029600249729e-06,
      "loss": 1.3115,
      "step": 57970
    },
    {
      "epoch": 9.581491427391034,
      "grad_norm": 18.110830307006836,
      "learning_rate": 2.3228483813511084e-06,
      "loss": 1.2563,
      "step": 57980
    },
    {
      "epoch": 9.583143978516835,
      "grad_norm": 14.122636795043945,
      "learning_rate": 2.313667162452487e-06,
      "loss": 1.3451,
      "step": 57990
    },
    {
      "epoch": 9.584796529642636,
      "grad_norm": 14.215288162231445,
      "learning_rate": 2.3044859435538664e-06,
      "loss": 1.2917,
      "step": 58000
    },
    {
      "epoch": 9.586449080768435,
      "grad_norm": 100.20287322998047,
      "learning_rate": 2.2953047246552453e-06,
      "loss": 1.2025,
      "step": 58010
    },
    {
      "epoch": 9.588101631894236,
      "grad_norm": 12.98987102508545,
      "learning_rate": 2.2861235057566245e-06,
      "loss": 1.2286,
      "step": 58020
    },
    {
      "epoch": 9.589754183020037,
      "grad_norm": 17.084440231323242,
      "learning_rate": 2.2769422868580033e-06,
      "loss": 1.2378,
      "step": 58030
    },
    {
      "epoch": 9.591406734145838,
      "grad_norm": 17.176267623901367,
      "learning_rate": 2.2677610679593826e-06,
      "loss": 1.4078,
      "step": 58040
    },
    {
      "epoch": 9.593059285271638,
      "grad_norm": 38.53559112548828,
      "learning_rate": 2.2585798490607614e-06,
      "loss": 1.3422,
      "step": 58050
    },
    {
      "epoch": 9.594711836397439,
      "grad_norm": 11.070537567138672,
      "learning_rate": 2.2493986301621402e-06,
      "loss": 1.3479,
      "step": 58060
    },
    {
      "epoch": 9.59636438752324,
      "grad_norm": 19.474384307861328,
      "learning_rate": 2.2402174112635195e-06,
      "loss": 1.2747,
      "step": 58070
    },
    {
      "epoch": 9.598016938649039,
      "grad_norm": 15.456579208374023,
      "learning_rate": 2.2310361923648987e-06,
      "loss": 1.3125,
      "step": 58080
    },
    {
      "epoch": 9.59966948977484,
      "grad_norm": 15.075743675231934,
      "learning_rate": 2.221854973466277e-06,
      "loss": 1.3164,
      "step": 58090
    },
    {
      "epoch": 9.60132204090064,
      "grad_norm": 25.315460205078125,
      "learning_rate": 2.2126737545676564e-06,
      "loss": 1.347,
      "step": 58100
    },
    {
      "epoch": 9.60297459202644,
      "grad_norm": 21.648412704467773,
      "learning_rate": 2.2034925356690356e-06,
      "loss": 1.332,
      "step": 58110
    },
    {
      "epoch": 9.604627143152241,
      "grad_norm": 15.424687385559082,
      "learning_rate": 2.194311316770415e-06,
      "loss": 1.3377,
      "step": 58120
    },
    {
      "epoch": 9.606279694278042,
      "grad_norm": 36.263980865478516,
      "learning_rate": 2.1851300978717933e-06,
      "loss": 1.3506,
      "step": 58130
    },
    {
      "epoch": 9.607932245403843,
      "grad_norm": 28.31026840209961,
      "learning_rate": 2.1759488789731725e-06,
      "loss": 1.2666,
      "step": 58140
    },
    {
      "epoch": 9.609584796529642,
      "grad_norm": 17.763492584228516,
      "learning_rate": 2.1667676600745518e-06,
      "loss": 1.3172,
      "step": 58150
    },
    {
      "epoch": 9.611237347655443,
      "grad_norm": 16.161304473876953,
      "learning_rate": 2.1575864411759306e-06,
      "loss": 1.2002,
      "step": 58160
    },
    {
      "epoch": 9.612889898781244,
      "grad_norm": 15.862537384033203,
      "learning_rate": 2.1484052222773094e-06,
      "loss": 1.4035,
      "step": 58170
    },
    {
      "epoch": 9.614542449907043,
      "grad_norm": 13.134415626525879,
      "learning_rate": 2.1392240033786887e-06,
      "loss": 1.2955,
      "step": 58180
    },
    {
      "epoch": 9.616195001032844,
      "grad_norm": 14.605551719665527,
      "learning_rate": 2.1300427844800675e-06,
      "loss": 1.4105,
      "step": 58190
    },
    {
      "epoch": 9.617847552158645,
      "grad_norm": 18.962343215942383,
      "learning_rate": 2.1208615655814467e-06,
      "loss": 1.2919,
      "step": 58200
    },
    {
      "epoch": 9.619500103284444,
      "grad_norm": 13.857765197753906,
      "learning_rate": 2.1116803466828256e-06,
      "loss": 1.2513,
      "step": 58210
    },
    {
      "epoch": 9.621152654410245,
      "grad_norm": 23.394207000732422,
      "learning_rate": 2.102499127784205e-06,
      "loss": 1.3572,
      "step": 58220
    },
    {
      "epoch": 9.622805205536046,
      "grad_norm": 14.980634689331055,
      "learning_rate": 2.0933179088855836e-06,
      "loss": 1.2912,
      "step": 58230
    },
    {
      "epoch": 9.624457756661847,
      "grad_norm": 20.3941593170166,
      "learning_rate": 2.084136689986963e-06,
      "loss": 1.3388,
      "step": 58240
    },
    {
      "epoch": 9.626110307787647,
      "grad_norm": 14.584732055664062,
      "learning_rate": 2.0749554710883417e-06,
      "loss": 1.252,
      "step": 58250
    },
    {
      "epoch": 9.627762858913448,
      "grad_norm": 23.42914581298828,
      "learning_rate": 2.0657742521897205e-06,
      "loss": 1.2462,
      "step": 58260
    },
    {
      "epoch": 9.629415410039249,
      "grad_norm": 10.313939094543457,
      "learning_rate": 2.0565930332910998e-06,
      "loss": 1.3573,
      "step": 58270
    },
    {
      "epoch": 9.631067961165048,
      "grad_norm": 11.711010932922363,
      "learning_rate": 2.047411814392479e-06,
      "loss": 1.3872,
      "step": 58280
    },
    {
      "epoch": 9.632720512290849,
      "grad_norm": 16.497787475585938,
      "learning_rate": 2.038230595493858e-06,
      "loss": 1.3352,
      "step": 58290
    },
    {
      "epoch": 9.63437306341665,
      "grad_norm": 9.749482154846191,
      "learning_rate": 2.0290493765952367e-06,
      "loss": 1.2596,
      "step": 58300
    },
    {
      "epoch": 9.63602561454245,
      "grad_norm": 27.247074127197266,
      "learning_rate": 2.019868157696616e-06,
      "loss": 1.1729,
      "step": 58310
    },
    {
      "epoch": 9.63767816566825,
      "grad_norm": 28.22972869873047,
      "learning_rate": 2.010686938797995e-06,
      "loss": 1.2017,
      "step": 58320
    },
    {
      "epoch": 9.639330716794051,
      "grad_norm": 16.79508399963379,
      "learning_rate": 2.001505719899374e-06,
      "loss": 1.2413,
      "step": 58330
    },
    {
      "epoch": 9.640983267919852,
      "grad_norm": 15.924692153930664,
      "learning_rate": 1.992324501000753e-06,
      "loss": 1.2305,
      "step": 58340
    },
    {
      "epoch": 9.642635819045651,
      "grad_norm": 12.7982816696167,
      "learning_rate": 1.983143282102132e-06,
      "loss": 1.3867,
      "step": 58350
    },
    {
      "epoch": 9.644288370171452,
      "grad_norm": 15.45654296875,
      "learning_rate": 1.973962063203511e-06,
      "loss": 1.2799,
      "step": 58360
    },
    {
      "epoch": 9.645940921297253,
      "grad_norm": 16.585647583007812,
      "learning_rate": 1.96478084430489e-06,
      "loss": 1.3226,
      "step": 58370
    },
    {
      "epoch": 9.647593472423052,
      "grad_norm": 23.49603843688965,
      "learning_rate": 1.955599625406269e-06,
      "loss": 1.2294,
      "step": 58380
    },
    {
      "epoch": 9.649246023548853,
      "grad_norm": 15.694708824157715,
      "learning_rate": 1.9464184065076478e-06,
      "loss": 1.1692,
      "step": 58390
    },
    {
      "epoch": 9.650898574674654,
      "grad_norm": 13.695332527160645,
      "learning_rate": 1.937237187609027e-06,
      "loss": 1.2909,
      "step": 58400
    },
    {
      "epoch": 9.652551125800455,
      "grad_norm": 9.972579002380371,
      "learning_rate": 1.9280559687104063e-06,
      "loss": 1.2341,
      "step": 58410
    },
    {
      "epoch": 9.654203676926254,
      "grad_norm": 14.860621452331543,
      "learning_rate": 1.918874749811785e-06,
      "loss": 1.2629,
      "step": 58420
    },
    {
      "epoch": 9.655856228052055,
      "grad_norm": 11.948851585388184,
      "learning_rate": 1.909693530913164e-06,
      "loss": 1.2555,
      "step": 58430
    },
    {
      "epoch": 9.657508779177856,
      "grad_norm": 22.112714767456055,
      "learning_rate": 1.9005123120145432e-06,
      "loss": 1.3156,
      "step": 58440
    },
    {
      "epoch": 9.659161330303656,
      "grad_norm": 24.64064598083496,
      "learning_rate": 1.8913310931159222e-06,
      "loss": 1.4058,
      "step": 58450
    },
    {
      "epoch": 9.660813881429457,
      "grad_norm": 14.680100440979004,
      "learning_rate": 1.882149874217301e-06,
      "loss": 1.357,
      "step": 58460
    },
    {
      "epoch": 9.662466432555258,
      "grad_norm": 13.727468490600586,
      "learning_rate": 1.87296865531868e-06,
      "loss": 1.3138,
      "step": 58470
    },
    {
      "epoch": 9.664118983681057,
      "grad_norm": 17.917619705200195,
      "learning_rate": 1.8637874364200593e-06,
      "loss": 1.286,
      "step": 58480
    },
    {
      "epoch": 9.665771534806858,
      "grad_norm": 13.282672882080078,
      "learning_rate": 1.8546062175214381e-06,
      "loss": 1.253,
      "step": 58490
    },
    {
      "epoch": 9.667424085932659,
      "grad_norm": 13.979046821594238,
      "learning_rate": 1.8454249986228172e-06,
      "loss": 1.2243,
      "step": 58500
    },
    {
      "epoch": 9.66907663705846,
      "grad_norm": 16.821653366088867,
      "learning_rate": 1.8362437797241962e-06,
      "loss": 1.2361,
      "step": 58510
    },
    {
      "epoch": 9.670729188184259,
      "grad_norm": 14.560995101928711,
      "learning_rate": 1.8270625608255755e-06,
      "loss": 1.2431,
      "step": 58520
    },
    {
      "epoch": 9.67238173931006,
      "grad_norm": 15.146513938903809,
      "learning_rate": 1.8178813419269543e-06,
      "loss": 1.1976,
      "step": 58530
    },
    {
      "epoch": 9.674034290435861,
      "grad_norm": 15.486896514892578,
      "learning_rate": 1.8087001230283333e-06,
      "loss": 1.3143,
      "step": 58540
    },
    {
      "epoch": 9.67568684156166,
      "grad_norm": 18.165647506713867,
      "learning_rate": 1.7995189041297124e-06,
      "loss": 1.2991,
      "step": 58550
    },
    {
      "epoch": 9.677339392687461,
      "grad_norm": 17.75432586669922,
      "learning_rate": 1.7903376852310912e-06,
      "loss": 1.1735,
      "step": 58560
    },
    {
      "epoch": 9.678991943813262,
      "grad_norm": 17.992963790893555,
      "learning_rate": 1.7811564663324704e-06,
      "loss": 1.3001,
      "step": 58570
    },
    {
      "epoch": 9.680644494939063,
      "grad_norm": 65.19438171386719,
      "learning_rate": 1.7719752474338495e-06,
      "loss": 1.3303,
      "step": 58580
    },
    {
      "epoch": 9.682297046064862,
      "grad_norm": 18.94664192199707,
      "learning_rate": 1.7627940285352283e-06,
      "loss": 1.2772,
      "step": 58590
    },
    {
      "epoch": 9.683949597190663,
      "grad_norm": 22.916940689086914,
      "learning_rate": 1.7536128096366073e-06,
      "loss": 1.3774,
      "step": 58600
    },
    {
      "epoch": 9.685602148316464,
      "grad_norm": 17.095354080200195,
      "learning_rate": 1.7444315907379866e-06,
      "loss": 1.2863,
      "step": 58610
    },
    {
      "epoch": 9.687254699442263,
      "grad_norm": 12.96331787109375,
      "learning_rate": 1.7352503718393656e-06,
      "loss": 1.4177,
      "step": 58620
    },
    {
      "epoch": 9.688907250568064,
      "grad_norm": 20.66131019592285,
      "learning_rate": 1.7260691529407444e-06,
      "loss": 1.3123,
      "step": 58630
    },
    {
      "epoch": 9.690559801693865,
      "grad_norm": 15.276062965393066,
      "learning_rate": 1.7168879340421235e-06,
      "loss": 1.334,
      "step": 58640
    },
    {
      "epoch": 9.692212352819665,
      "grad_norm": 30.0185546875,
      "learning_rate": 1.7077067151435027e-06,
      "loss": 1.3818,
      "step": 58650
    },
    {
      "epoch": 9.693864903945466,
      "grad_norm": 14.805960655212402,
      "learning_rate": 1.6985254962448813e-06,
      "loss": 1.3392,
      "step": 58660
    },
    {
      "epoch": 9.695517455071267,
      "grad_norm": 17.313312530517578,
      "learning_rate": 1.6893442773462606e-06,
      "loss": 1.421,
      "step": 58670
    },
    {
      "epoch": 9.697170006197066,
      "grad_norm": 15.689269065856934,
      "learning_rate": 1.6801630584476396e-06,
      "loss": 1.375,
      "step": 58680
    },
    {
      "epoch": 9.698822557322867,
      "grad_norm": 27.808439254760742,
      "learning_rate": 1.6709818395490184e-06,
      "loss": 1.3845,
      "step": 58690
    },
    {
      "epoch": 9.700475108448668,
      "grad_norm": 16.34505271911621,
      "learning_rate": 1.6618006206503975e-06,
      "loss": 1.2094,
      "step": 58700
    },
    {
      "epoch": 9.702127659574469,
      "grad_norm": 30.347618103027344,
      "learning_rate": 1.6526194017517767e-06,
      "loss": 1.4331,
      "step": 58710
    },
    {
      "epoch": 9.703780210700268,
      "grad_norm": 32.67634201049805,
      "learning_rate": 1.6434381828531558e-06,
      "loss": 1.3411,
      "step": 58720
    },
    {
      "epoch": 9.705432761826069,
      "grad_norm": 13.503914833068848,
      "learning_rate": 1.6342569639545346e-06,
      "loss": 1.2915,
      "step": 58730
    },
    {
      "epoch": 9.70708531295187,
      "grad_norm": 30.984130859375,
      "learning_rate": 1.6250757450559136e-06,
      "loss": 1.2693,
      "step": 58740
    },
    {
      "epoch": 9.70873786407767,
      "grad_norm": 22.482847213745117,
      "learning_rate": 1.6158945261572929e-06,
      "loss": 1.2009,
      "step": 58750
    },
    {
      "epoch": 9.71039041520347,
      "grad_norm": 51.00187301635742,
      "learning_rate": 1.6067133072586717e-06,
      "loss": 1.3464,
      "step": 58760
    },
    {
      "epoch": 9.712042966329271,
      "grad_norm": 16.98764991760254,
      "learning_rate": 1.5975320883600507e-06,
      "loss": 1.2747,
      "step": 58770
    },
    {
      "epoch": 9.713695517455072,
      "grad_norm": 14.666571617126465,
      "learning_rate": 1.5883508694614298e-06,
      "loss": 1.2513,
      "step": 58780
    },
    {
      "epoch": 9.715348068580871,
      "grad_norm": 11.236737251281738,
      "learning_rate": 1.579169650562809e-06,
      "loss": 1.2613,
      "step": 58790
    },
    {
      "epoch": 9.717000619706672,
      "grad_norm": 22.248416900634766,
      "learning_rate": 1.5699884316641878e-06,
      "loss": 1.3245,
      "step": 58800
    },
    {
      "epoch": 9.718653170832473,
      "grad_norm": 10.243622779846191,
      "learning_rate": 1.5608072127655669e-06,
      "loss": 1.3724,
      "step": 58810
    },
    {
      "epoch": 9.720305721958272,
      "grad_norm": 16.06927490234375,
      "learning_rate": 1.551625993866946e-06,
      "loss": 1.3333,
      "step": 58820
    },
    {
      "epoch": 9.721958273084073,
      "grad_norm": 15.943928718566895,
      "learning_rate": 1.542444774968325e-06,
      "loss": 1.261,
      "step": 58830
    },
    {
      "epoch": 9.723610824209874,
      "grad_norm": 17.844417572021484,
      "learning_rate": 1.533263556069704e-06,
      "loss": 1.3135,
      "step": 58840
    },
    {
      "epoch": 9.725263375335674,
      "grad_norm": 16.729440689086914,
      "learning_rate": 1.5240823371710828e-06,
      "loss": 1.3187,
      "step": 58850
    },
    {
      "epoch": 9.726915926461475,
      "grad_norm": 15.814526557922363,
      "learning_rate": 1.514901118272462e-06,
      "loss": 1.3089,
      "step": 58860
    },
    {
      "epoch": 9.728568477587276,
      "grad_norm": 12.397698402404785,
      "learning_rate": 1.5057198993738409e-06,
      "loss": 1.3365,
      "step": 58870
    },
    {
      "epoch": 9.730221028713077,
      "grad_norm": 19.588912963867188,
      "learning_rate": 1.4965386804752201e-06,
      "loss": 1.3647,
      "step": 58880
    },
    {
      "epoch": 9.731873579838876,
      "grad_norm": 19.40388298034668,
      "learning_rate": 1.487357461576599e-06,
      "loss": 1.228,
      "step": 58890
    },
    {
      "epoch": 9.733526130964677,
      "grad_norm": 9.690013885498047,
      "learning_rate": 1.478176242677978e-06,
      "loss": 1.198,
      "step": 58900
    },
    {
      "epoch": 9.735178682090478,
      "grad_norm": 21.138465881347656,
      "learning_rate": 1.468995023779357e-06,
      "loss": 1.2491,
      "step": 58910
    },
    {
      "epoch": 9.736831233216277,
      "grad_norm": 13.240848541259766,
      "learning_rate": 1.459813804880736e-06,
      "loss": 1.0939,
      "step": 58920
    },
    {
      "epoch": 9.738483784342078,
      "grad_norm": 11.486268043518066,
      "learning_rate": 1.450632585982115e-06,
      "loss": 1.2259,
      "step": 58930
    },
    {
      "epoch": 9.740136335467879,
      "grad_norm": 18.410409927368164,
      "learning_rate": 1.4414513670834941e-06,
      "loss": 1.2794,
      "step": 58940
    },
    {
      "epoch": 9.741788886593678,
      "grad_norm": 13.786377906799316,
      "learning_rate": 1.432270148184873e-06,
      "loss": 1.2376,
      "step": 58950
    },
    {
      "epoch": 9.74344143771948,
      "grad_norm": 17.038511276245117,
      "learning_rate": 1.4230889292862522e-06,
      "loss": 1.2173,
      "step": 58960
    },
    {
      "epoch": 9.74509398884528,
      "grad_norm": 15.250676155090332,
      "learning_rate": 1.413907710387631e-06,
      "loss": 1.3068,
      "step": 58970
    },
    {
      "epoch": 9.746746539971081,
      "grad_norm": 28.942930221557617,
      "learning_rate": 1.4047264914890103e-06,
      "loss": 1.3075,
      "step": 58980
    },
    {
      "epoch": 9.74839909109688,
      "grad_norm": 12.493021965026855,
      "learning_rate": 1.395545272590389e-06,
      "loss": 1.3052,
      "step": 58990
    },
    {
      "epoch": 9.750051642222681,
      "grad_norm": 15.28077220916748,
      "learning_rate": 1.3863640536917681e-06,
      "loss": 1.3504,
      "step": 59000
    },
    {
      "epoch": 9.751704193348482,
      "grad_norm": 20.08050537109375,
      "learning_rate": 1.3771828347931472e-06,
      "loss": 1.2934,
      "step": 59010
    },
    {
      "epoch": 9.753356744474281,
      "grad_norm": 25.60179901123047,
      "learning_rate": 1.3680016158945262e-06,
      "loss": 1.2961,
      "step": 59020
    },
    {
      "epoch": 9.755009295600082,
      "grad_norm": 14.154158592224121,
      "learning_rate": 1.3588203969959052e-06,
      "loss": 1.4102,
      "step": 59030
    },
    {
      "epoch": 9.756661846725883,
      "grad_norm": 12.870734214782715,
      "learning_rate": 1.3496391780972843e-06,
      "loss": 1.134,
      "step": 59040
    },
    {
      "epoch": 9.758314397851684,
      "grad_norm": 9.939929962158203,
      "learning_rate": 1.3404579591986633e-06,
      "loss": 1.2146,
      "step": 59050
    },
    {
      "epoch": 9.759966948977484,
      "grad_norm": 14.322998046875,
      "learning_rate": 1.3312767403000423e-06,
      "loss": 1.1763,
      "step": 59060
    },
    {
      "epoch": 9.761619500103285,
      "grad_norm": 18.574724197387695,
      "learning_rate": 1.3220955214014214e-06,
      "loss": 1.3397,
      "step": 59070
    },
    {
      "epoch": 9.763272051229086,
      "grad_norm": 12.602153778076172,
      "learning_rate": 1.3129143025028004e-06,
      "loss": 1.245,
      "step": 59080
    },
    {
      "epoch": 9.764924602354885,
      "grad_norm": 21.25696563720703,
      "learning_rate": 1.3037330836041794e-06,
      "loss": 1.2831,
      "step": 59090
    },
    {
      "epoch": 9.766577153480686,
      "grad_norm": 24.119489669799805,
      "learning_rate": 1.2945518647055583e-06,
      "loss": 1.2615,
      "step": 59100
    },
    {
      "epoch": 9.768229704606487,
      "grad_norm": 20.59731101989746,
      "learning_rate": 1.2853706458069375e-06,
      "loss": 1.2008,
      "step": 59110
    },
    {
      "epoch": 9.769882255732286,
      "grad_norm": 17.279531478881836,
      "learning_rate": 1.2761894269083163e-06,
      "loss": 1.2945,
      "step": 59120
    },
    {
      "epoch": 9.771534806858087,
      "grad_norm": 15.187520027160645,
      "learning_rate": 1.2670082080096956e-06,
      "loss": 1.3367,
      "step": 59130
    },
    {
      "epoch": 9.773187357983888,
      "grad_norm": 24.268413543701172,
      "learning_rate": 1.2578269891110744e-06,
      "loss": 1.2033,
      "step": 59140
    },
    {
      "epoch": 9.774839909109687,
      "grad_norm": 10.84234619140625,
      "learning_rate": 1.2486457702124534e-06,
      "loss": 1.1307,
      "step": 59150
    },
    {
      "epoch": 9.776492460235488,
      "grad_norm": 19.090682983398438,
      "learning_rate": 1.2394645513138325e-06,
      "loss": 1.2864,
      "step": 59160
    },
    {
      "epoch": 9.778145011361289,
      "grad_norm": 16.51657485961914,
      "learning_rate": 1.2302833324152115e-06,
      "loss": 1.2448,
      "step": 59170
    },
    {
      "epoch": 9.77979756248709,
      "grad_norm": 10.925396919250488,
      "learning_rate": 1.2211021135165906e-06,
      "loss": 1.2297,
      "step": 59180
    },
    {
      "epoch": 9.78145011361289,
      "grad_norm": 19.890026092529297,
      "learning_rate": 1.2119208946179696e-06,
      "loss": 1.2474,
      "step": 59190
    },
    {
      "epoch": 9.78310266473869,
      "grad_norm": 21.863983154296875,
      "learning_rate": 1.2027396757193484e-06,
      "loss": 1.3076,
      "step": 59200
    },
    {
      "epoch": 9.784755215864491,
      "grad_norm": 20.163433074951172,
      "learning_rate": 1.1935584568207277e-06,
      "loss": 1.253,
      "step": 59210
    },
    {
      "epoch": 9.78640776699029,
      "grad_norm": 13.780261039733887,
      "learning_rate": 1.1843772379221065e-06,
      "loss": 1.2075,
      "step": 59220
    },
    {
      "epoch": 9.788060318116091,
      "grad_norm": 15.498116493225098,
      "learning_rate": 1.1751960190234857e-06,
      "loss": 1.3903,
      "step": 59230
    },
    {
      "epoch": 9.789712869241892,
      "grad_norm": 12.257755279541016,
      "learning_rate": 1.1660148001248646e-06,
      "loss": 1.379,
      "step": 59240
    },
    {
      "epoch": 9.791365420367693,
      "grad_norm": 15.42638111114502,
      "learning_rate": 1.1568335812262436e-06,
      "loss": 1.2928,
      "step": 59250
    },
    {
      "epoch": 9.793017971493493,
      "grad_norm": 17.679750442504883,
      "learning_rate": 1.1476523623276226e-06,
      "loss": 1.2155,
      "step": 59260
    },
    {
      "epoch": 9.794670522619294,
      "grad_norm": 13.531704902648926,
      "learning_rate": 1.1384711434290017e-06,
      "loss": 1.396,
      "step": 59270
    },
    {
      "epoch": 9.796323073745095,
      "grad_norm": 21.790496826171875,
      "learning_rate": 1.1292899245303807e-06,
      "loss": 1.3374,
      "step": 59280
    },
    {
      "epoch": 9.797975624870894,
      "grad_norm": 20.123689651489258,
      "learning_rate": 1.1201087056317597e-06,
      "loss": 1.3,
      "step": 59290
    },
    {
      "epoch": 9.799628175996695,
      "grad_norm": 16.618236541748047,
      "learning_rate": 1.1109274867331386e-06,
      "loss": 1.2193,
      "step": 59300
    },
    {
      "epoch": 9.801280727122496,
      "grad_norm": 87.83232879638672,
      "learning_rate": 1.1017462678345178e-06,
      "loss": 1.283,
      "step": 59310
    },
    {
      "epoch": 9.802933278248297,
      "grad_norm": 19.9451847076416,
      "learning_rate": 1.0925650489358966e-06,
      "loss": 1.2853,
      "step": 59320
    },
    {
      "epoch": 9.804585829374096,
      "grad_norm": 20.5788631439209,
      "learning_rate": 1.0833838300372759e-06,
      "loss": 1.3428,
      "step": 59330
    },
    {
      "epoch": 9.806238380499897,
      "grad_norm": 13.178668975830078,
      "learning_rate": 1.0742026111386547e-06,
      "loss": 1.3449,
      "step": 59340
    },
    {
      "epoch": 9.807890931625698,
      "grad_norm": 9.858376502990723,
      "learning_rate": 1.0650213922400337e-06,
      "loss": 1.2549,
      "step": 59350
    },
    {
      "epoch": 9.809543482751497,
      "grad_norm": 14.950618743896484,
      "learning_rate": 1.0558401733414128e-06,
      "loss": 1.1921,
      "step": 59360
    },
    {
      "epoch": 9.811196033877298,
      "grad_norm": 15.29289436340332,
      "learning_rate": 1.0466589544427918e-06,
      "loss": 1.2611,
      "step": 59370
    },
    {
      "epoch": 9.812848585003099,
      "grad_norm": 10.011371612548828,
      "learning_rate": 1.0374777355441709e-06,
      "loss": 1.3245,
      "step": 59380
    },
    {
      "epoch": 9.814501136128898,
      "grad_norm": 39.90605545043945,
      "learning_rate": 1.0282965166455499e-06,
      "loss": 1.1861,
      "step": 59390
    },
    {
      "epoch": 9.8161536872547,
      "grad_norm": 23.458520889282227,
      "learning_rate": 1.019115297746929e-06,
      "loss": 1.3771,
      "step": 59400
    },
    {
      "epoch": 9.8178062383805,
      "grad_norm": 24.068479537963867,
      "learning_rate": 1.009934078848308e-06,
      "loss": 1.1251,
      "step": 59410
    },
    {
      "epoch": 9.8194587895063,
      "grad_norm": 20.833627700805664,
      "learning_rate": 1.000752859949687e-06,
      "loss": 1.4204,
      "step": 59420
    },
    {
      "epoch": 9.8211113406321,
      "grad_norm": 13.489704132080078,
      "learning_rate": 9.91571641051066e-07,
      "loss": 1.3008,
      "step": 59430
    },
    {
      "epoch": 9.822763891757901,
      "grad_norm": 35.63694381713867,
      "learning_rate": 9.82390422152445e-07,
      "loss": 1.3852,
      "step": 59440
    },
    {
      "epoch": 9.824416442883702,
      "grad_norm": 17.55329132080078,
      "learning_rate": 9.732092032538239e-07,
      "loss": 1.3671,
      "step": 59450
    },
    {
      "epoch": 9.826068994009502,
      "grad_norm": 19.870128631591797,
      "learning_rate": 9.640279843552031e-07,
      "loss": 1.3378,
      "step": 59460
    },
    {
      "epoch": 9.827721545135303,
      "grad_norm": 31.158733367919922,
      "learning_rate": 9.54846765456582e-07,
      "loss": 1.2725,
      "step": 59470
    },
    {
      "epoch": 9.829374096261104,
      "grad_norm": 23.401962280273438,
      "learning_rate": 9.456655465579611e-07,
      "loss": 1.3456,
      "step": 59480
    },
    {
      "epoch": 9.831026647386903,
      "grad_norm": 11.159757614135742,
      "learning_rate": 9.3648432765934e-07,
      "loss": 1.2201,
      "step": 59490
    },
    {
      "epoch": 9.832679198512704,
      "grad_norm": 18.284528732299805,
      "learning_rate": 9.273031087607191e-07,
      "loss": 1.3702,
      "step": 59500
    },
    {
      "epoch": 9.834331749638505,
      "grad_norm": 22.221506118774414,
      "learning_rate": 9.181218898620981e-07,
      "loss": 1.1929,
      "step": 59510
    },
    {
      "epoch": 9.835984300764306,
      "grad_norm": 14.718652725219727,
      "learning_rate": 9.089406709634771e-07,
      "loss": 1.3242,
      "step": 59520
    },
    {
      "epoch": 9.837636851890105,
      "grad_norm": 30.458106994628906,
      "learning_rate": 8.997594520648562e-07,
      "loss": 1.2832,
      "step": 59530
    },
    {
      "epoch": 9.839289403015906,
      "grad_norm": 17.770442962646484,
      "learning_rate": 8.905782331662352e-07,
      "loss": 1.375,
      "step": 59540
    },
    {
      "epoch": 9.840941954141707,
      "grad_norm": 13.754855155944824,
      "learning_rate": 8.813970142676141e-07,
      "loss": 1.1981,
      "step": 59550
    },
    {
      "epoch": 9.842594505267506,
      "grad_norm": 28.639034271240234,
      "learning_rate": 8.722157953689933e-07,
      "loss": 1.3355,
      "step": 59560
    },
    {
      "epoch": 9.844247056393307,
      "grad_norm": 34.25663757324219,
      "learning_rate": 8.630345764703722e-07,
      "loss": 1.5129,
      "step": 59570
    },
    {
      "epoch": 9.845899607519108,
      "grad_norm": 8.183585166931152,
      "learning_rate": 8.538533575717514e-07,
      "loss": 1.262,
      "step": 59580
    },
    {
      "epoch": 9.847552158644907,
      "grad_norm": 27.922536849975586,
      "learning_rate": 8.446721386731303e-07,
      "loss": 1.2981,
      "step": 59590
    },
    {
      "epoch": 9.849204709770708,
      "grad_norm": 11.618725776672363,
      "learning_rate": 8.354909197745092e-07,
      "loss": 1.2361,
      "step": 59600
    },
    {
      "epoch": 9.85085726089651,
      "grad_norm": 21.355430603027344,
      "learning_rate": 8.263097008758884e-07,
      "loss": 1.3164,
      "step": 59610
    },
    {
      "epoch": 9.85250981202231,
      "grad_norm": 17.563817977905273,
      "learning_rate": 8.171284819772673e-07,
      "loss": 1.1633,
      "step": 59620
    },
    {
      "epoch": 9.85416236314811,
      "grad_norm": 15.469801902770996,
      "learning_rate": 8.079472630786464e-07,
      "loss": 1.3304,
      "step": 59630
    },
    {
      "epoch": 9.85581491427391,
      "grad_norm": 12.664761543273926,
      "learning_rate": 7.987660441800254e-07,
      "loss": 1.3019,
      "step": 59640
    },
    {
      "epoch": 9.857467465399711,
      "grad_norm": 18.20703125,
      "learning_rate": 7.895848252814045e-07,
      "loss": 1.3006,
      "step": 59650
    },
    {
      "epoch": 9.85912001652551,
      "grad_norm": 14.753276824951172,
      "learning_rate": 7.804036063827834e-07,
      "loss": 1.3308,
      "step": 59660
    },
    {
      "epoch": 9.860772567651312,
      "grad_norm": 12.871907234191895,
      "learning_rate": 7.712223874841625e-07,
      "loss": 1.3225,
      "step": 59670
    },
    {
      "epoch": 9.862425118777113,
      "grad_norm": 10.84438419342041,
      "learning_rate": 7.620411685855414e-07,
      "loss": 1.1678,
      "step": 59680
    },
    {
      "epoch": 9.864077669902912,
      "grad_norm": 15.910385131835938,
      "learning_rate": 7.528599496869204e-07,
      "loss": 1.3399,
      "step": 59690
    },
    {
      "epoch": 9.865730221028713,
      "grad_norm": 15.912435531616211,
      "learning_rate": 7.436787307882995e-07,
      "loss": 1.251,
      "step": 59700
    },
    {
      "epoch": 9.867382772154514,
      "grad_norm": 14.724613189697266,
      "learning_rate": 7.344975118896785e-07,
      "loss": 1.3633,
      "step": 59710
    },
    {
      "epoch": 9.869035323280315,
      "grad_norm": 19.542469024658203,
      "learning_rate": 7.253162929910575e-07,
      "loss": 1.4641,
      "step": 59720
    },
    {
      "epoch": 9.870687874406114,
      "grad_norm": 15.426322937011719,
      "learning_rate": 7.161350740924365e-07,
      "loss": 1.4261,
      "step": 59730
    },
    {
      "epoch": 9.872340425531915,
      "grad_norm": 12.275348663330078,
      "learning_rate": 7.069538551938155e-07,
      "loss": 1.3722,
      "step": 59740
    },
    {
      "epoch": 9.873992976657716,
      "grad_norm": 13.315044403076172,
      "learning_rate": 6.977726362951945e-07,
      "loss": 1.3351,
      "step": 59750
    },
    {
      "epoch": 9.875645527783515,
      "grad_norm": 24.9056453704834,
      "learning_rate": 6.885914173965736e-07,
      "loss": 1.3371,
      "step": 59760
    },
    {
      "epoch": 9.877298078909316,
      "grad_norm": 14.293803215026855,
      "learning_rate": 6.794101984979526e-07,
      "loss": 1.1908,
      "step": 59770
    },
    {
      "epoch": 9.878950630035117,
      "grad_norm": 20.57039451599121,
      "learning_rate": 6.702289795993317e-07,
      "loss": 1.2033,
      "step": 59780
    },
    {
      "epoch": 9.880603181160918,
      "grad_norm": 13.297779083251953,
      "learning_rate": 6.610477607007107e-07,
      "loss": 1.163,
      "step": 59790
    },
    {
      "epoch": 9.882255732286717,
      "grad_norm": 13.230541229248047,
      "learning_rate": 6.518665418020897e-07,
      "loss": 1.4123,
      "step": 59800
    },
    {
      "epoch": 9.883908283412518,
      "grad_norm": 24.931100845336914,
      "learning_rate": 6.426853229034688e-07,
      "loss": 1.3383,
      "step": 59810
    },
    {
      "epoch": 9.88556083453832,
      "grad_norm": 17.547151565551758,
      "learning_rate": 6.335041040048478e-07,
      "loss": 1.3894,
      "step": 59820
    },
    {
      "epoch": 9.887213385664118,
      "grad_norm": 16.77924346923828,
      "learning_rate": 6.243228851062267e-07,
      "loss": 1.32,
      "step": 59830
    },
    {
      "epoch": 9.88886593678992,
      "grad_norm": 16.8702449798584,
      "learning_rate": 6.151416662076058e-07,
      "loss": 1.3243,
      "step": 59840
    },
    {
      "epoch": 9.89051848791572,
      "grad_norm": 15.003113746643066,
      "learning_rate": 6.059604473089848e-07,
      "loss": 1.2222,
      "step": 59850
    },
    {
      "epoch": 9.89217103904152,
      "grad_norm": 15.460107803344727,
      "learning_rate": 5.967792284103638e-07,
      "loss": 1.2588,
      "step": 59860
    },
    {
      "epoch": 9.89382359016732,
      "grad_norm": 17.36907958984375,
      "learning_rate": 5.875980095117429e-07,
      "loss": 1.3467,
      "step": 59870
    },
    {
      "epoch": 9.895476141293122,
      "grad_norm": 13.089510917663574,
      "learning_rate": 5.784167906131218e-07,
      "loss": 1.3112,
      "step": 59880
    },
    {
      "epoch": 9.89712869241892,
      "grad_norm": 22.797977447509766,
      "learning_rate": 5.692355717145008e-07,
      "loss": 1.379,
      "step": 59890
    },
    {
      "epoch": 9.898781243544722,
      "grad_norm": 18.484411239624023,
      "learning_rate": 5.600543528158799e-07,
      "loss": 1.3037,
      "step": 59900
    },
    {
      "epoch": 9.900433794670523,
      "grad_norm": 10.942911148071289,
      "learning_rate": 5.508731339172589e-07,
      "loss": 1.3234,
      "step": 59910
    },
    {
      "epoch": 9.902086345796324,
      "grad_norm": 17.38446044921875,
      "learning_rate": 5.416919150186379e-07,
      "loss": 1.3274,
      "step": 59920
    },
    {
      "epoch": 9.903738896922123,
      "grad_norm": 140.1362762451172,
      "learning_rate": 5.325106961200169e-07,
      "loss": 1.3595,
      "step": 59930
    },
    {
      "epoch": 9.905391448047924,
      "grad_norm": 16.535696029663086,
      "learning_rate": 5.233294772213959e-07,
      "loss": 1.3435,
      "step": 59940
    },
    {
      "epoch": 9.907043999173725,
      "grad_norm": 17.555944442749023,
      "learning_rate": 5.141482583227749e-07,
      "loss": 1.3091,
      "step": 59950
    },
    {
      "epoch": 9.908696550299524,
      "grad_norm": 13.292182922363281,
      "learning_rate": 5.04967039424154e-07,
      "loss": 1.3891,
      "step": 59960
    },
    {
      "epoch": 9.910349101425325,
      "grad_norm": 15.89194393157959,
      "learning_rate": 4.95785820525533e-07,
      "loss": 1.2585,
      "step": 59970
    },
    {
      "epoch": 9.912001652551126,
      "grad_norm": 11.141946792602539,
      "learning_rate": 4.866046016269119e-07,
      "loss": 1.3311,
      "step": 59980
    },
    {
      "epoch": 9.913654203676927,
      "grad_norm": 18.01121711730957,
      "learning_rate": 4.77423382728291e-07,
      "loss": 1.2728,
      "step": 59990
    },
    {
      "epoch": 9.915306754802726,
      "grad_norm": 14.889471054077148,
      "learning_rate": 4.6824216382967e-07,
      "loss": 1.2427,
      "step": 60000
    },
    {
      "epoch": 9.916959305928527,
      "grad_norm": 24.612850189208984,
      "learning_rate": 4.5906094493104905e-07,
      "loss": 1.2339,
      "step": 60010
    },
    {
      "epoch": 9.918611857054328,
      "grad_norm": 11.734923362731934,
      "learning_rate": 4.498797260324281e-07,
      "loss": 1.3032,
      "step": 60020
    },
    {
      "epoch": 9.920264408180127,
      "grad_norm": 20.81572723388672,
      "learning_rate": 4.4069850713380707e-07,
      "loss": 1.2575,
      "step": 60030
    },
    {
      "epoch": 9.921916959305928,
      "grad_norm": 17.165307998657227,
      "learning_rate": 4.315172882351861e-07,
      "loss": 1.3268,
      "step": 60040
    },
    {
      "epoch": 9.92356951043173,
      "grad_norm": 10.70225715637207,
      "learning_rate": 4.2233606933656514e-07,
      "loss": 1.3622,
      "step": 60050
    },
    {
      "epoch": 9.925222061557529,
      "grad_norm": 36.28559494018555,
      "learning_rate": 4.131548504379442e-07,
      "loss": 1.2658,
      "step": 60060
    },
    {
      "epoch": 9.92687461268333,
      "grad_norm": 17.466758728027344,
      "learning_rate": 4.039736315393232e-07,
      "loss": 1.3854,
      "step": 60070
    },
    {
      "epoch": 9.92852716380913,
      "grad_norm": 19.143348693847656,
      "learning_rate": 3.9479241264070225e-07,
      "loss": 1.0911,
      "step": 60080
    },
    {
      "epoch": 9.930179714934932,
      "grad_norm": 18.350982666015625,
      "learning_rate": 3.8561119374208123e-07,
      "loss": 1.2973,
      "step": 60090
    },
    {
      "epoch": 9.93183226606073,
      "grad_norm": 12.689515113830566,
      "learning_rate": 3.764299748434602e-07,
      "loss": 1.3574,
      "step": 60100
    },
    {
      "epoch": 9.933484817186532,
      "grad_norm": 8.145671844482422,
      "learning_rate": 3.6724875594483925e-07,
      "loss": 1.4449,
      "step": 60110
    },
    {
      "epoch": 9.935137368312333,
      "grad_norm": 23.987871170043945,
      "learning_rate": 3.5806753704621824e-07,
      "loss": 1.298,
      "step": 60120
    },
    {
      "epoch": 9.936789919438132,
      "grad_norm": 18.037437438964844,
      "learning_rate": 3.4888631814759727e-07,
      "loss": 1.2382,
      "step": 60130
    },
    {
      "epoch": 9.938442470563933,
      "grad_norm": 16.41983413696289,
      "learning_rate": 3.397050992489763e-07,
      "loss": 1.3517,
      "step": 60140
    },
    {
      "epoch": 9.940095021689734,
      "grad_norm": 16.38405418395996,
      "learning_rate": 3.3052388035035534e-07,
      "loss": 1.2899,
      "step": 60150
    },
    {
      "epoch": 9.941747572815533,
      "grad_norm": 16.4738826751709,
      "learning_rate": 3.213426614517344e-07,
      "loss": 1.29,
      "step": 60160
    },
    {
      "epoch": 9.943400123941334,
      "grad_norm": 22.556604385375977,
      "learning_rate": 3.1216144255311336e-07,
      "loss": 1.2654,
      "step": 60170
    },
    {
      "epoch": 9.945052675067135,
      "grad_norm": 11.890191078186035,
      "learning_rate": 3.029802236544924e-07,
      "loss": 1.0929,
      "step": 60180
    },
    {
      "epoch": 9.946705226192936,
      "grad_norm": 72.20085906982422,
      "learning_rate": 2.9379900475587143e-07,
      "loss": 1.2483,
      "step": 60190
    },
    {
      "epoch": 9.948357777318735,
      "grad_norm": 17.972667694091797,
      "learning_rate": 2.846177858572504e-07,
      "loss": 1.2809,
      "step": 60200
    },
    {
      "epoch": 9.950010328444536,
      "grad_norm": 21.140539169311523,
      "learning_rate": 2.7543656695862945e-07,
      "loss": 1.4348,
      "step": 60210
    },
    {
      "epoch": 9.951662879570337,
      "grad_norm": 17.385190963745117,
      "learning_rate": 2.6625534806000844e-07,
      "loss": 1.3182,
      "step": 60220
    },
    {
      "epoch": 9.953315430696136,
      "grad_norm": 9.468734741210938,
      "learning_rate": 2.5707412916138747e-07,
      "loss": 1.1968,
      "step": 60230
    },
    {
      "epoch": 9.954967981821937,
      "grad_norm": 9.555684089660645,
      "learning_rate": 2.478929102627665e-07,
      "loss": 1.3367,
      "step": 60240
    },
    {
      "epoch": 9.956620532947738,
      "grad_norm": 11.08018684387207,
      "learning_rate": 2.387116913641455e-07,
      "loss": 1.392,
      "step": 60250
    },
    {
      "epoch": 9.95827308407354,
      "grad_norm": 16.899229049682617,
      "learning_rate": 2.2953047246552453e-07,
      "loss": 1.2918,
      "step": 60260
    },
    {
      "epoch": 9.959925635199339,
      "grad_norm": 15.151360511779785,
      "learning_rate": 2.2034925356690354e-07,
      "loss": 1.2946,
      "step": 60270
    },
    {
      "epoch": 9.96157818632514,
      "grad_norm": 8.768255233764648,
      "learning_rate": 2.1116803466828257e-07,
      "loss": 1.3185,
      "step": 60280
    },
    {
      "epoch": 9.96323073745094,
      "grad_norm": 11.71931266784668,
      "learning_rate": 2.019868157696616e-07,
      "loss": 1.3727,
      "step": 60290
    },
    {
      "epoch": 9.96488328857674,
      "grad_norm": 25.73476219177246,
      "learning_rate": 1.9280559687104062e-07,
      "loss": 1.2354,
      "step": 60300
    },
    {
      "epoch": 9.96653583970254,
      "grad_norm": 25.98105239868164,
      "learning_rate": 1.8362437797241963e-07,
      "loss": 1.2941,
      "step": 60310
    },
    {
      "epoch": 9.968188390828342,
      "grad_norm": 35.79945373535156,
      "learning_rate": 1.7444315907379864e-07,
      "loss": 1.2131,
      "step": 60320
    },
    {
      "epoch": 9.969840941954141,
      "grad_norm": 13.044464111328125,
      "learning_rate": 1.6526194017517767e-07,
      "loss": 1.2468,
      "step": 60330
    },
    {
      "epoch": 9.971493493079942,
      "grad_norm": 25.407047271728516,
      "learning_rate": 1.5608072127655668e-07,
      "loss": 1.379,
      "step": 60340
    },
    {
      "epoch": 9.973146044205743,
      "grad_norm": 13.202648162841797,
      "learning_rate": 1.4689950237793572e-07,
      "loss": 1.185,
      "step": 60350
    },
    {
      "epoch": 9.974798595331544,
      "grad_norm": 38.73662567138672,
      "learning_rate": 1.3771828347931473e-07,
      "loss": 1.2843,
      "step": 60360
    },
    {
      "epoch": 9.976451146457343,
      "grad_norm": 14.0643310546875,
      "learning_rate": 1.2853706458069374e-07,
      "loss": 1.294,
      "step": 60370
    },
    {
      "epoch": 9.978103697583144,
      "grad_norm": 12.296748161315918,
      "learning_rate": 1.1935584568207275e-07,
      "loss": 1.2442,
      "step": 60380
    },
    {
      "epoch": 9.979756248708945,
      "grad_norm": 19.61980628967285,
      "learning_rate": 1.1017462678345177e-07,
      "loss": 1.1628,
      "step": 60390
    },
    {
      "epoch": 9.981408799834744,
      "grad_norm": 15.85641860961914,
      "learning_rate": 1.009934078848308e-07,
      "loss": 1.2077,
      "step": 60400
    },
    {
      "epoch": 9.983061350960545,
      "grad_norm": 18.99192237854004,
      "learning_rate": 9.181218898620981e-08,
      "loss": 1.1939,
      "step": 60410
    },
    {
      "epoch": 9.984713902086346,
      "grad_norm": 34.235374450683594,
      "learning_rate": 8.263097008758884e-08,
      "loss": 1.3344,
      "step": 60420
    },
    {
      "epoch": 9.986366453212145,
      "grad_norm": 13.520135879516602,
      "learning_rate": 7.344975118896786e-08,
      "loss": 1.2839,
      "step": 60430
    },
    {
      "epoch": 9.988019004337946,
      "grad_norm": 18.85852813720703,
      "learning_rate": 6.426853229034687e-08,
      "loss": 1.2802,
      "step": 60440
    },
    {
      "epoch": 9.989671555463747,
      "grad_norm": 16.009300231933594,
      "learning_rate": 5.5087313391725884e-08,
      "loss": 1.2258,
      "step": 60450
    },
    {
      "epoch": 9.991324106589548,
      "grad_norm": 14.094069480895996,
      "learning_rate": 4.5906094493104907e-08,
      "loss": 1.3012,
      "step": 60460
    },
    {
      "epoch": 9.992976657715348,
      "grad_norm": 23.66718101501465,
      "learning_rate": 3.672487559448393e-08,
      "loss": 1.329,
      "step": 60470
    },
    {
      "epoch": 9.994629208841149,
      "grad_norm": 18.014339447021484,
      "learning_rate": 2.7543656695862942e-08,
      "loss": 1.3193,
      "step": 60480
    },
    {
      "epoch": 9.99628175996695,
      "grad_norm": 12.735496520996094,
      "learning_rate": 1.8362437797241965e-08,
      "loss": 1.2747,
      "step": 60490
    },
    {
      "epoch": 9.997934311092749,
      "grad_norm": 29.70014190673828,
      "learning_rate": 9.181218898620982e-09,
      "loss": 1.3249,
      "step": 60500
    },
    {
      "epoch": 9.99958686221855,
      "grad_norm": 17.789243698120117,
      "learning_rate": 0.0,
      "loss": 1.3595,
      "step": 60510
    },
    {
      "epoch": 9.99958686221855,
      "eval_accuracy": 0.33699815576677544,
      "eval_loss": 2.4853203296661377,
      "eval_runtime": 861.2243,
      "eval_samples_per_second": 32.739,
      "eval_steps_per_second": 8.185,
      "step": 60510
    },
    {
      "epoch": 9.99958686221855,
      "step": 60510,
      "total_flos": 9.925024630549355e+19,
      "train_loss": 1.502550286699811,
      "train_runtime": 81421.0241,
      "train_samples_per_second": 11.891,
      "train_steps_per_second": 0.743
    }
  ],
  "logging_steps": 10,
  "max_steps": 60510,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 9.925024630549355e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}